test_new_dataloader.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Table of contents**<a id='toc0_'></a>    \n",
    "- [Image classification example](#toc1_)    \n",
    "- [Pretrained Image classification example - ResNet18](#toc2_)    \n",
    "- [Pretrained Image classification example - Transformer](#toc3_)    \n",
    "- [Sequential classification (Unsupported for now)](#toc4_)    \n",
    "\n",
    "<!-- vscode-jupyter-toc-config\n",
    "\tnumbering=false\n",
    "\tanchor=true\n",
    "\tflat=false\n",
    "\tminLevel=1\n",
    "\tmaxLevel=6\n",
    "\t/vscode-jupyter-toc-config -->\n",
    "<!-- THIS CELL WILL BE REPLACED ON TOC UPDATE. DO NOT WRITE YOUR TEXT IN THIS CELL -->"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = ''"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Image classification task"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-09-23T12:01:33.514283Z",
     "start_time": "2024-09-23T12:01:32.116044Z"
    }
   },
   "outputs": [],
   "source": [
    "import torch.nn\n",
    "import torch.optim\n",
    "\n",
    "import openml_pytorch.config\n",
    "import openml\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# Suppress FutureWarning messages\n",
    "warnings.simplefilter(action='ignore')\n",
    "\n",
    "############################################################################\n",
    "# Enable logging in order to observe the progress while running the example.\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "from openml_pytorch.trainer import OpenMLTrainerModule\n",
    "from openml_pytorch.trainer import OpenMLDataModule\n",
    "from torchvision.transforms import Compose, Resize, ToPILImage, ToTensor, Lambda\n",
    "import torchvision\n",
    "\n",
    "from openml_pytorch.trainer import convert_to_rgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-09-23T12:01:39.788930Z",
     "start_time": "2024-09-23T12:01:34.041129Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Received uncompressed content from OpenML for https://api.openml.org/data/v1/download/22120926/tiniest-imagenet-200.arff.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 0 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.640577377507716, tensor(0.0034, device='mps:0')]\n",
      "valid: [5.368189832899305, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.7403, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 1 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.676578173225309, tensor(0.0031, device='mps:0')]\n",
      "valid: [5.3457007514105905, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.2909, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 2 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.6762894241898145, tensor(0.0034, device='mps:0')]\n",
      "valid: [5.445003594292535, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.5449, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 3 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.618812090084877, tensor(0.0034, device='mps:0')]\n",
      "valid: [5.329725816514757, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.2780, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 4 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.6880232445987655, tensor(0.0037, device='mps:0')]\n",
      "valid: [5.357654147677952, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.3656, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 5 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.670154562114197, tensor(0.0049, device='mps:0')]\n",
      "valid: [5.3565253363715275, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.3674, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 6 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.632607542438271, tensor(0.0046, device='mps:0')]\n",
      "valid: [5.383794487847222, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.4994, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 7 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.633710093557099, tensor(0.0040, device='mps:0')]\n",
      "valid: [5.288886176215278, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.3511, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 8 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.6644796489197535, tensor(0.0049, device='mps:0')]\n",
      "valid: [5.352758449978299, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.4141, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n",
      "INFO:openml.config:Going to run model ResNet(\n",
      "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
      "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  (relu): ReLU(inplace=True)\n",
      "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
      "  (layer1): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer2): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer3): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (layer4): Sequential(\n",
      "    (0): BasicBlock(\n",
      "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (downsample): Sequential(\n",
      "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
      "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      )\n",
      "    )\n",
      "    (1): BasicBlock(\n",
      "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (relu): ReLU(inplace=True)\n",
      "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
      "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
      "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 9 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.632195818865741, tensor(0.0043, device='mps:0')]\n",
      "valid: [5.409883287217882, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(6.2667, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:openml.config:Executed Task 362128 on local Flow with name torch.nn.ResNet.2a12408ac25b016a.\n"
     ]
    }
   ],
   "source": [
    "transform = Compose(\n",
    "    [\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(\n",
    "            convert_to_rgb\n",
    "        ),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize(\n",
    "            (64, 64)\n",
    "        ),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")\n",
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"image\",\n",
    "    file_dir=\"datasets\",\n",
    "    filename_col=\"image_path\",\n",
    "    target_mode=\"categorical\",\n",
    "    target_column=\"label\",\n",
    "    batch_size = 64,\n",
    "    transform=transform\n",
    ")\n",
    "# Download the OpenML task for tiniest imagenet\n",
    "task = openml.tasks.get_task(362128)\n",
    "model = torchvision.models.resnet18(num_classes=200)\n",
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    data_module=data_module,\n",
    "    verbose = True,\n",
    "    epoch_count = 1,\n",
    "    # optimizer_gen = torch.optim.AdamW\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:openml.datasets.functions:Could not download file from https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq: Object at 'https://openml1.win.tue.nl/datasets/0004/46347/dataset_46347.pq' does not exist.\n",
      "WARNING:openml.datasets.functions:Failed to download parquet, fallback on ARFF.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "OpenML Run\n",
       "==========\n",
       "Uploader Name...................: None\n",
       "Metric..........................: None\n",
       "Local Result - Accuracy (+- STD): 0.0055 +- 0.0033\n",
       "Run ID..........................: 10595158\n",
       "Run URL.........................: https://www.openml.org/r/10595158\n",
       "Task ID.........................: 362128\n",
       "Task Type.......................: None\n",
       "Task URL........................: https://www.openml.org/t/362128\n",
       "Flow ID.........................: 25011\n",
       "Flow Name.......................: torch.nn.ResNet.2a12408ac25b016a\n",
       "Flow URL........................: https://www.openml.org/f/25011\n",
       "Setup ID........................: None\n",
       "Setup String....................: Python_3.11.9. Torch_2.4.1. NumPy_1.26.4. SciPy_1.14.1. ResNet(\n",
       "  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (relu): ReLU(inplace=True)\n",
       "  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "  (layer1): Sequential(\n",
       "    (0): BasicBlock(\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    )\n",
       "    (1): BasicBlock(\n",
       "      (conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    )\n",
       "  )\n",
       "  (layer2): Sequential(\n",
       "    (0): BasicBlock(\n",
       "      (conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): BasicBlock(\n",
       "      (conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    )\n",
       "  )\n",
       "  (layer3): Sequential(\n",
       "    (0): BasicBlock(\n",
       "      (conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): BasicBlock(\n",
       "      (conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    )\n",
       "  )\n",
       "  (layer4): Sequential(\n",
       "    (0): BasicBlock(\n",
       "      (conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (downsample): Sequential(\n",
       "        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      )\n",
       "    )\n",
       "    (1): BasicBlock(\n",
       "      (conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (relu): ReLU(inplace=True)\n",
       "      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    )\n",
       "  )\n",
       "  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "  (fc): Linear(in_features=512, out_features=200, bias=True)\n",
       ")\n",
       "Dataset ID......................: 46347\n",
       "Dataset URL.....................: https://www.openml.org/d/46347"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['n01443537', 'n01629819', 'n01641577', 'n01644900', 'n01698640',\n",
       "       'n01742172', 'n01768244', 'n01770393', 'n01774384', 'n01774750',\n",
       "       'n01784675', 'n01855672', 'n01882714', 'n01910747', 'n01917289',\n",
       "       'n01944390', 'n01945685', 'n01950731', 'n01983481', 'n01984695',\n",
       "       'n02002724', 'n02056570', 'n02058221', 'n02074367', 'n02085620',\n",
       "       'n02094433', 'n02099601', 'n02099712', 'n02106662', 'n02113799',\n",
       "       'n02123045', 'n02123394', 'n02124075', 'n02125311', 'n02129165',\n",
       "       'n02132136', 'n02165456', 'n02190166', 'n02206856', 'n02226429',\n",
       "       'n02231487', 'n02233338', 'n02236044', 'n02268443', 'n02279972',\n",
       "       'n02281406', 'n02321529', 'n02364673', 'n02395406', 'n02403003',\n",
       "       'n02410509', 'n02415577', 'n02423022', 'n02437312', 'n02480495',\n",
       "       'n02481823', 'n02486410', 'n02504458', 'n02509815', 'n02666196',\n",
       "       'n02669723', 'n02699494', 'n02730930', 'n02769748', 'n02788148',\n",
       "       'n02791270', 'n02793495', 'n02795169', 'n02802426', 'n02808440',\n",
       "       'n02814533', 'n02814860', 'n02815834', 'n02823428', 'n02837789',\n",
       "       'n02841315', 'n02843684', 'n02883205', 'n02892201', 'n02906734',\n",
       "       'n02909870', 'n02917067', 'n02927161', 'n02948072', 'n02950826',\n",
       "       'n02963159', 'n02977058', 'n02988304', 'n02999410', 'n03014705',\n",
       "       'n03026506', 'n03042490', 'n03085013', 'n03089624', 'n03100240',\n",
       "       'n03126707', 'n03160309', 'n03179701', 'n03201208', 'n03250847',\n",
       "       'n03255030', 'n03355925', 'n03388043', 'n03393912', 'n03400231',\n",
       "       'n03404251', 'n03424325', 'n03444034', 'n03447447', 'n03544143',\n",
       "       'n03584254', 'n03599486', 'n03617480', 'n03637318', 'n03649909',\n",
       "       'n03662601', 'n03670208', 'n03706229', 'n03733131', 'n03763968',\n",
       "       'n03770439', 'n03796401', 'n03804744', 'n03814639', 'n03837869',\n",
       "       'n03838899', 'n03854065', 'n03891332', 'n03902125', 'n03930313',\n",
       "       'n03937543', 'n03970156', 'n03976657', 'n03977966', 'n03980874',\n",
       "       'n03983396', 'n03992509', 'n04008634', 'n04023962', 'n04067472',\n",
       "       'n04070727', 'n04074963', 'n04099969', 'n04118538', 'n04133789',\n",
       "       'n04146614', 'n04149813', 'n04179913', 'n04251144', 'n04254777',\n",
       "       'n04259630', 'n04265275', 'n04275548', 'n04285008', 'n04311004',\n",
       "       'n04328186', 'n04356056', 'n04366367', 'n04371430', 'n04376876',\n",
       "       'n04398044', 'n04399382', 'n04417672', 'n04456115', 'n04465501',\n",
       "       'n04486054', 'n04487081', 'n04501370', 'n04507155', 'n04532106',\n",
       "       'n04532670', 'n04540053', 'n04560804', 'n04562935', 'n04596742',\n",
       "       'n04597913', 'n06596364', 'n07579787', 'n07583066', 'n07614500',\n",
       "       'n07615774', 'n07695742', 'n07711569', 'n07715103', 'n07720875',\n",
       "       'n07734744', 'n07747607', 'n07749582', 'n07753592', 'n07768694',\n",
       "       'n07871810', 'n07873807', 'n07875152', 'n07920052', 'n09193705',\n",
       "       'n09246464', 'n09256479', 'n09332890', 'n09428293', 'n12267677'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.learn.model_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sequential classification (Old example MNIST)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import torch.nn\n",
    "import torch.optim\n",
    "\n",
    "import openml_pytorch.config\n",
    "import openml\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# Suppress FutureWarning messages\n",
    "warnings.simplefilter(action='ignore')\n",
    "\n",
    "############################################################################\n",
    "# Enable logging in order to observe the progress while running the example.\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "############################################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from openml_pytorch.trainer import OpenMLTrainerModule\n",
    "from openml_pytorch.trainer import OpenMLDataModule"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "############################################################################\n",
    "# Define a sequential network that does the initial image reshaping\n",
    "# and normalization model.\n",
    "processing_net = torch.nn.Sequential(\n",
    "    openml_pytorch.layers.Functional(function=torch.Tensor.reshape,\n",
    "                                                shape=(-1, 1, 28, 28)),\n",
    "    torch.nn.BatchNorm2d(num_features=1)\n",
    ")\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "# Define a sequential network that does the extracts the features from the\n",
    "# image.\n",
    "features_net = torch.nn.Sequential(\n",
    "    torch.nn.Conv2d(in_channels=1, out_channels=32, kernel_size=5),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.MaxPool2d(kernel_size=2),\n",
    "    torch.nn.Conv2d(in_channels=32, out_channels=64, kernel_size=5),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.MaxPool2d(kernel_size=2),\n",
    ")\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "# Define a sequential network that flattens the features and compiles the\n",
    "# results into probabilities for each digit.\n",
    "results_net = torch.nn.Sequential(\n",
    "    openml_pytorch.layers.Functional(function=torch.Tensor.reshape,\n",
    "                                                shape=(-1, 4 * 4 * 64)),\n",
    "    torch.nn.Linear(in_features=4 * 4 * 64, out_features=256),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.Dropout(),\n",
    "    torch.nn.Linear(in_features=256, out_features=10),\n",
    ")\n",
    "############################################################################\n",
    "# openml.config.apikey = 'key'\n",
    "\n",
    "############################################################################\n",
    "# The main network, composed of the above specified networks.\n",
    "model = torch.nn.Sequential(\n",
    "    processing_net,\n",
    "    features_net,\n",
    "    results_net\n",
    ")\n",
    "############################################################################\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"dataframe\",\n",
    "    filename_col=\"class\",\n",
    "    target_mode=\"categorical\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    data_module=data_module,\n",
    "    verbose = True,\n",
    "    epoch_count = 1,\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Download the OpenML task for the mnist 784 dataset.\n",
    "task = openml.tasks.get_task(3573)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 0 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.37000785383597884, tensor(0.8802, device='mps:0')]\n",
      "valid: [0.06105882432725694, tensor(0.9798, device='mps:0')]\n",
      "Loss tensor(0.1303, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 1 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3643697296626984, tensor(0.8875, device='mps:0')]\n",
      "valid: [0.05847218831380208, tensor(0.9806, device='mps:0')]\n",
      "Loss tensor(0.2014, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 2 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3116477210097002, tensor(0.9034, device='mps:0')]\n",
      "valid: [0.05281951904296875, tensor(0.9852, device='mps:0')]\n",
      "Loss tensor(0.1903, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 3 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3643473048941799, tensor(0.8850, device='mps:0')]\n",
      "valid: [0.07235263400607639, tensor(0.9790, device='mps:0')]\n",
      "Loss tensor(0.2772, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 4 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.33402395419973546, tensor(0.8938, device='mps:0')]\n",
      "valid: [0.06707420228019594, tensor(0.9817, device='mps:0')]\n",
      "Loss tensor(0.2481, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 5 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3635254422949735, tensor(0.8826, device='mps:0')]\n",
      "valid: [0.06506180051773314, tensor(0.9811, device='mps:0')]\n",
      "Loss tensor(0.0397, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 6 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.38557877259700174, tensor(0.8777, device='mps:0')]\n",
      "valid: [0.06910077776227679, tensor(0.9816, device='mps:0')]\n",
      "Loss tensor(0.0020, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 7 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3358298542217813, tensor(0.8956, device='mps:0')]\n",
      "valid: [0.07069795154389881, tensor(0.9813, device='mps:0')]\n",
      "Loss tensor(0.0143, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 8 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.33729483575837743, tensor(0.8953, device='mps:0')]\n",
      "valid: [0.07018110971602183, tensor(0.9821, device='mps:0')]\n",
      "Loss tensor(0.1584, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model Sequential(\n",
      "  (0): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "  )\n",
      "  (1): Sequential(\n",
      "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (1): LeakyReLU(negative_slope=0.01)\n",
      "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
      "    (4): LeakyReLU(negative_slope=0.01)\n",
      "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
      "  )\n",
      "  (2): Sequential(\n",
      "    (0): Functional()\n",
      "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
      "    (2): LeakyReLU(negative_slope=0.01)\n",
      "    (3): Dropout(p=0.5, inplace=False)\n",
      "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
      "  )\n",
      ") on dataset mnist_784 for repeat 0 fold 9 sample 0\n",
      "                                                  \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.3252566964285714, tensor(0.8973, device='mps:0')]\n",
      "valid: [0.07023202563089037, tensor(0.9816, device='mps:0')]\n",
      "Loss tensor(0.3708, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:openml.config:Executed Task 3573 on local Flow with name torch.nn.Sequential.bb8de941c9933dd2.\n"
     ]
    }
   ],
   "source": [
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OpenML Run\n",
       "==========\n",
       "Uploader Name...................: None\n",
       "Metric..........................: None\n",
       "Local Result - Accuracy (+- STD): 0.9814 +- 0.0015\n",
       "Run ID..........................: 10595159\n",
       "Run URL.........................: https://www.openml.org/r/10595159\n",
       "Task ID.........................: 3573\n",
       "Task Type.......................: None\n",
       "Task URL........................: https://www.openml.org/t/3573\n",
       "Flow ID.........................: 25079\n",
       "Flow Name.......................: torch.nn.Sequential.bb8de941c9933dd2\n",
       "Flow URL........................: https://www.openml.org/f/25079\n",
       "Setup ID........................: None\n",
       "Setup String....................: Python_3.11.9. Torch_2.4.1. NumPy_1.26.4. SciPy_1.14.1. Sequential(\n",
       "  (0): Sequential(\n",
       "    (0): Functional()\n",
       "    (1): BatchNorm2d(1, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  )\n",
       "  (1): Sequential(\n",
       "    (0): Conv2d(1, 32, kernel_size=(5, 5), stride=(1, 1))\n",
       "    (1): LeakyReLU(negative_slope=0.01)\n",
       "    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "    (3): Conv2d(32, 64, kernel_size=(5, 5), stride=(1, 1))\n",
       "    (4): LeakyReLU(negative_slope=0.01)\n",
       "    (5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  )\n",
       "  (2): Sequential(\n",
       "    (0): Functional()\n",
       "    (1): Linear(in_features=1024, out_features=256, bias=True)\n",
       "    (2): LeakyReLU(negative_slope=0.01)\n",
       "    (3): Dropout(p=0.5, inplace=False)\n",
       "    (4): Linear(in_features=256, out_features=10, bias=True)\n",
       "  )\n",
       ")\n",
       "Dataset ID......................: 554\n",
       "Dataset URL.....................: https://www.openml.org/d/554"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <a id='toc3_'></a>[Pretrained Image classification example - Transformer](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn\n",
    "import torch.optim\n",
    "\n",
    "import openml\n",
    "import openml_pytorch\n",
    "import openml_pytorch.layers\n",
    "import openml_pytorch.config\n",
    "import logging\n",
    "import warnings\n",
    "from torchvision.transforms import Compose, Resize, ToPILImage, ToTensor, Lambda\n",
    "from openml_pytorch.trainer import convert_to_rgb\n",
    "# Suppress FutureWarning messages\n",
    "warnings.simplefilter(action='ignore')\n",
    "\n",
    "############################################################################\n",
    "# Enable logging in order to observe the progress while running the example.\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# openml.config.apikey = 'key'\n",
    "from openml_pytorch.trainer import OpenMLTrainerModule\n",
    "from openml_pytorch.trainer import OpenMLDataModule\n",
    "from openml_pytorch.trainer import Callback"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from openml import OpenMLTask\n",
    "def custom_optimizer_gen(model: torch.nn.Module, task: OpenMLTask) -> torch.optim.Optimizer:\n",
    "    return torch.optim.Adam(model.fc.parameters())\n",
    "\n",
    "############################################################################\n",
    "transform = Compose(\n",
    "    [\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(\n",
    "            convert_to_rgb\n",
    "        ),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize(\n",
    "            (64, 64)\n",
    "        ),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")\n",
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"image\",\n",
    "    file_dir=\"datasets\",\n",
    "    filename_col=\"image_path\",\n",
    "    target_mode=\"categorical\",\n",
    "    target_column=\"label\",\n",
    "    batch_size = 64,\n",
    "    transform=transform\n",
    ")\n",
    "# Download the OpenML task for tiniest imagenet\n",
    "task = openml.tasks.get_task(362128)\n",
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    data_module=data_module,\n",
    "    verbose = True,\n",
    "    epoch_count = 1,\n",
    "    optimizer = custom_optimizer_gen\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example model. You can do better :)\n",
    "import torchvision.models as models\n",
    "\n",
    "# Load the pre-trained ResNet model\n",
    "model = models.efficientnet_b0(pretrained=True)\n",
    "\n",
    "# Modify the last fully connected layer to the required number of classes\n",
    "num_classes = 200\n",
    "in_features = model.classifier[-1].in_features\n",
    "# model.fc = nn.Linear(in_features, num_classes)\n",
    "model.classifier = nn.Sequential(\n",
    "    nn.Dropout(p=0.2, inplace=True),\n",
    "    nn.Linear(in_features, num_classes),\n",
    ")\n",
    "\n",
    "# Optional: If you're fine-tuning, you may want to freeze the pre-trained layers\n",
    "# for param in model.parameters():\n",
    "#     param.requires_grad = False\n",
    "\n",
    "# # If you want to train the last layer only (the newly added layer)\n",
    "# for param in model.fc.parameters():\n",
    "#     param.requires_grad = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 0 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.486830873842592, tensor(0.0065, device='mps:0')]\n",
      "valid: [5.498468017578125, tensor(0.0083, device='mps:0')]\n",
      "Loss tensor(5.6186, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 1 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.534859664351852, tensor(0.0043, device='mps:0')]\n",
      "valid: [5.557652452256945, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.6452, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 2 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.5046585648148145, tensor(0.0059, device='mps:0')]\n",
      "valid: [5.664307657877604, tensor(0.0028, device='mps:0')]\n",
      "Loss tensor(5.8150, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 3 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.49383861400463, tensor(0.0046, device='mps:0')]\n",
      "valid: [5.487301974826389, tensor(0.0028, device='mps:0')]\n",
      "Loss tensor(5.5008, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 4 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.507443576388889, tensor(0.0062, device='mps:0')]\n",
      "valid: [5.3962646484375, tensor(0.0083, device='mps:0')]\n",
      "Loss tensor(5.3565, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 5 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.524650366512346, tensor(0.0052, device='mps:0')]\n",
      "valid: [5.57498541937934, tensor(0., device='mps:0')]\n",
      "Loss tensor(5.4743, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 6 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.530463927469135, tensor(0.0068, device='mps:0')]\n",
      "valid: [5.956352403428819, tensor(0.0083, device='mps:0')]\n",
      "Loss tensor(6.2169, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 7 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.506794945987655, tensor(0.0056, device='mps:0')]\n",
      "valid: [5.932855902777778, tensor(0., device='mps:0')]\n",
      "Loss tensor(6.6748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 8 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.516219979745371, tensor(0.0049, device='mps:0')]\n",
      "valid: [6.410263400607639, tensor(0.0056, device='mps:0')]\n",
      "Loss tensor(5.7348, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model EfficientNet(\n",
      "  (features): Sequential(\n",
      "    (0): Conv2dNormActivation(\n",
      "      (0): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "    (1): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=32, bias=False)\n",
      "            (1): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(32, 8, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(8, 32, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (2): Conv2dNormActivation(\n",
      "            (0): Conv2d(32, 16, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(16, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (2): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(16, 96, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 96, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=96, bias=False)\n",
      "            (1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(96, 4, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(4, 96, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(96, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0125, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 24, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.025, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (3): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(24, 144, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 144, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=144, bias=False)\n",
      "            (1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(144, 6, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(6, 144, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(144, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.037500000000000006, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 40, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(40, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.05, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (4): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(40, 240, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 240, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), groups=240, bias=False)\n",
      "            (1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(240, 10, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(10, 240, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(240, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.0625, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.07500000000000001, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 80, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(80, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.08750000000000001, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (5): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(80, 480, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 480, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=480, bias=False)\n",
      "            (1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(480, 20, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(20, 480, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(480, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1125, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 112, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.125, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (6): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(112, 672, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 672, kernel_size=(5, 5), stride=(2, 2), padding=(2, 2), groups=672, bias=False)\n",
      "            (1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(672, 28, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(28, 672, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1375, mode=row)\n",
      "      )\n",
      "      (1): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.15000000000000002, mode=row)\n",
      "      )\n",
      "      (2): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1625, mode=row)\n",
      "      )\n",
      "      (3): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.17500000000000002, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (7): Sequential(\n",
      "      (0): MBConv(\n",
      "        (block): Sequential(\n",
      "          (0): Conv2dNormActivation(\n",
      "            (0): Conv2d(192, 1152, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (1): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 1152, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), groups=1152, bias=False)\n",
      "            (1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "            (2): SiLU(inplace=True)\n",
      "          )\n",
      "          (2): SqueezeExcitation(\n",
      "            (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "            (fc1): Conv2d(1152, 48, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (fc2): Conv2d(48, 1152, kernel_size=(1, 1), stride=(1, 1))\n",
      "            (activation): SiLU(inplace=True)\n",
      "            (scale_activation): Sigmoid()\n",
      "          )\n",
      "          (3): Conv2dNormActivation(\n",
      "            (0): Conv2d(1152, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "            (1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "          )\n",
      "        )\n",
      "        (stochastic_depth): StochasticDepth(p=0.1875, mode=row)\n",
      "      )\n",
      "    )\n",
      "    (8): Conv2dNormActivation(\n",
      "      (0): Conv2d(320, 1280, kernel_size=(1, 1), stride=(1, 1), bias=False)\n",
      "      (1): BatchNorm2d(1280, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
      "      (2): SiLU(inplace=True)\n",
      "    )\n",
      "  )\n",
      "  (avgpool): AdaptiveAvgPool2d(output_size=1)\n",
      "  (classifier): Sequential(\n",
      "    (0): Dropout(p=0.2, inplace=True)\n",
      "    (1): Linear(in_features=1280, out_features=200, bias=True)\n",
      "  )\n",
      ") on dataset tiniest-imagenet-200 for repeat 0 fold 9 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [5.494152681327161, tensor(0.0065, device='mps:0')]\n",
      "valid: [5.389304606119792, tensor(0.0083, device='mps:0')]\n",
      "Loss tensor(5.4360, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:openml.config:Executed Task 362128 on local Flow with name torch.nn.EfficientNet.e18645453fee0ec6.\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "# Run the model on the task (requires an API key).m\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_lr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tabular classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import torch.nn\n",
    "import torch.optim\n",
    "\n",
    "import openml\n",
    "import openml_pytorch\n",
    "import openml_pytorch.layers\n",
    "import openml_pytorch.config\n",
    "import logging\n",
    "\n",
    "\n",
    "############################################################################\n",
    "# Enable logging in order to observe the progress while running the example.\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "############################################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "from openml_pytorch.trainer import OpenMLTrainerModule\n",
    "from openml_pytorch.trainer import OpenMLDataModule\n",
    "from openml_pytorch.trainer import Callback\n",
    "import torchvision"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# supervised credit-g classification\n",
    "task = openml.tasks.get_task(31)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"dataframe\",\n",
    "    target_column=\"class\",\n",
    "    target_mode=\"categorical\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    data_module=data_module,\n",
    "    verbose = True,\n",
    "    epoch_count = 5,\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = task.get_dataset().get_data()[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TabularClassificationmodel(torch.nn.Module):\n",
    "    def __init__(self, input_size, output_size):\n",
    "        super(TabularClassificationmodel, self).__init__()\n",
    "        self.fc1 = torch.nn.Linear(input_size, 128)\n",
    "        self.fc2 = torch.nn.Linear(128, 64)\n",
    "        self.fc3 = torch.nn.Linear(64, output_size)\n",
    "        self.relu = torch.nn.ReLU()\n",
    "        self.softmax = torch.nn.Softmax(dim=1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.fc1(x)\n",
    "        x = self.relu(x)\n",
    "        x = self.fc2(x)\n",
    "        x = self.relu(x)\n",
    "        x = self.fc3(x)\n",
    "        x = self.softmax(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = TabularClassificationmodel(20, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 0 sample 0\n",
      "                                               \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 1 sample 0\n",
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 2 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                              \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 3 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 4 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 5 sample 0\n",
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 6 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 7 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 8 sample 0\n",
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DEBUG:openml.config:Requested to create log handlers, but they are already created.\n",
      "INFO:openml.config:Going to run model TabularClassificationmodel(\n",
      "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
      "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
      "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
      "  (relu): ReLU()\n",
      "  (softmax): Softmax(dim=1)\n",
      ") on dataset credit-g for repeat 0 fold 9 sample 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                      \r"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:openml.config:Executed Task 31 on local Flow with name torch.nn.TabularClassificationmodel.481f5d46717292fa.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: [0.6132615454402971, tensor(0.7000, device='mps:0')]\n",
      "valid: [0.6132616255018446, tensor(0.7000, device='mps:0')]\n",
      "Loss tensor(0.7748, device='mps:0')\n"
     ]
    }
   ],
   "source": [
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OpenML Run\n",
       "==========\n",
       "Uploader Name...................: None\n",
       "Metric..........................: None\n",
       "Local Result - Accuracy (+- STD): 0.7000 +- 0.0000\n",
       "Run ID..........................: 10595160\n",
       "Run URL.........................: https://www.openml.org/r/10595160\n",
       "Task ID.........................: 31\n",
       "Task Type.......................: None\n",
       "Task URL........................: https://www.openml.org/t/31\n",
       "Flow ID.........................: 25096\n",
       "Flow Name.......................: torch.nn.TabularClassificationmodel.481f5d46717292fa\n",
       "Flow URL........................: https://www.openml.org/f/25096\n",
       "Setup ID........................: None\n",
       "Setup String....................: Python_3.11.9. Torch_2.4.1. NumPy_1.26.4. SciPy_1.14.1. TabularClassificationmodel(\n",
       "  (fc1): Linear(in_features=20, out_features=128, bias=True)\n",
       "  (fc2): Linear(in_features=128, out_features=64, bias=True)\n",
       "  (fc3): Linear(in_features=64, out_features=2, bias=True)\n",
       "  (relu): ReLU()\n",
       "  (softmax): Softmax(dim=1)\n",
       ")\n",
       "Dataset ID......................: 31\n",
       "Dataset URL.....................: https://www.openml.org/d/31"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_lr()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "openmlpytorch",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}