From 17145ba62996367f60c9bc6d1d84a668c78810c6 Mon Sep 17 00:00:00 2001
From: Akira <tidemark0105@gmail.com>
Date: Tue, 16 Jan 2024 00:08:38 +0900
Subject: [PATCH] adding convlstm exampes

---
 .gitignore                        |  1 +
 README.md                         | 13 ++++--
 examples/__init__.py              |  0
 examples/moving_mnist_convlstm.py | 68 +++++++++++++++++++++++++++++++
 4 files changed, 78 insertions(+), 4 deletions(-)
 create mode 100644 examples/__init__.py
 create mode 100644 examples/moving_mnist_convlstm.py

diff --git a/.gitignore b/.gitignore
index 781d569..553baaa 100644
--- a/.gitignore
+++ b/.gitignore
@@ -165,3 +165,4 @@ poetry.lock
 
 MovingMNIST/
 tmp/
+*.pt
diff --git a/README.md b/README.md
index ae20d35..e0868b1 100644
--- a/README.md
+++ b/README.md
@@ -7,21 +7,26 @@
 
 ## Examples
 
+```bash
+python -m examples.moving_mnist_convlstm
+  
+```
+
 ## Directories
 
-## `convlstm/`
+### `convlstm/`
 
 ConvLSTM implementation based on [Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting](https://paperswithcode.com/paper/convolutional-lstm-network-a-machine-learning).
 
-## `self_attention_convlstm/`
+### `self_attention_convlstm/`
 
 Self Attention ConvLSTM implementation based on [Self-Attention ConvLSTM for Spatiotemporal Prediction](https://ojs.aaai.org/index.php/AAAI/article/view/6819/6673).
 
-## `self_attention_memory_convlstm/`
+### `self_attention_memory_convlstm/`
 
 Self-Attention ConvLSTM (with memory module) implementation based on [Self-Attention ConvLSTM for Spatiotemporal Prediction](https://ojs.aaai.org/index.php/AAAI/article/view/6819/6673).
 
-## Visualized Attention Maps
+### Visualized Attention Maps
 
 ![sa-convlstm](fig/sa-convlstm.png)
 
diff --git a/examples/__init__.py b/examples/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/examples/moving_mnist_convlstm.py b/examples/moving_mnist_convlstm.py
new file mode 100644
index 0000000..b9e9c13
--- /dev/null
+++ b/examples/moving_mnist_convlstm.py
@@ -0,0 +1,68 @@
+from torch import nn
+from torch.optim import Adam
+
+from convlstm.seq2seq import Seq2Seq, Seq2SeqParams
+from core.constants import WeightsInitializer
+from data_loaders.moving_mnist import MovingMNISTDataLoaders
+from pipelines.experimenter import Experimenter
+from pipelines.trainer import TrainingParams
+from pipelines.utils.early_stopping import EarlyStopping
+
+
+def main():
+    ###
+    # Common Params
+    ###
+    artifact_dir = "./tmp"
+    input_seq_length = 10
+    train_batch_size = 32
+    validation_bath_size = 16
+    ###
+    # Setup Pipeline
+    ###
+    model_params: Seq2SeqParams = {
+        "input_seq_length": input_seq_length,
+        "num_layers": 2,
+        "num_kernels": 64,
+        "return_sequences": False,
+        "convlstm_params": {
+            "in_channels": 1,
+            "out_channels": 1,
+            "kernel_size": (3, 3),
+            "padding": "same",
+            "activation": "relu",
+            "frame_size": (64, 64),
+            "weights_initializer": WeightsInitializer.He,
+        },
+    }
+
+    model = Seq2Seq(**model_params)
+
+    training_params: TrainingParams = {
+        "epochs": 1,
+        "loss_criterion": nn.BCELoss(reduction="sum"),
+        "accuracy_criterion": nn.L1Loss(),
+        "optimizer": Adam(model.parameters(), lr=1e-4),
+        "early_stopping": EarlyStopping(
+            patience=30,
+            verbose=True,
+            delta=0.0001,
+        ),
+        "metrics_filename": "metrics.csv",
+    }
+
+    print("Loading dataset ...")
+    data_loaders = MovingMNISTDataLoaders(
+        train_batch_size=train_batch_size,
+        validation_batch_size=validation_bath_size,
+        input_frames=model_params["input_seq_length"],
+        label_frames=1,
+        split_ratios=[0.7, 0.299, 0.001],
+    )
+
+    experimenter = Experimenter(artifact_dir, data_loaders, model, training_params)
+    experimenter.run()
+
+
+if __name__ == "__main__":
+    main()