generator_transformer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import muspy\n",
    "\n",
    "# emopia = muspy.EMOPIADataset(\"data/emopia/\", download_and_extract=True)\n",
    "# emopia.convert()\n",
    "# music = emopia[0]\n",
    "# print(music.annotations[0].annotation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<torch._C.Generator at 0x7ff8fbb3a070>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "from data.process_data import MidiEncoder, MIDIEncoderREMI\n",
    "import pickle as pkl\n",
    "from torch.utils.data import DataLoader\n",
    "from data.dataset import TransformerDatasetREMI\n",
    "import numpy as np\n",
    "import glob\n",
    "import torch\n",
    "torch.manual_seed(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "path_to_midi = \"data/emopia/EMOPIA_2.2/midis/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate your MidiEncoder and MidiEncoderREMI\n",
    "midi_files_list = [os.path.join(path_to_midi, file) for file in os.listdir(path_to_midi) if file.endswith(\".mid\")]\n",
    "midi_encoder = MidiEncoder(steps_per_sec=100, num_vel_bins=32, min_pitch=21, max_pitch=108)\n",
    "midi_encoder_remi = MIDIEncoderREMI(dict_path=\"data/encoder_dict.pkl\", midi_files_list=midi_files_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data/emopia/EMOPIA_2.2/midis/Q1_9v2WSpn4FCw_10.mid\n",
      "0.03267407417297363\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_dtS02mrDMsM_1.mid\n",
      "0.04438495635986328\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_3ZnxqCZ7qGg_0.mid\n",
      "0.010029077529907227\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vpTguZtJAFA_2.mid\n",
      "0.021282196044921875\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_5.mid\n",
      "0.01467442512512207\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JP3QKZlyQz4_0.mid\n",
      "0.010650634765625\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Y5JcZQ0xg4Y_3.mid\n",
      "0.03576493263244629\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1Qc15G0ZHIg_3.mid\n",
      "0.029140710830688477\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ZgT7yq2jsBk_0.mid\n",
      "0.02819538116455078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1kny88W533Q_4.mid\n",
      "0.0304105281829834\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_xIsvaT20pZ0_1.mid\n",
      "0.0209348201751709\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_k-FNDbK6Qhg_2.mid\n",
      "0.1541440486907959\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_YAAxPW1GB7w_1.mid\n",
      "0.01327371597290039\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Q5b5unyP8BM_0.mid\n",
      "0.03690671920776367\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JxSU49jFKwM_1.mid\n",
      "0.014066934585571289\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_PK8YIUaV3Xw_1.mid\n",
      "0.013200759887695312\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Jn9r0avp0fY_3.mid\n",
      "0.0336766242980957\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TonQX8XbvX8_2.mid\n",
      "0.015815258026123047\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iHPKusssXzk_2.mid\n",
      "0.020525217056274414\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_POaIGvLsp5M_1.mid\n",
      "0.0173490047454834\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_eVMSeElk81Q_2.mid\n",
      "0.014029979705810547\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1vjy9oMFa8c_2.mid\n",
      "0.015770912170410156\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_1Q3MoBFh6eU_2.mid\n",
      "0.0200347900390625\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_HY9vPoHbgaI_1.mid\n",
      "0.0187070369720459\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_f2b4kpdw7_c_0.mid\n",
      "0.014348983764648438\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_3ahg_eQZhxs_1.mid\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.02319788932800293\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_f2b4kpdw7_c_1.mid\n",
      "0.01906728744506836\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_oTi57hZfArE_1.mid\n",
      "0.010872125625610352\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0H4rq0L9OSw_0.mid\n",
      "0.0477757453918457\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_3.mid\n",
      "0.02422308921813965\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_l-4Kce3YxxQ_3.mid\n",
      "0.03947329521179199\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_2.mid\n",
      "0.019085168838500977\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1EDThWjNxOI_1.mid\n",
      "0.013672590255737305\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_xrhWli_R98g_0.mid\n",
      "0.020205974578857422\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_fey-8bOR95E_0.mid\n",
      "0.03303861618041992\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_vv6nrZ2myXw_4.mid\n",
      "0.022354841232299805\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6Uf9XBUD3wE_0.mid\n",
      "0.01603841781616211\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zvWX_dwQPtU_0.mid\n",
      "0.017418384552001953\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_38i12IPkL5c_2.mid\n",
      "0.0343632698059082\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ivCNV47tsRw_0.mid\n",
      "0.03774428367614746\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_7JIdJLkJ0S4_3.mid\n",
      "0.02518486976623535\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_9v2WSpn4FCw_2.mid\n",
      "0.022332191467285156\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_10.mid\n",
      "0.023865461349487305\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Dg935IbDggo_0.mid\n",
      "0.03461575508117676\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kMHiU0fhWVw_1.mid\n",
      "0.021877050399780273\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_wzSr6qR3Ra0_0.mid\n",
      "0.017403841018676758\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lQCrA2TE-fo_0.mid\n",
      "0.01370096206665039\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_SQDuF0qxGQw_1.mid\n",
      "0.03101968765258789\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_jTPXwbDtIpA_0.mid\n",
      "0.013531208038330078\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Cg2u_Ldjv8g_0.mid\n",
      "0.023227691650390625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_DhnX54kOW8U_0.mid\n",
      "0.0296783447265625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_dtS02mrDMsM_2.mid\n",
      "0.0332794189453125\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jO2zX6Ul8GM_0.mid\n",
      "0.011593103408813477\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_im4Qxn3GQvo_1.mid\n",
      "0.02245020866394043\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_9Yb9OEVwups_0.mid\n",
      "0.0175015926361084\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_XC_SiJszQx0_1.mid\n",
      "0.02791309356689453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_vv6nrZ2myXw_3.mid\n",
      "0.020064592361450195\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UJcnNfeCN1c_1.mid\n",
      "0.0155487060546875\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0wRQRwnQiqY_0.mid\n",
      "0.024707794189453125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_AacIP9mDLw8_1.mid\n",
      "0.017148256301879883\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wfXSdMsd4q8_2.mid\n",
      "0.024832963943481445\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_im4Qxn3GQvo_0.mid\n",
      "0.023717880249023438\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pRuwHN-lI44_0.mid\n",
      "0.01848745346069336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_CxFXrYgdSSI_0.mid\n",
      "0.014703035354614258\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_4Er-eFHSHA4_3.mid\n",
      "0.021746158599853516\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ii1Jw1-7Ka4_1.mid\n",
      "0.018873214721679688\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_4.mid\n",
      "0.02309703826904297\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_fuCxYrru2S4_1.mid\n",
      "0.02156996726989746\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lTxVVdhFE6Q_0.mid\n",
      "0.020159244537353516\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0lEL_XHyhuU_2.mid\n",
      "0.012591838836669922\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_7gifuoofgpk_1.mid\n",
      "0.010950803756713867\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_DX1IK3z1w10_0.mid\n",
      "0.02025604248046875\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ZvD8QSO7NPw_2.mid\n",
      "0.048964500427246094\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_aYe-2Glruu4_4.mid\n",
      "0.04223775863647461\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UbKAl4roB8k_1.mid\n",
      "0.02564835548400879\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Oku8sIpTMAE_2.mid\n",
      "0.0050868988037109375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_qlbazHayULg_3.mid\n",
      "0.03388404846191406\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_9Xg69XrVaYU_2.mid\n",
      "0.02163100242614746\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jNzZCplNNyY_0.mid\n",
      "0.02176690101623535\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_fOYX0uH8mSQ_0.mid\n",
      "0.015030145645141602\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_X1p8xcarVu0_0.mid\n",
      "0.013941764831542969\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bQS8eAXAhyE_2.mid\n",
      "0.030132293701171875\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_H6_7DjsbROQ_3.mid\n",
      "0.03170967102050781\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_K6OFDxBU370_1.mid\n",
      "0.021013259887695312\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ADJ_PDauy-g_0.mid\n",
      "0.010095596313476562\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TonQX8XbvX8_1.mid\n",
      "0.015611410140991211\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yrvKReuwVlY_3.mid\n",
      "0.02451467514038086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_VXh4IaR5C6s_0.mid\n",
      "0.021411418914794922\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_9eMxCs-LXCE_2.mid\n",
      "0.02166461944580078\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_I2MrA-o5H8I_1.mid\n",
      "0.030054330825805664\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wlOa1jkTn_U_0.mid\n",
      "0.029903650283813477\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0v2N1ROvEI0_1.mid\n",
      "0.01955103874206543\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_tw_lArYiHTo_0.mid\n",
      "0.029898643493652344\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_8TYY0qG-KOw_0.mid\n",
      "0.0207369327545166\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_2.mid\n",
      "0.014037370681762695\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_eVMSeElk81Q_0.mid\n",
      "0.01665973663330078\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e838BE_MH6s_3.mid\n",
      "0.03359794616699219\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_4ydjOX3pWds_0.mid\n",
      "0.023152589797973633\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_By3JCii1k3w_2.mid\n",
      "0.013607263565063477\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Jn9r0avp0fY_1.mid\n",
      "0.0439455509185791\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_5.mid\n",
      "0.02643132209777832\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__kJtgm1OUNA_2.mid\n",
      "0.016275882720947266\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_HYVmgq5Y93g_0.mid\n",
      "0.019586563110351562\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ldCQ6N9G6Mk_2.mid\n",
      "0.009305715560913086\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ilCcZpD39HU_1.mid\n",
      "0.01329803466796875\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_EtAbPc20mn8_0.mid\n",
      "0.017735719680786133\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_12.mid\n",
      "0.023153305053710938\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_xrhWli_R98g_3.mid\n",
      "0.018746376037597656\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Iq6g_4AwUWs_0.mid\n",
      "0.0352022647857666\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XPLYvpyfKUM_0.mid\n",
      "0.01308441162109375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_miqLU2739dk_0.mid\n",
      "0.028919458389282227\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_TcFVq0leii4_0.mid\n",
      "0.01508641242980957\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cFWdWpyvwag_0.mid\n",
      "0.02149677276611328\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_BaHHHgDd0BU_0.mid\n",
      "0.02330470085144043\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ISCVfmBvYoM_0.mid\n",
      "0.021205902099609375\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_7.mid\n",
      "0.019290924072265625\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KxlqB3j0zys_4.mid\n",
      "0.013355731964111328\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_doM_LWo-74A_1.mid\n",
      "0.03217124938964844\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_nBIls0laAAU_2.mid\n",
      "0.030531883239746094\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_VXgRfsVd_o0_2.mid\n",
      "0.026664018630981445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pzeHffxmdcE_0.mid\n",
      "0.02460503578186035\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_FfwKrQyQ7WU_3.mid\n",
      "0.02036738395690918\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_AjN_5CdVUdw_0.mid\n",
      "0.015820980072021484\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OgkB7PfqWzA_1.mid\n",
      "0.010307073593139648\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tc684nD8CEA_1.mid\n",
      "0.012668609619140625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jO2zX6Ul8GM_1.mid\n",
      "0.013913154602050781\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UbKAl4roB8k_2.mid\n",
      "0.01844310760498047\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_8izVTDgBQPc_0.mid\n",
      "0.023288488388061523\n",
      "data/emopia/EMOPIA_2.2/midis/Q3__vZOEQCYSaY_2.mid\n",
      "0.027747392654418945\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ZvD8QSO7NPw_0.mid\n",
      "0.01268625259399414\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Jn9r0avp0fY_2.mid\n",
      "0.033779144287109375\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Q49PWE0RJLY_0.mid\n",
      "0.02425694465637207\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_D8Ed5PZXfF4_0.mid\n",
      "0.016022682189941406\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_E6MARl3zjnQ_0.mid\n",
      "0.028652667999267578\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_pouxqiySdI8_1.mid\n",
      "0.024605751037597656\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_JNzqL24ZcZ8_2.mid\n",
      "0.021144866943359375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_rrD7XMtPs0Q_0.mid\n",
      "0.01704692840576172\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_eGQUZ9FlUJ4_0.mid\n",
      "0.03742790222167969\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rW1I58_lMEg_1.mid\n",
      "0.015704870223999023\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_KxlqB3j0zys_1.mid\n",
      "0.012321233749389648\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jCLgDtk695I_4.mid\n",
      "0.01350545883178711\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_7n32BtYb4d4_0.mid\n",
      "0.15372562408447266\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_08t73Qgjkt4_2.mid\n",
      "0.04218173027038574\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_V3Y9L4UOcpk_1.mid\n",
      "0.021821975708007812\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_7n32BtYb4d4_2.mid\n",
      "0.033577680587768555\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_HY9vPoHbgaI_2.mid\n",
      "0.020380735397338867\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UYh88SRZC24_0.mid\n",
      "0.030426740646362305\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0lEL_XHyhuU_3.mid\n",
      "0.008795738220214844\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_qlbazHayULg_9.mid\n",
      "0.015145301818847656\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GY3f6ckBVkA_0.mid\n",
      "0.0177154541015625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jrVXitpLGMk_2.mid\n",
      "0.0446927547454834\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_tfuul78deOs_1.mid\n",
      "0.01639699935913086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_1.mid\n",
      "0.01403665542602539\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Hwd6l9jK7bo_1.mid\n",
      "0.012140274047851562\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_0.mid\n",
      "0.019736528396606445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vBHoSz8cJIE_3.mid\n",
      "0.020656347274780273\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UJcnNfeCN1c_2.mid\n",
      "0.015546083450317383\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XPLYvpyfKUM_1.mid\n",
      "0.018095970153808594\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_DYBBfpEx-JI_0.mid\n",
      "0.030568838119506836\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rW1I58_lMEg_2.mid\n",
      "0.01745462417602539\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_gwmVjvR-sVs_1.mid\n",
      "0.017636775970458984\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Uu-oRTRT6uQ_0.mid\n",
      "0.03790760040283203\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FSzOgraSKhs_1.mid\n",
      "0.016026735305786133\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pRuwHN-lI44_3.mid\n",
      "0.015271663665771484\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TYkTOwBfFB8_0.mid\n",
      "0.022231340408325195\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_TNoaOTrTkTQ_2.mid\n",
      "0.017415761947631836\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_doM_LWo-74A_2.mid\n",
      "0.023511171340942383\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UJcnNfeCN1c_3.mid\n",
      "0.015276193618774414\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_NP0lwB-_-og_0.mid\n",
      "0.014503955841064453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_5NW0zDu6IYM_2.mid\n",
      "0.05845212936401367\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tc684nD8CEA_2.mid\n",
      "0.015099287033081055\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lvfNo4KOTmM_1.mid\n",
      "0.013983726501464844\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ZvD8QSO7NPw_1.mid\n",
      "0.026231765747070312\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_4.mid\n",
      "0.05327272415161133\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UYh88SRZC24_1.mid\n",
      "0.03542327880859375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_yZGBxiNPu3k_2.mid\n",
      "0.02498483657836914\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_6.mid\n",
      "0.10467290878295898\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_AacIP9mDLw8_0.mid\n",
      "0.021152257919311523\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_PZN7XLKzLDI_0.mid\n",
      "0.02729320526123047\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_5.mid\n",
      "0.03371906280517578\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_l-4Kce3YxxQ_1.mid\n",
      "0.04287552833557129\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Hwd6l9jK7bo_0.mid\n",
      "0.013434171676635742\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_OO7zYVeK814_1.mid\n",
      "0.017437219619750977\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_0.mid\n",
      "0.013621330261230469\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_hoPp_GLXQis_1.mid\n",
      "0.014218807220458984\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_2Z9SjI131jA_6.mid\n",
      "0.019286394119262695\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Ie5koh4qvJc_12.mid\n",
      "0.07465577125549316\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3N2G21U7guk_1.mid\n",
      "0.01484370231628418\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_D3onzY4MNxE_0.mid\n",
      "0.01453256607055664\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_9v2WSpn4FCw_0.mid\n",
      "0.018184900283813477\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_aQ4PIaRMLwE_0.mid\n",
      "0.010393857955932617\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_0.mid\n",
      "0.06300544738769531\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lvfNo4KOTmM_0.mid\n",
      "0.015413284301757812\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_heBmBQDWj-M_2.mid\n",
      "0.013103723526000977\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_p2qihp_PnLY_1.mid\n",
      "0.025304794311523438\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Xsn9zT-05ns_1.mid\n",
      "0.04117608070373535\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_0.mid\n",
      "0.023484230041503906\n",
      "data/emopia/EMOPIA_2.2/midis/Q2__vZOEQCYSaY_1.mid\n",
      "0.018209457397460938\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_6oLmO9ZSJt0_1.mid\n",
      "0.007556438446044922\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_61EA0xRX8gE_2.mid\n",
      "0.021797657012939453\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ezoAduuCkUs_0.mid\n",
      "0.04362988471984863\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_jziI_cuN_H8_1.mid\n",
      "0.03161478042602539\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_7.mid\n",
      "0.02888655662536621\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_TcFVq0leii4_2.mid\n",
      "0.04236459732055664\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_f2BXXa7lPuE_3.mid\n",
      "0.039191246032714844\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_1kny88W533Q_1.mid\n",
      "0.014436483383178711\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_JT1XJnVmABo_6.mid\n",
      "0.03381538391113281\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yrvKReuwVlY_2.mid\n",
      "0.04930853843688965\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_prUUays30Mg_1.mid\n",
      "0.011603355407714844\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Xsn9zT-05ns_0.mid\n",
      "0.03316640853881836\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bIx4DwkZxFY_0.mid\n",
      "0.014056682586669922\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_k6aLLVX1D7Y_1.mid\n",
      "0.0261993408203125\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Cg2u_Ldjv8g_1.mid\n",
      "0.02688431739807129\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0J7lX-Tcx-w_2.mid\n",
      "0.028176546096801758\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_1.mid\n",
      "0.19663095474243164\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_CxFXrYgdSSI_2.mid\n",
      "0.05733180046081543\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kfYaGryaGzI_0.mid\n",
      "0.01399993896484375\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vpTguZtJAFA_1.mid\n",
      "0.009949922561645508\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_SQDuF0qxGQw_2.mid\n",
      "0.0294492244720459\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ltxNPJda7zE_2.mid\n",
      "0.017607927322387695\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_RiQMuhk_SuQ_0.mid\n",
      "0.008322477340698242\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tc684nD8CEA_3.mid\n",
      "0.017477035522460938\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_CxFXrYgdSSI_1.mid\n",
      "0.035668373107910156\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_K6OFDxBU370_0.mid\n",
      "0.01990056037902832\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zjr_-ql5Avk_1.mid\n",
      "0.00874948501586914\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_x4B0raxVODA_2.mid\n",
      "0.011646747589111328\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FOcdHIhI0_s_1.mid\n",
      "0.014681577682495117\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_PLfFWFZflQU_0.mid\n",
      "0.01952528953552246\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_bbU31JLtlug_1.mid\n",
      "0.0357670783996582\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__kJtgm1OUNA_0.mid\n",
      "0.02048778533935547\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_-_FjI4bT2Wo_0.mid\n",
      "0.022528409957885742\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_77z6Ep3aOmg_0.mid\n",
      "0.019437789916992188\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_g15aRCHUJEI_0.mid\n",
      "0.01263570785522461\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5J_zI7QE9W8_2.mid\n",
      "0.0200345516204834\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lQCrA2TE-fo_1.mid\n",
      "0.013975381851196289\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_WBxql6cuU7c_0.mid\n",
      "0.040450096130371094\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pRuwHN-lI44_2.mid\n",
      "0.012715578079223633\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_-jJVb0xvbdg_0.mid\n",
      "0.0184018611907959\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_jziI_cuN_H8_2.mid\n",
      "0.042517900466918945\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_1.mid\n",
      "0.01187586784362793\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_-vAz_HTFEXs_0.mid\n",
      "0.015336036682128906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TYkTOwBfFB8_1.mid\n",
      "0.025318384170532227\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_qRceKZtwRWA_0.mid\n",
      "0.011829853057861328\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_sUJ4a74Skk8_0.mid\n",
      "0.017545461654663086\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_1cgO42sh-ZY_0.mid\n",
      "0.008828401565551758\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_X1p8xcarVu0_2.mid\n",
      "0.012773752212524414\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e8NQ2NH0nc8_1.mid\n",
      "0.02086806297302246\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ilCcZpD39HU_0.mid\n",
      "0.013802289962768555\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wS0RzrBL9UI_0.mid\n",
      "0.013694524765014648\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_iwxwLEvqeD8_0.mid\n",
      "0.010201215744018555\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_b8HVQtIoBYU_1.mid\n",
      "0.029644012451171875\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_7n32BtYb4d4_3.mid\n",
      "0.03047943115234375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_kDGmND1BgmA_1.mid\n",
      "0.006482124328613281\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UAm0aWvzFI8_1.mid\n",
      "0.0352933406829834\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_DvIH5yUrboc_1.mid\n",
      "0.029935121536254883\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_r_sD61KeUQU_0.mid\n",
      "0.019701004028320312\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_8.mid\n",
      "0.055329084396362305\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_g-yM0Lsp4lc_0.mid\n",
      "0.01553487777709961\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cQp1BYDGcRo_1.mid\n",
      "0.03511333465576172\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_PLfFWFZflQU_1.mid\n",
      "0.02804255485534668\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_DYBBfpEx-JI_3.mid\n",
      "0.03449130058288574\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ury1cdB79s0_0.mid\n",
      "0.01159811019897461\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_15.mid\n",
      "0.014508962631225586\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_H6_7DjsbROQ_0.mid\n",
      "0.027251720428466797\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_vVTP0DOL_2Q_1.mid\n",
      "0.014670133590698242\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Kk60F8a7-Jw_0.mid\n",
      "0.033757925033569336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2__FVzelKlBFs_2.mid\n",
      "0.02233576774597168\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jCLgDtk695I_0.mid\n",
      "0.011476278305053711\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_2.mid\n",
      "0.041574954986572266\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_UJcnNfeCN1c_0.mid\n",
      "0.012511491775512695\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_02t1bj7ZABY_1.mid\n",
      "0.015856027603149414\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_12tDlCQVRtA_1.mid\n",
      "0.01815176010131836\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Xf2fox7zhCQ_0.mid\n",
      "0.01644587516784668\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6kRPHamGDSo_3.mid\n",
      "0.018971681594848633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KRlAez34QfE_1.mid\n",
      "0.017079591751098633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_-_FjI4bT2Wo_2.mid\n",
      "0.017414569854736328\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Uu-oRTRT6uQ_1.mid\n",
      "0.04368281364440918\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_12.mid\n",
      "0.03311324119567871\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_RiQMuhk_SuQ_2.mid\n",
      "0.08267593383789062\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wqc8iqbDsGM_2.mid\n",
      "0.021039247512817383\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_VXgRfsVd_o0_1.mid\n",
      "0.040708065032958984\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Dg935IbDggo_1.mid\n",
      "0.0372776985168457\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_T92R7xjce34_1.mid\n",
      "0.0180666446685791\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_x4B0raxVODA_1.mid\n",
      "0.007962465286254883\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5DqIntcmdRI_1.mid\n",
      "0.020210981369018555\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_epX33OVpkmA_1.mid\n",
      "0.0416102409362793\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_9.mid\n",
      "0.039861440658569336\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltxNPJda7zE_1.mid\n",
      "0.01148843765258789\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_d_49EtXDMFE_1.mid\n",
      "0.012009859085083008\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_94ROf10d1iA_0.mid\n",
      "0.011781930923461914\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_vv6nrZ2myXw_2.mid\n",
      "0.017958879470825195\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_VrWAZLRdQO8_0.mid\n",
      "0.008085489273071289\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_xrhWli_R98g_1.mid\n",
      "0.03239083290100098\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iGHgeUCPotc_0.mid\n",
      "0.015141725540161133\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Y5JcZQ0xg4Y_1.mid\n",
      "0.02164769172668457\n",
      "data/emopia/EMOPIA_2.2/midis/Q3__vZOEQCYSaY_0.mid\n",
      "0.02140974998474121\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_l-4Kce3YxxQ_0.mid\n",
      "0.0435030460357666\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_78C_TpQDvVE_1.mid\n",
      "0.023138046264648438\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XC_SiJszQx0_0.mid\n",
      "0.013248682022094727\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_veG92Oi-DlU_0.mid\n",
      "0.020359277725219727\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_dfNdpy8TUzA_1.mid\n",
      "0.0161893367767334\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_fShAWhXIits_1.mid\n",
      "0.01618671417236328\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ZTrEoB8T9YA_0.mid\n",
      "0.03255486488342285\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_17.mid\n",
      "0.011849164962768555\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_TNoaOTrTkTQ_1.mid\n",
      "0.022306442260742188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yhIncZ_Ue4E_0.mid\n",
      "0.018192529678344727\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_7yW9c7t8Hq0_3.mid\n",
      "0.04304862022399902\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_31.mid\n",
      "0.01664900779724121\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_XfA2KXodrOE_0.mid\n",
      "0.03225350379943848\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_BGf5nCdzPOc_0.mid\n",
      "0.044531822204589844\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bQS8eAXAhyE_0.mid\n",
      "0.01602339744567871\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_DYBBfpEx-JI_2.mid\n",
      "0.036210060119628906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zvWX_dwQPtU_1.mid\n",
      "0.023058414459228516\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_mX-xs3OVhTs_1.mid\n",
      "0.017551660537719727\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_QwsQ8ejbMKg_0.mid\n",
      "0.035428762435913086\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_3N2G21U7guk_5.mid\n",
      "0.022418975830078125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_QJlnTN7HRwE_0.mid\n",
      "0.009322404861450195\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ldCQ6N9G6Mk_5.mid\n",
      "0.14975309371948242\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_4Er-eFHSHA4_1.mid\n",
      "0.019088029861450195\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_DvIH5yUrboc_0.mid\n",
      "0.025742769241333008\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_L3_TadjSQQ0_0.mid\n",
      "0.0482485294342041\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UOSlDydo94E_0.mid\n",
      "0.01927804946899414\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__SJQaaRzD-A_1.mid\n",
      "0.02475118637084961\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_b8HVQtIoBYU_0.mid\n",
      "0.018715858459472656\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1EDThWjNxOI_0.mid\n",
      "0.012134075164794922\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Oku8sIpTMAE_0.mid\n",
      "0.005368471145629883\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_2w5XkAt_3WU_0.mid\n",
      "0.011853694915771484\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Dw7M69CidJ8_0.mid\n",
      "0.011422872543334961\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_WQrPcuy50aA_1.mid\n",
      "0.030360698699951172\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_vVTP0DOL_2Q_3.mid\n",
      "0.021236181259155273\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_mlHheZbblT0_1.mid\n",
      "0.023216724395751953\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1Qc15G0ZHIg_1.mid\n",
      "0.025957107543945312\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1vjy9oMFa8c_4.mid\n",
      "0.012287616729736328\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_L3_TadjSQQ0_1.mid\n",
      "0.0390477180480957\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_61EA0xRX8gE_4.mid\n",
      "0.02434229850769043\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_AacIP9mDLw8_2.mid\n",
      "0.024717330932617188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_l-4Kce3YxxQ_4.mid\n",
      "0.04142403602600098\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GY3f6ckBVkA_1.mid\n",
      "0.014547348022460938\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_f2BXXa7lPuE_0.mid\n",
      "0.021674394607543945\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KRlAez34QfE_2.mid\n",
      "0.01544046401977539\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_JNzqL24ZcZ8_0.mid\n",
      "0.022713899612426758\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_epX33OVpkmA_2.mid\n",
      "0.032058000564575195\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_s6-SbDSZzEU_0.mid\n",
      "0.020465612411499023\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_fnqb8zqHqdY_1.mid\n",
      "0.010112285614013672\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pkz8OkWhwnk_1.mid\n",
      "0.01657724380493164\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FH28sPhm9Dc_1.mid\n",
      "0.025762557983398438\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_POaIGvLsp5M_0.mid\n",
      "0.016265869140625\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_WpZ1rWHOCAQ_0.mid\n",
      "0.01437234878540039\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_cXxGBDgFCs8_0.mid\n",
      "0.020853042602539062\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_7yW9c7t8Hq0_2.mid\n",
      "0.02111053466796875\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_fShAWhXIits_0.mid\n",
      "0.03458428382873535\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_dfNdpy8TUzA_2.mid\n",
      "0.017712116241455078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UTo7E0evQWw_0.mid\n",
      "0.02023029327392578\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_61EA0xRX8gE_1.mid\n",
      "0.03225827217102051\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_94ROf10d1iA_2.mid\n",
      "0.006329536437988281\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_snfsc7pEzlk_1.mid\n",
      "0.03685808181762695\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_z5ewpQwGDNw_0.mid\n",
      "0.015154838562011719\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_FUAK5TBaNY8_1.mid\n",
      "0.022530317306518555\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0H4rq0L9OSw_1.mid\n",
      "0.05353903770446777\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_S7Bo4-NCEDk_1.mid\n",
      "0.03927445411682129\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_REq37pDAm3A_3.mid\n",
      "0.024233579635620117\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltNK_MY1HkM_2.mid\n",
      "0.006232500076293945\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_yZGBxiNPu3k_1.mid\n",
      "0.016615629196166992\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ZdyiXIlKe_Y_1.mid\n",
      "0.02157759666442871\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_BzqX-9TA-GY_0.mid\n",
      "0.004341840744018555\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_sA9ac1rod84_0.mid\n",
      "0.020302534103393555\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_PdwxeMPXOCk_1.mid\n",
      "0.02786707878112793\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_cCfEDTkuMvg_0.mid\n",
      "0.010910511016845703\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_d_49EtXDMFE_2.mid\n",
      "0.01091909408569336\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OoxI6k0Qsno_0.mid\n",
      "0.011130809783935547\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_TcFVq0leii4_1.mid\n",
      "0.03612065315246582\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0lEL_XHyhuU_0.mid\n",
      "0.00822758674621582\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wxXpvnmq_-E_2.mid\n",
      "0.01168966293334961\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_JT1XJnVmABo_3.mid\n",
      "0.010805130004882812\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KRlAez34QfE_0.mid\n",
      "0.015827655792236328\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tc684nD8CEA_0.mid\n",
      "0.013695478439331055\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D-gDPh9ZpFk_2.mid\n",
      "0.03503608703613281\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D5kyjnlDNZs_0.mid\n",
      "0.03246951103210449\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tc684nD8CEA_4.mid\n",
      "0.01523280143737793\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_4dXC1cC7crw_1.mid\n",
      "0.02914118766784668\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OUb9uaOlWAM_0.mid\n",
      "0.009826421737670898\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cFWdWpyvwag_1.mid\n",
      "0.055533409118652344\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_4Er-eFHSHA4_2.mid\n",
      "0.01926565170288086\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e8NQ2NH0nc8_0.mid\n",
      "0.022043466567993164\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_V_jbWvQNh5c_1.mid\n",
      "0.009764671325683594\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_snfsc7pEzlk_0.mid\n",
      "0.016889572143554688\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yrvKReuwVlY_0.mid\n",
      "0.01612114906311035\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_9eMxCs-LXCE_0.mid\n",
      "0.008817434310913086\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ldCQ6N9G6Mk_4.mid\n",
      "0.019922494888305664\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0EWWyyV7RJA_0.mid\n",
      "0.02476048469543457\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UpHutdJvZMI_0.mid\n",
      "0.012812614440917969\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__8v0MFBZoco_0.mid\n",
      "0.05359148979187012\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_v3Nl5rUxLqE_1.mid\n",
      "0.007315397262573242\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OoxI6k0Qsno_1.mid\n",
      "0.01096343994140625\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_H6_7DjsbROQ_1.mid\n",
      "0.019832849502563477\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_pNwQ9Tu_bCs_1.mid\n",
      "0.023314714431762695\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5J_zI7QE9W8_1.mid\n",
      "0.015281438827514648\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ijfxkQPAQdE_2.mid\n",
      "0.05411982536315918\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_MT2_mXn-vGU_0.mid\n",
      "0.015339136123657227\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_VCSDjQxAJzc_1.mid\n",
      "0.017185449600219727\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6wFJhmhNeeg_0.mid\n",
      "0.05155754089355469\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_xYZ8n8ULaNo_1.mid\n",
      "0.01740288734436035\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_3.mid\n",
      "0.015209436416625977\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3TYeU9idRGI_0.mid\n",
      "0.008874893188476562\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Xlv4PPGrRRs_1.mid\n",
      "0.017066001892089844\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_JMt4m608s3k_0.mid\n",
      "0.025246858596801758\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1TAKQK3s6qQ_0.mid\n",
      "0.04494524002075195\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_3.mid\n",
      "0.03537917137145996\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kMHiU0fhWVw_2.mid\n",
      "0.01601719856262207\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Jn9r0avp0fY_0.mid\n",
      "0.03957867622375488\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_cXxGBDgFCs8_2.mid\n",
      "0.015720129013061523\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_x5ibvz38jOs_0.mid\n",
      "0.02308797836303711\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_xYZ8n8ULaNo_0.mid\n",
      "0.014403343200683594\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3N2G21U7guk_4.mid\n",
      "0.014656543731689453\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_o20YCzej5_s_0.mid\n",
      "0.04887223243713379\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FyK_c-TIcCA_0.mid\n",
      "0.029451370239257812\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_mGYKWCVnMwQ_1.mid\n",
      "0.024910688400268555\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Fc1qk52SaKY_0.mid\n",
      "0.029920578002929688\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_BzqX-9TA-GY_2.mid\n",
      "0.014051675796508789\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_OCqMDeD6Fmc_0.mid\n",
      "0.023710966110229492\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_s5b8viFsVJE_1.mid\n",
      "0.037758827209472656\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Cir1kZyB7QM_0.mid\n",
      "0.016867637634277344\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_21.mid\n",
      "0.035331010818481445\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1kny88W533Q_2.mid\n",
      "0.036499977111816406\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_wfXSdMsd4q8_5.mid\n",
      "0.02942824363708496\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jSWItJI-Gmk_1.mid\n",
      "0.01567387580871582\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_7n32BtYb4d4_1.mid\n",
      "0.01835918426513672\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_eGQUZ9FlUJ4_1.mid\n",
      "0.0391383171081543\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_E5qEloUO3SM_0.mid\n",
      "0.020305633544921875\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wlOa1jkTn_U_1.mid\n",
      "0.02272772789001465\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_TOXg4-ZGb-g_2.mid\n",
      "0.016067028045654297\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_pNwQ9Tu_bCs_2.mid\n",
      "0.024777889251708984\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_tfuul78deOs_0.mid\n",
      "0.015386581420898438\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_5.mid\n",
      "0.034380435943603516\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_k-FNDbK6Qhg_1.mid\n",
      "0.0223846435546875\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XWk5XUEBXmg_0.mid\n",
      "0.014729976654052734\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zHwJG8Rrx6c_1.mid\n",
      "0.017096519470214844\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_qlbazHayULg_13.mid\n",
      "0.03606128692626953\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_6.mid\n",
      "0.03737449645996094\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_RaUSISlnhKw_2.mid\n",
      "0.034880638122558594\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_16.mid\n",
      "0.005972623825073242\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_24.mid\n",
      "0.13988447189331055\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FoTXpYZXxJs_0.mid\n",
      "0.017604589462280273\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_gxsdWKc1QaM_0.mid\n",
      "0.025727272033691406\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_BIRDDxz0E4c_1.mid\n",
      "0.012253522872924805\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_12tDlCQVRtA_0.mid\n",
      "0.011548042297363281\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_aQ4PIaRMLwE_1.mid\n",
      "0.018389225006103516\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_SlsXsqotUis_0.mid\n",
      "0.03780722618103027\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Y1F0c3zgMJo_1.mid\n",
      "0.012170076370239258\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_--2B4d4lQhs_1.mid\n",
      "0.06860065460205078\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_s5b8viFsVJE_2.mid\n",
      "0.03336286544799805\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1kny88W533Q_3.mid\n",
      "0.031705379486083984\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_38i12IPkL5c_1.mid\n",
      "0.031899452209472656\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Xlv4PPGrRRs_0.mid\n",
      "0.014019250869750977\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_FfwKrQyQ7WU_1.mid\n",
      "0.021544456481933594\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_eVMSeElk81Q_1.mid\n",
      "0.0398867130279541\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_oTi57hZfArE_0.mid\n",
      "0.012375116348266602\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jIKX3ShvLxo_1.mid\n",
      "0.054917335510253906\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Gcbk0GPT3-g_0.mid\n",
      "0.018108606338500977\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_JYVXM0qNQAg_1.mid\n",
      "0.04210615158081055\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1fdxsFbnsX4_0.mid\n",
      "0.017331838607788086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_IgpNyJVWcBo_0.mid\n",
      "0.028363704681396484\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_X1p8xcarVu0_3.mid\n",
      "0.014800548553466797\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5Ju9q1N2x0E_1.mid\n",
      "0.03693032264709473\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_u_7b9CwyM8Q_0.mid\n",
      "0.01630687713623047\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_7.mid\n",
      "0.0363316535949707\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_cB-Xh8H_7-Q_0.mid\n",
      "0.038457393646240234\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_16OaOpfOPZ0_1.mid\n",
      "0.019273042678833008\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TonQX8XbvX8_0.mid\n",
      "0.008408546447753906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_egYSmNuIFGk_2.mid\n",
      "0.018537282943725586\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_J6X3rVU1H-c_1.mid\n",
      "0.02406907081604004\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_D3onzY4MNxE_1.mid\n",
      "0.01215815544128418\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_rEz0D3VohFA_2.mid\n",
      "0.04929780960083008\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0v2N1ROvEI0_0.mid\n",
      "0.07205057144165039\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_f0NbGeyJYms_0.mid\n",
      "0.02169060707092285\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Dw7M69CidJ8_1.mid\n",
      "0.019387245178222656\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_iwxwLEvqeD8_1.mid\n",
      "0.013441085815429688\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_60LLKmpgzRM_0.mid\n",
      "0.005851268768310547\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_9.mid\n",
      "0.017677783966064453\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_2.mid\n",
      "0.052600860595703125\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XC_SiJszQx0_2.mid\n",
      "0.011746644973754883\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_dtS02mrDMsM_3.mid\n",
      "0.04184222221374512\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_v2an-9szi8M_0.mid\n",
      "0.014657020568847656\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wqc8iqbDsGM_3.mid\n",
      "0.012456655502319336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0qYNvMCEwaw_0.mid\n",
      "0.031180858612060547\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_8.mid\n",
      "0.031032562255859375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_K84TcgjCRt4_0.mid\n",
      "0.011384248733520508\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_16OaOpfOPZ0_3.mid\n",
      "0.021365642547607422\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UE0y8MHqT-g_2.mid\n",
      "0.03603982925415039\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Q49PWE0RJLY_1.mid\n",
      "0.017724275588989258\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_OO7zYVeK814_2.mid\n",
      "0.01020956039428711\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_fey-8bOR95E_1.mid\n",
      "0.03459882736206055\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_13.mid\n",
      "0.02399730682373047\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_3.mid\n",
      "0.04633307456970215\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_BGf5nCdzPOc_1.mid\n",
      "0.03803443908691406\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ZdyiXIlKe_Y_0.mid\n",
      "0.024945497512817383\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_snfsc7pEzlk_2.mid\n",
      "0.03700590133666992\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_7yW9c7t8Hq0_1.mid\n",
      "0.036496877670288086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_xrhWli_R98g_2.mid\n",
      "0.015259027481079102\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_k-FNDbK6Qhg_0.mid\n",
      "0.04179644584655762\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_g15aRCHUJEI_2.mid\n",
      "0.013239145278930664\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FyK_c-TIcCA_2.mid\n",
      "0.09755659103393555\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_B3aRl8iTEKw_1.mid\n",
      "0.026515483856201172\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ufRISSD28XA_1.mid\n",
      "0.022248268127441406\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iHPKusssXzk_1.mid\n",
      "0.017317771911621094\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ldCQ6N9G6Mk_1.mid\n",
      "0.014097929000854492\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ZTrEoB8T9YA_2.mid\n",
      "0.026227235794067383\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wfXSdMsd4q8_0.mid\n",
      "0.014818906784057617\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_4.mid\n",
      "0.0204160213470459\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_uqRLEByE6pU_2.mid\n",
      "0.050002336502075195\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_nFunbdBrukY_0.mid\n",
      "0.014004945755004883\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_nBao46YGZDg_0.mid\n",
      "0.011089563369750977\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_K84TcgjCRt4_2.mid\n",
      "0.023228168487548828\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Xf2fox7zhCQ_2.mid\n",
      "0.023380279541015625\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_nGkpcWNAxDU_2.mid\n",
      "0.02346944808959961\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_PZN7XLKzLDI_1.mid\n",
      "0.02319169044494629\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_k4lT-sXg3iI_2.mid\n",
      "0.027243375778198242\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cm6E860vDjY_2.mid\n",
      "0.03215909004211426\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ivCNV47tsRw_1.mid\n",
      "0.15006637573242188\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_IH2KrGjKXw0_3.mid\n",
      "0.04444384574890137\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1vjy9oMFa8c_5.mid\n",
      "0.03886294364929199\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kMHiU0fhWVw_3.mid\n",
      "0.01733112335205078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_GwdQliEv6Z0_0.mid\n",
      "0.0293731689453125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_19.mid\n",
      "0.013559103012084961\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__kJtgm1OUNA_1.mid\n",
      "0.03165030479431152\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_GwdQliEv6Z0_1.mid\n",
      "0.02244710922241211\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_2eIsQtm4YNs_1.mid\n",
      "0.019784212112426758\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JP3QKZlyQz4_2.mid\n",
      "0.009644031524658203\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_qRceKZtwRWA_1.mid\n",
      "0.018060684204101562\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D-gDPh9ZpFk_1.mid\n",
      "0.03737473487854004\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Ie5koh4qvJc_6.mid\n",
      "0.03593087196350098\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_RaUSISlnhKw_0.mid\n",
      "0.020167112350463867\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_3_MxhSS86oU_0.mid\n",
      "0.04223203659057617\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_0.mid\n",
      "0.02583479881286621\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_o5AIp2Yc01M_1.mid\n",
      "0.046500444412231445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_HYVmgq5Y93g_2.mid\n",
      "0.016113758087158203\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FUAK5TBaNY8_0.mid\n",
      "0.02444171905517578\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vBHoSz8cJIE_2.mid\n",
      "0.007874011993408203\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Xf2fox7zhCQ_1.mid\n",
      "0.014252424240112305\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_epMOLn5Axw0_0.mid\n",
      "0.022408008575439453\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_7gifuoofgpk_0.mid\n",
      "0.01407003402709961\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_nBIls0laAAU_1.mid\n",
      "0.01995992660522461\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_c6CwY8Gbw0c_4.mid\n",
      "0.03209066390991211\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OgkB7PfqWzA_0.mid\n",
      "0.009536981582641602\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_p2qihp_PnLY_0.mid\n",
      "0.0273740291595459\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_2v6gJi03LlA_0.mid\n",
      "0.018553972244262695\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_dCoUuknChhE_0.mid\n",
      "0.010849237442016602\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UOSlDydo94E_1.mid\n",
      "0.016957521438598633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1Qc15G0ZHIg_0.mid\n",
      "0.013405561447143555\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_7JIdJLkJ0S4_2.mid\n",
      "0.014976024627685547\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_x5ibvz38jOs_1.mid\n",
      "0.028074026107788086\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__BK2o77sTc0_1.mid\n",
      "0.021663904190063477\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_z5ewpQwGDNw_2.mid\n",
      "0.023589372634887695\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_5CEAeMiXKaA_0.mid\n",
      "0.017870664596557617\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_sEQf5lcnj_o_2.mid\n",
      "0.05610847473144531\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_VrWAZLRdQO8_1.mid\n",
      "0.008920907974243164\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_11.mid\n",
      "0.04093337059020996\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_j8Ir-ssM-AA_0.mid\n",
      "0.03019547462463379\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Oku8sIpTMAE_1.mid\n",
      "0.0053861141204833984\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_jziI_cuN_H8_0.mid\n",
      "0.025127649307250977\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_doM_LWo-74A_3.mid\n",
      "0.016073942184448242\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_10.mid\n",
      "0.012923002243041992\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ivCNV47tsRw_2.mid\n",
      "0.05538630485534668\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_PXOWy7NiZhk_0.mid\n",
      "0.0209810733795166\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wS0RzrBL9UI_2.mid\n",
      "0.02478504180908203\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_2Z9SjI131jA_5.mid\n",
      "0.012323379516601562\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_vv6nrZ2myXw_0.mid\n",
      "0.019795894622802734\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_DYBBfpEx-JI_1.mid\n",
      "0.0219724178314209\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GbUV3TXUzeQ_2.mid\n",
      "0.03568673133850098\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_JMt4m608s3k_2.mid\n",
      "0.038985490798950195\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_8rupdevqfuI_1.mid\n",
      "0.025334835052490234\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ii1Jw1-7Ka4_2.mid\n",
      "0.02123737335205078\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_QJlnTN7HRwE_1.mid\n",
      "0.01191258430480957\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1vjy9oMFa8c_3.mid\n",
      "0.010513782501220703\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_K84TcgjCRt4_1.mid\n",
      "0.016376733779907227\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_tLmXAWfbGqs_0.mid\n",
      "0.01574873924255371\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_12tDlCQVRtA_2.mid\n",
      "0.021143436431884766\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_CkjAksZstEA_1.mid\n",
      "0.013508319854736328\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_CkjAksZstEA_0.mid\n",
      "0.010314702987670898\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_5.mid\n",
      "0.0437467098236084\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_22.mid\n",
      "0.017607450485229492\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jrVXitpLGMk_0.mid\n",
      "0.021572589874267578\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_NTI9ode6bRk_1.mid\n",
      "0.018291950225830078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jrVXitpLGMk_1.mid\n",
      "0.034613609313964844\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1A9Hh2GrDTs_0.mid\n",
      "0.022028446197509766\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0c2dTJTDUR4_0.mid\n",
      "0.008440971374511719\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5J_zI7QE9W8_0.mid\n",
      "0.02311086654663086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kMHiU0fhWVw_0.mid\n",
      "0.01592421531677246\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_78C_TpQDvVE_0.mid\n",
      "0.011719465255737305\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_x4B0raxVODA_3.mid\n",
      "0.009850025177001953\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wfXSdMsd4q8_1.mid\n",
      "0.014912605285644531\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_k4dxAlI5N-k_0.mid\n",
      "0.04542851448059082\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ADJ_PDauy-g_1.mid\n",
      "0.022236108779907227\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_fuCxYrru2S4_0.mid\n",
      "0.02190566062927246\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_VCSDjQxAJzc_2.mid\n",
      "0.013746261596679688\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vBHoSz8cJIE_4.mid\n",
      "0.007816076278686523\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_VXgRfsVd_o0_0.mid\n",
      "0.014035224914550781\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_sEQf5lcnj_o_0.mid\n",
      "0.0391690731048584\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cm6E860vDjY_0.mid\n",
      "0.03410601615905762\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_UuLTvhERcnk_0.mid\n",
      "0.013617753982543945\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_k4dxAlI5N-k_1.mid\n",
      "0.06734490394592285\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_egYSmNuIFGk_0.mid\n",
      "0.0213472843170166\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_QmK_SND2jAw_1.mid\n",
      "0.039518117904663086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_I2MrA-o5H8I_2.mid\n",
      "0.03586435317993164\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Adil-zpJdEs_0.mid\n",
      "0.033754587173461914\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltNK_MY1HkM_1.mid\n",
      "0.006692171096801758\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_170KC2CbJ7Y_1.mid\n",
      "0.013330221176147461\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_NGE9ynTJABg_2.mid\n",
      "0.023529529571533203\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_g15aRCHUJEI_1.mid\n",
      "0.012892007827758789\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vBHoSz8cJIE_0.mid\n",
      "0.009831428527832031\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_GAvPwG2wYZM_1.mid\n",
      "0.015372753143310547\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jCLgDtk695I_3.mid\n",
      "0.013053178787231445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jCLgDtk695I_1.mid\n",
      "0.013541936874389648\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_8c7kcmsnf-s_0.mid\n",
      "0.03197741508483887\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_gwmVjvR-sVs_0.mid\n",
      "0.028429508209228516\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_dfNdpy8TUzA_0.mid\n",
      "0.013284683227539062\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OyaY4213c2A_0.mid\n",
      "0.02411651611328125\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Gcbk0GPT3-g_1.mid\n",
      "0.015913009643554688\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_77z6Ep3aOmg_1.mid\n",
      "0.027301549911499023\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zHwJG8Rrx6c_3.mid\n",
      "0.02138805389404297\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_z5ewpQwGDNw_1.mid\n",
      "0.017404794692993164\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_biROWEwkDQQ_2.mid\n",
      "0.013927221298217773\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_uqRLEByE6pU_0.mid\n",
      "0.040572166442871094\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_gvWDOIiocuE_2.mid\n",
      "0.012457132339477539\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jTPXwbDtIpA_1.mid\n",
      "0.1717517375946045\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_iFgSjUqI7iM_0.mid\n",
      "0.02472233772277832\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_NP0lwB-_-og_1.mid\n",
      "0.03656578063964844\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_5NW0zDu6IYM_1.mid\n",
      "0.027909278869628906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_16OaOpfOPZ0_0.mid\n",
      "0.013463735580444336\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_B3aRl8iTEKw_0.mid\n",
      "0.024259567260742188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0J7lX-Tcx-w_0.mid\n",
      "0.029517412185668945\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_veG92Oi-DlU_2.mid\n",
      "0.015755653381347656\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_nOIBJHkqrE8_1.mid\n",
      "0.024122238159179688\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wxXpvnmq_-E_0.mid\n",
      "0.016007184982299805\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_bfopzItCYrE_1.mid\n",
      "0.020810842514038086\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ivCNV47tsRw_3.mid\n",
      "0.033740997314453125\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_JMt4m608s3k_1.mid\n",
      "0.03220558166503906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wlOa1jkTn_U_2.mid\n",
      "0.013080358505249023\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_MT2_mXn-vGU_1.mid\n",
      "0.01904129981994629\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_0c2dTJTDUR4_1.mid\n",
      "0.012900829315185547\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_RiQMuhk_SuQ_3.mid\n",
      "0.05269265174865723\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_VCSDjQxAJzc_0.mid\n",
      "0.018595218658447266\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wqc8iqbDsGM_1.mid\n",
      "0.013319015502929688\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_PdwxeMPXOCk_0.mid\n",
      "0.03420758247375488\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_sA9ac1rod84_1.mid\n",
      "0.025386333465576172\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_dtS02mrDMsM_4.mid\n",
      "0.03098750114440918\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_e838BE_MH6s_0.mid\n",
      "0.023977279663085938\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_j8Ir-ssM-AA_1.mid\n",
      "0.016817092895507812\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_RiQMuhk_SuQ_1.mid\n",
      "0.030833959579467773\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GbUV3TXUzeQ_1.mid\n",
      "0.02758169174194336\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5DqIntcmdRI_0.mid\n",
      "0.030392169952392578\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_IH2KrGjKXw0_2.mid\n",
      "0.013901472091674805\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_mlHheZbblT0_0.mid\n",
      "0.02841949462890625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_61EA0xRX8gE_5.mid\n",
      "0.02682638168334961\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_4Er-eFHSHA4_0.mid\n",
      "0.014074087142944336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_8TYY0qG-KOw_1.mid\n",
      "0.024638891220092773\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_3ZnxqCZ7qGg_1.mid\n",
      "0.01547384262084961\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_heBmBQDWj-M_1.mid\n",
      "0.019881010055541992\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yhIncZ_Ue4E_2.mid\n",
      "0.0188596248626709\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ItGNJM6skM4_1.mid\n",
      "0.0511014461517334\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_16OaOpfOPZ0_2.mid\n",
      "0.028797626495361328\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_V_jbWvQNh5c_0.mid\n",
      "0.014766454696655273\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zjr_-ql5Avk_0.mid\n",
      "0.009330511093139648\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_OCqMDeD6Fmc_3.mid\n",
      "0.04949688911437988\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_02t1bj7ZABY_2.mid\n",
      "0.019103288650512695\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_MbvPxCUMSek_2.mid\n",
      "0.008748292922973633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UpHutdJvZMI_1.mid\n",
      "0.021831274032592773\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ilCcZpD39HU_4.mid\n",
      "0.020946979522705078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D5kyjnlDNZs_2.mid\n",
      "0.06462430953979492\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rDhbGjFlKF8_0.mid\n",
      "0.024235010147094727\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_im4Qxn3GQvo_2.mid\n",
      "0.03301548957824707\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1fdxsFbnsX4_1.mid\n",
      "0.016291141510009766\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_MbvPxCUMSek_0.mid\n",
      "0.012089967727661133\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_7.mid\n",
      "0.03429818153381348\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_g-yM0Lsp4lc_1.mid\n",
      "0.008042097091674805\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ufRISSD28XA_0.mid\n",
      "0.019867420196533203\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltxNPJda7zE_3.mid\n",
      "0.016210079193115234\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_3.mid\n",
      "0.01445627212524414\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_BaHHHgDd0BU_1.mid\n",
      "0.025834083557128906\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_J6X3rVU1H-c_0.mid\n",
      "0.010395526885986328\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_HY9vPoHbgaI_3.mid\n",
      "0.01295328140258789\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e838BE_MH6s_1.mid\n",
      "0.018395662307739258\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_biROWEwkDQQ_3.mid\n",
      "0.022644996643066406\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3TYeU9idRGI_1.mid\n",
      "0.00786590576171875\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Wy2-my19YnY_1.mid\n",
      "0.02635502815246582\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UE0y8MHqT-g_1.mid\n",
      "0.04336953163146973\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_1Q3MoBFh6eU_0.mid\n",
      "0.018907785415649414\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_QJlnTN7HRwE_3.mid\n",
      "0.013145685195922852\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_r6laJv1-HMg_0.mid\n",
      "0.025534391403198242\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3As1t6JySDE_0.mid\n",
      "0.021122217178344727\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_nBIls0laAAU_0.mid\n",
      "0.019963979721069336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_f2BXXa7lPuE_2.mid\n",
      "0.042772531509399414\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_30.mid\n",
      "0.024466991424560547\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_By3JCii1k3w_1.mid\n",
      "0.014814138412475586\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_NGE9ynTJABg_0.mid\n",
      "0.040805816650390625\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_OCqMDeD6Fmc_1.mid\n",
      "0.03333926200866699\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JxSU49jFKwM_0.mid\n",
      "0.016267776489257812\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wxXpvnmq_-E_1.mid\n",
      "0.014075756072998047\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Oku8sIpTMAE_4.mid\n",
      "0.011995077133178711\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltNK_MY1HkM_0.mid\n",
      "0.009334087371826172\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FH28sPhm9Dc_0.mid\n",
      "0.01884174346923828\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_rEz0D3VohFA_1.mid\n",
      "0.037218332290649414\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yrvKReuwVlY_1.mid\n",
      "0.013031482696533203\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_QwsQ8ejbMKg_1.mid\n",
      "0.021305561065673828\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_S1T3UF1vhSk_0.mid\n",
      "0.036252498626708984\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_HQ8ISDX6PiI_0.mid\n",
      "0.021952152252197266\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_JT1XJnVmABo_5.mid\n",
      "0.018685102462768555\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_9eMxCs-LXCE_1.mid\n",
      "0.01169729232788086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_YhXRyOl5pi0_0.mid\n",
      "0.02560877799987793\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_kfYaGryaGzI_1.mid\n",
      "0.006352424621582031\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GY3f6ckBVkA_3.mid\n",
      "0.02092742919921875\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iXiSFSGws-c_2.mid\n",
      "0.016344070434570312\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_WQrPcuy50aA_0.mid\n",
      "0.029157400131225586\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Dg935IbDggo_2.mid\n",
      "0.029478073120117188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Dg935IbDggo_3.mid\n",
      "0.04624009132385254\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_wnB2cjd6zbE_0.mid\n",
      "0.03491377830505371\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_8c7kcmsnf-s_2.mid\n",
      "0.03459954261779785\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bIx4DwkZxFY_1.mid\n",
      "0.02359628677368164\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_08t73Qgjkt4_0.mid\n",
      "0.021080493927001953\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cm6E860vDjY_1.mid\n",
      "0.04089546203613281\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_5NW0zDu6IYM_0.mid\n",
      "0.025212764739990234\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_JYVXM0qNQAg_2.mid\n",
      "0.021463394165039062\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_v3Nl5rUxLqE_0.mid\n",
      "0.011512994766235352\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ijfxkQPAQdE_1.mid\n",
      "0.07168030738830566\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UaYYMRkHMYc_2.mid\n",
      "0.021605491638183594\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1vjy9oMFa8c_0.mid\n",
      "0.020531654357910156\n",
      "data/emopia/EMOPIA_2.2/midis/Q2__FVzelKlBFs_1.mid\n",
      "0.16099190711975098\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_MT2_mXn-vGU_2.mid\n",
      "0.01604151725769043\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_vVTP0DOL_2Q_2.mid\n",
      "0.017586231231689453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6Uf9XBUD3wE_1.mid\n",
      "0.04883527755737305\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_4ydjOX3pWds_1.mid\n",
      "0.01978468894958496\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_28.mid\n",
      "0.020974397659301758\n",
      "data/emopia/EMOPIA_2.2/midis/Q3__S991V8N-8s_0.mid\n",
      "0.025865554809570312\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iXiSFSGws-c_0.mid\n",
      "0.017798662185668945\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zmZHNy9T8Pg_1.mid\n",
      "0.03901505470275879\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_3ahg_eQZhxs_0.mid\n",
      "0.02955770492553711\n",
      "data/emopia/EMOPIA_2.2/midis/Q2__FVzelKlBFs_0.mid\n",
      "0.0406947135925293\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_27.mid\n",
      "0.025330543518066406\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_biROWEwkDQQ_0.mid\n",
      "0.017482757568359375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_8.mid\n",
      "0.030812501907348633\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_sZe3IB1OVfw_1.mid\n",
      "0.020721912384033203\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_PXOWy7NiZhk_1.mid\n",
      "0.013898849487304688\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_REq37pDAm3A_2.mid\n",
      "0.013831615447998047\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tjno89OVuRI_0.mid\n",
      "0.014877796173095703\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_8c7kcmsnf-s_1.mid\n",
      "0.041019439697265625\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_2.mid\n",
      "0.02588176727294922\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_veG92Oi-DlU_1.mid\n",
      "0.017911434173583984\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_3N2G21U7guk_6.mid\n",
      "0.029342174530029297\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_2v6gJi03LlA_1.mid\n",
      "0.02499532699584961\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_bfopzItCYrE_0.mid\n",
      "0.01086878776550293\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_6oLmO9ZSJt0_0.mid\n",
      "0.03411865234375\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_x4B0raxVODA_0.mid\n",
      "0.0073359012603759766\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_k4lT-sXg3iI_0.mid\n",
      "0.021229028701782227\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_5Ju9q1N2x0E_2.mid\n",
      "0.044994354248046875\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_dtS02mrDMsM_0.mid\n",
      "0.013997554779052734\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_fuCxYrru2S4_2.mid\n",
      "0.03319120407104492\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_14.mid\n",
      "0.0150146484375\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_d_49EtXDMFE_0.mid\n",
      "0.014282464981079102\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jTPXwbDtIpA_2.mid\n",
      "0.042771339416503906\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vpTguZtJAFA_0.mid\n",
      "0.014655351638793945\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_08t73Qgjkt4_1.mid\n",
      "0.01630091667175293\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_22S3w4idugs_0.mid\n",
      "0.011931180953979492\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_TOXg4-ZGb-g_1.mid\n",
      "0.011231184005737305\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_13.mid\n",
      "0.05413413047790527\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_7JIdJLkJ0S4_0.mid\n",
      "0.02124309539794922\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_nGkpcWNAxDU_0.mid\n",
      "0.021251201629638672\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_-_FjI4bT2Wo_1.mid\n",
      "0.01985788345336914\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_kDGmND1BgmA_0.mid\n",
      "0.00722050666809082\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_1.mid\n",
      "0.026130199432373047\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_9v2WSpn4FCw_9.mid\n",
      "0.017005443572998047\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_G4rL_OtfFAU_0.mid\n",
      "0.03141379356384277\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ANZf1QXsNrY_6.mid\n",
      "0.02235865592956543\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rW1I58_lMEg_0.mid\n",
      "0.018242359161376953\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_48EYLEAgaBc_1.mid\n",
      "0.005810737609863281\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Dw7M69CidJ8_2.mid\n",
      "0.03509020805358887\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UbKAl4roB8k_0.mid\n",
      "0.02492666244506836\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ilCcZpD39HU_2.mid\n",
      "0.015205621719360352\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6kRPHamGDSo_1.mid\n",
      "0.01732349395751953\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_7.mid\n",
      "0.03696584701538086\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_f2BXXa7lPuE_1.mid\n",
      "0.03915667533874512\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_PK8YIUaV3Xw_0.mid\n",
      "0.017430543899536133\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_prUUays30Mg_0.mid\n",
      "0.013817548751831055\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_0vLPYiPN7qY_2.mid\n",
      "0.031404972076416016\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FUAK5TBaNY8_3.mid\n",
      "0.015525102615356445\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zHwJG8Rrx6c_0.mid\n",
      "0.014293432235717773\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KxlqB3j0zys_3.mid\n",
      "0.025432825088500977\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0lEL_XHyhuU_1.mid\n",
      "0.012208938598632812\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_NGE9ynTJABg_1.mid\n",
      "0.02001786231994629\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pkz8OkWhwnk_0.mid\n",
      "0.013889551162719727\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_TOXg4-ZGb-g_0.mid\n",
      "0.0166318416595459\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_1.mid\n",
      "0.013515472412109375\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zmZHNy9T8Pg_2.mid\n",
      "0.021727800369262695\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_11.mid\n",
      "0.04880261421203613\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_2.mid\n",
      "0.019118547439575195\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_FfwKrQyQ7WU_0.mid\n",
      "0.02542591094970703\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_02t1bj7ZABY_0.mid\n",
      "0.007191658020019531\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_REq37pDAm3A_4.mid\n",
      "0.022761821746826172\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wqc8iqbDsGM_0.mid\n",
      "0.016284942626953125\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ADJ_PDauy-g_2.mid\n",
      "0.016378402709960938\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_IH2KrGjKXw0_0.mid\n",
      "0.013204813003540039\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ZTrEoB8T9YA_1.mid\n",
      "0.017310142517089844\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_7yW9c7t8Hq0_0.mid\n",
      "0.010921001434326172\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_YAAxPW1GB7w_0.mid\n",
      "0.01160740852355957\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jCLgDtk695I_2.mid\n",
      "0.014191389083862305\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D-gDPh9ZpFk_0.mid\n",
      "0.027149200439453125\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_OyaY4213c2A_1.mid\n",
      "0.017129898071289062\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_QmK_SND2jAw_0.mid\n",
      "0.025053739547729492\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wfXSdMsd4q8_3.mid\n",
      "0.02760910987854004\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_XC_SiJszQx0_3.mid\n",
      "0.03258681297302246\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_NTI9ode6bRk_0.mid\n",
      "0.016153573989868164\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_8rupdevqfuI_0.mid\n",
      "0.025146961212158203\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_8.mid\n",
      "0.02028489112854004\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vJ77eDxHezE_0.mid\n",
      "0.02838754653930664\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_KxlqB3j0zys_0.mid\n",
      "0.01542043685913086\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_rEz0D3VohFA_0.mid\n",
      "0.007854938507080078\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_BzqX-9TA-GY_1.mid\n",
      "0.01163935661315918\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_fnqb8zqHqdY_2.mid\n",
      "0.016338586807250977\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_aYe-2Glruu4_1.mid\n",
      "0.03727364540100098\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_PdwxeMPXOCk_2.mid\n",
      "0.03294801712036133\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_DX1IK3z1w10_1.mid\n",
      "0.0290524959564209\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_By3JCii1k3w_0.mid\n",
      "0.02009892463684082\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_H6_7DjsbROQ_2.mid\n",
      "0.01607656478881836\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rW1I58_lMEg_3.mid\n",
      "0.024875879287719727\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_UAm0aWvzFI8_0.mid\n",
      "0.0452580451965332\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wfXSdMsd4q8_4.mid\n",
      "0.02157115936279297\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_sZe3IB1OVfw_0.mid\n",
      "0.021422147750854492\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_XfA2KXodrOE_1.mid\n",
      "0.0357363224029541\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_15.mid\n",
      "0.02473902702331543\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_61EA0xRX8gE_0.mid\n",
      "0.04087567329406738\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_qlbazHayULg_8.mid\n",
      "0.042560577392578125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_7JIdJLkJ0S4_1.mid\n",
      "0.02955770492553711\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FoTXpYZXxJs_1.mid\n",
      "0.011708498001098633\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_--2B4d4lQhs_0.mid\n",
      "0.042557477951049805\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Y5JcZQ0xg4Y_2.mid\n",
      "0.027269601821899414\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_rDhbGjFlKF8_1.mid\n",
      "0.013274192810058594\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_ANZf1QXsNrY_4.mid\n",
      "0.02931666374206543\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ldCQ6N9G6Mk_0.mid\n",
      "0.005745410919189453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_TNoaOTrTkTQ_0.mid\n",
      "0.037145376205444336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_38i12IPkL5c_0.mid\n",
      "0.025066614151000977\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_X1p8xcarVu0_1.mid\n",
      "0.016347408294677734\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UpHutdJvZMI_2.mid\n",
      "0.0203249454498291\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_V3Y9L4UOcpk_0.mid\n",
      "0.016982555389404297\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_20.mid\n",
      "0.0212857723236084\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_osxmQReE_2o_0.mid\n",
      "0.028096437454223633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4__BK2o77sTc0_2.mid\n",
      "0.02636861801147461\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_4dXC1cC7crw_0.mid\n",
      "0.0236968994140625\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_QJlnTN7HRwE_2.mid\n",
      "0.010537862777709961\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_pRuwHN-lI44_1.mid\n",
      "0.01676797866821289\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_q4T8Znozvkk_0.mid\n",
      "0.012881994247436523\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UaYYMRkHMYc_1.mid\n",
      "0.04482388496398926\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_4.mid\n",
      "0.012448310852050781\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_-jJVb0xvbdg_1.mid\n",
      "0.03152871131896973\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_23.mid\n",
      "0.021757841110229492\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_94ROf10d1iA_3.mid\n",
      "0.026740312576293945\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_uj3Gif77SYM_6.mid\n",
      "0.026311159133911133\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FSzOgraSKhs_0.mid\n",
      "0.011494874954223633\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_lTxVVdhFE6Q_1.mid\n",
      "0.01935720443725586\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wqc8iqbDsGM_4.mid\n",
      "0.01812458038330078\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_170KC2CbJ7Y_0.mid\n",
      "0.0210726261138916\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zHwJG8Rrx6c_2.mid\n",
      "0.029301166534423828\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_doM_LWo-74A_4.mid\n",
      "0.040665388107299805\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_yC0fAQxGgr0_1.mid\n",
      "0.028226852416992188\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zMa78n8ULtE_0.mid\n",
      "0.01072549819946289\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_UaYYMRkHMYc_0.mid\n",
      "0.15878582000732422\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_FfwKrQyQ7WU_2.mid\n",
      "0.01906442642211914\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_cQp1BYDGcRo_0.mid\n",
      "0.037127017974853516\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_xgtwQGeB6_0_0.mid\n",
      "0.01956653594970703\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_hoPp_GLXQis_0.mid\n",
      "0.02570509910583496\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_RaUSISlnhKw_1.mid\n",
      "0.03957629203796387\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ury1cdB79s0_1.mid\n",
      "0.025757312774658203\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zvWX_dwQPtU_2.mid\n",
      "0.025813579559326172\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_IH2KrGjKXw0_1.mid\n",
      "0.02019023895263672\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iHPKusssXzk_0.mid\n",
      "0.021902799606323242\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_kKFwiXKlhC8_0.mid\n",
      "0.0233919620513916\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_k4lT-sXg3iI_1.mid\n",
      "0.02287745475769043\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltxNPJda7zE_0.mid\n",
      "0.008745193481445312\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_XWk5XUEBXmg_1.mid\n",
      "0.018500804901123047\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_yC0fAQxGgr0_0.mid\n",
      "0.015583038330078125\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Tjno89OVuRI_1.mid\n",
      "0.01950383186340332\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ldCQ6N9G6Mk_3.mid\n",
      "0.017913818359375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_3ZnxqCZ7qGg_2.mid\n",
      "0.022760629653930664\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_biROWEwkDQQ_1.mid\n",
      "0.022472620010375977\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_P-We4q3goKU_6.mid\n",
      "0.03341841697692871\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_cXxGBDgFCs8_1.mid\n",
      "0.014159440994262695\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_E5qEloUO3SM_1.mid\n",
      "0.01753687858581543\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_uqRLEByE6pU_1.mid\n",
      "0.01580047607421875\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_aYe-2Glruu4_3.mid\n",
      "0.02268838882446289\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_48EYLEAgaBc_0.mid\n",
      "0.008149385452270508\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_yFw_kO7DF-Y_2.mid\n",
      "0.044164180755615234\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_k6aLLVX1D7Y_2.mid\n",
      "0.019231557846069336\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_bfopzItCYrE_2.mid\n",
      "0.01754164695739746\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_29.mid\n",
      "0.007358074188232422\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ItGNJM6skM4_0.mid\n",
      "0.04209494590759277\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JP3QKZlyQz4_1.mid\n",
      "0.011746883392333984\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_GbUV3TXUzeQ_0.mid\n",
      "0.020267963409423828\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_T92R7xjce34_0.mid\n",
      "0.018584251403808594\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_D5kyjnlDNZs_1.mid\n",
      "0.03484058380126953\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jIKX3ShvLxo_2.mid\n",
      "0.044945716857910156\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Ie5koh4qvJc_26.mid\n",
      "0.020922422409057617\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_REq37pDAm3A_1.mid\n",
      "0.017453908920288086\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_iRu9KQvbtVw_0.mid\n",
      "0.020391464233398438\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_l-4Kce3YxxQ_2.mid\n",
      "0.047788381576538086\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Y1F0c3zgMJo_0.mid\n",
      "0.016190052032470703\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_BIRDDxz0E4c_0.mid\n",
      "0.01383829116821289\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_pouxqiySdI8_0.mid\n",
      "0.02433609962463379\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_3.mid\n",
      "0.04065728187561035\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_doM_LWo-74A_0.mid\n",
      "0.021623849868774414\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FOcdHIhI0_s_0.mid\n",
      "0.009388446807861328\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ijfxkQPAQdE_0.mid\n",
      "0.055986881256103516\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_uUtnyA6tLnI_0.mid\n",
      "0.02264094352722168\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_aQ4PIaRMLwE_2.mid\n",
      "0.02975010871887207\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_SQDuF0qxGQw_0.mid\n",
      "0.023522615432739258\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_k6aLLVX1D7Y_0.mid\n",
      "0.025661230087280273\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_8rupdevqfuI_2.mid\n",
      "0.017850160598754883\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_94ROf10d1iA_1.mid\n",
      "0.015451431274414062\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yZGBxiNPu3k_0.mid\n",
      "0.03209424018859863\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_pbVGv_g4n50_0.mid\n",
      "0.007588386535644531\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_6kRPHamGDSo_0.mid\n",
      "0.017235755920410156\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_1Qc15G0ZHIg_2.mid\n",
      "0.030794858932495117\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_JNzqL24ZcZ8_1.mid\n",
      "0.019245386123657227\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_a5QCcwEjxAk_1.mid\n",
      "0.012137651443481445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_heBmBQDWj-M_0.mid\n",
      "0.020823001861572266\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_UE0y8MHqT-g_0.mid\n",
      "0.016431331634521484\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_JYVXM0qNQAg_0.mid\n",
      "0.021768808364868164\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_DYBBfpEx-JI_4.mid\n",
      "0.0180051326751709\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bIx4DwkZxFY_2.mid\n",
      "0.017147541046142578\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_S7Bo4-NCEDk_0.mid\n",
      "0.04277944564819336\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Uu-oRTRT6uQ_2.mid\n",
      "0.02417159080505371\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_--2B4d4lQhs_3.mid\n",
      "0.016885757446289062\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_8izVTDgBQPc_1.mid\n",
      "0.024000167846679688\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_o5AIp2Yc01M_0.mid\n",
      "0.031935691833496094\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_9Xg69XrVaYU_0.mid\n",
      "0.021656036376953125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_OO7zYVeK814_0.mid\n",
      "0.01372981071472168\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_nGkpcWNAxDU_1.mid\n",
      "0.027597904205322266\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_NzJjMGJg1wE_0.mid\n",
      "0.017026185989379883\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jO2zX6Ul8GM_2.mid\n",
      "0.01932525634765625\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_TonQX8XbvX8_3.mid\n",
      "0.01522374153137207\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_vv6nrZ2myXw_1.mid\n",
      "0.02679300308227539\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_bbU31JLtlug_0.mid\n",
      "0.02821826934814453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e838BE_MH6s_2.mid\n",
      "0.029928922653198242\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Wy2-my19YnY_0.mid\n",
      "0.02583789825439453\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_0vLPYiPN7qY_0.mid\n",
      "0.015313148498535156\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_HYVmgq5Y93g_1.mid\n",
      "0.018862485885620117\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_yhIncZ_Ue4E_1.mid\n",
      "0.022689342498779297\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ilCcZpD39HU_3.mid\n",
      "0.018048524856567383\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_18.mid\n",
      "0.018447399139404297\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_5Ju9q1N2x0E_0.mid\n",
      "0.02557086944580078\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_G4rL_OtfFAU_1.mid\n",
      "0.060091495513916016\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_6kRPHamGDSo_2.mid\n",
      "0.01685309410095215\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zjr_-ql5Avk_2.mid\n",
      "0.006957054138183594\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_2uLw9Zs60A4_0.mid\n",
      "0.025287866592407227\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_14.mid\n",
      "0.01616811752319336\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_yFw_kO7DF-Y_0.mid\n",
      "0.03182172775268555\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_egYSmNuIFGk_1.mid\n",
      "0.019513845443725586\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vJ77eDxHezE_1.mid\n",
      "0.026709556579589844\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_5.mid\n",
      "0.0310819149017334\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_4.mid\n",
      "0.034014225006103516\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_RL_cmmNVLfs_0.mid\n",
      "0.014422416687011719\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1Q3MoBFh6eU_1.mid\n",
      "0.04228997230529785\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Iq6g_4AwUWs_1.mid\n",
      "0.02358865737915039\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_JT1XJnVmABo_4.mid\n",
      "0.0175936222076416\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ANZf1QXsNrY_7.mid\n",
      "0.03852033615112305\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_xIsvaT20pZ0_0.mid\n",
      "0.017764568328857422\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_gvWDOIiocuE_1.mid\n",
      "0.028793811798095703\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_9v2WSpn4FCw_1.mid\n",
      "0.013632774353027344\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_zmZHNy9T8Pg_0.mid\n",
      "0.02174067497253418\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JT1XJnVmABo_0.mid\n",
      "0.017595291137695312\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_iFgSjUqI7iM_1.mid\n",
      "0.021000146865844727\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_zMa78n8ULtE_1.mid\n",
      "0.009769439697265625\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_c6CwY8Gbw0c_2.mid\n",
      "0.01769566535949707\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jSWItJI-Gmk_0.mid\n",
      "0.01628565788269043\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__8v0MFBZoco_1.mid\n",
      "0.037782907485961914\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_6n8HgIcJ6vo_0.mid\n",
      "0.013116121292114258\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_61EA0xRX8gE_3.mid\n",
      "0.04079699516296387\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_1vjy9oMFa8c_1.mid\n",
      "0.009852409362792969\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_H6_7DjsbROQ_4.mid\n",
      "0.017538070678710938\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_3N2G21U7guk_3.mid\n",
      "0.03393220901489258\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_wS0RzrBL9UI_1.mid\n",
      "0.013781547546386719\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_0vLPYiPN7qY_3.mid\n",
      "0.035610198974609375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_S1T3UF1vhSk_2.mid\n",
      "0.021044492721557617\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_JT1XJnVmABo_1.mid\n",
      "0.019911527633666992\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_vBHoSz8cJIE_1.mid\n",
      "0.00996255874633789\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_s6-SbDSZzEU_1.mid\n",
      "0.02078104019165039\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_ItGNJM6skM4_2.mid\n",
      "0.16391730308532715\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_lFznCt5kvXM_0.mid\n",
      "0.04940223693847656\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_aYe-2Glruu4_0.mid\n",
      "0.027859926223754883\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_fey-8bOR95E_3.mid\n",
      "0.03406548500061035\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_1kny88W533Q_0.mid\n",
      "0.011762857437133789\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_a5QCcwEjxAk_0.mid\n",
      "0.014340639114379883\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_FUAK5TBaNY8_2.mid\n",
      "0.018521785736083984\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_e838BE_MH6s_4.mid\n",
      "0.03400778770446777\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_u_7b9CwyM8Q_1.mid\n",
      "0.013019084930419922\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_REq37pDAm3A_0.mid\n",
      "0.011442184448242188\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_c6CwY8Gbw0c_0.mid\n",
      "0.019066810607910156\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_WpZ1rWHOCAQ_1.mid\n",
      "0.015500307083129883\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Oku8sIpTMAE_3.mid\n",
      "0.016019821166992188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_S1T3UF1vhSk_1.mid\n",
      "0.013047933578491211\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_c6CwY8Gbw0c_3.mid\n",
      "0.02057814598083496\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_9v2WSpn4FCw_4.mid\n",
      "0.07351803779602051\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_-vAz_HTFEXs_1.mid\n",
      "0.018556594848632812\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_fey-8bOR95E_2.mid\n",
      "0.027799606323242188\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_aYe-2Glruu4_2.mid\n",
      "0.0330815315246582\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_sEQf5lcnj_o_3.mid\n",
      "0.05482006072998047\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_mGYKWCVnMwQ_0.mid\n",
      "0.024104833602905273\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_J6X3rVU1H-c_2.mid\n",
      "0.033724069595336914\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Ie5koh4qvJc_32.mid\n",
      "0.01694774627685547\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_JT1XJnVmABo_2.mid\n",
      "0.011451005935668945\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_oTi57hZfArE_2.mid\n",
      "0.013314485549926758\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Y5JcZQ0xg4Y_0.mid\n",
      "0.010271072387695312\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_HQ8ISDX6PiI_1.mid\n",
      "0.017453432083129883\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_iXiSFSGws-c_1.mid\n",
      "0.019034147262573242\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Dw7M69CidJ8_3.mid\n",
      "0.00861048698425293\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3N2G21U7guk_0.mid\n",
      "0.008478879928588867\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_pNwQ9Tu_bCs_0.mid\n",
      "0.020168542861938477\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_bQS8eAXAhyE_1.mid\n",
      "0.01635909080505371\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_P-We4q3goKU_0.mid\n",
      "0.016580581665039062\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_jNzZCplNNyY_1.mid\n",
      "0.02380084991455078\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_ltxNPJda7zE_4.mid\n",
      "0.007343292236328125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_ii1Jw1-7Ka4_0.mid\n",
      "0.0059337615966796875\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_UOSlDydo94E_2.mid\n",
      "0.02585458755493164\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_Cir1kZyB7QM_1.mid\n",
      "0.010336160659790039\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_11.mid\n",
      "0.017847299575805664\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_vVTP0DOL_2Q_0.mid\n",
      "0.014261722564697266\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_HY9vPoHbgaI_0.mid\n",
      "0.015569925308227539\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__SJQaaRzD-A_0.mid\n",
      "0.016737937927246094\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_lalnGhxT3PQ_1.mid\n",
      "0.03764533996582031\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_qlbazHayULg_10.mid\n",
      "0.033333778381347656\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_c6CwY8Gbw0c_1.mid\n",
      "0.019886016845703125\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_9Xg69XrVaYU_1.mid\n",
      "0.019220352172851562\n",
      "data/emopia/EMOPIA_2.2/midis/Q1__BK2o77sTc0_0.mid\n",
      "0.023807525634765625\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_gSwv8hZGM-s_1.mid\n",
      "0.022111892700195312\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_OCqMDeD6Fmc_2.mid\n",
      "0.054841041564941406\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_MbvPxCUMSek_1.mid\n",
      "0.008356094360351562\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_DvIH5yUrboc_2.mid\n",
      "0.02501988410949707\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_3N2G21U7guk_2.mid\n",
      "0.008449554443359375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_kKFwiXKlhC8_1.mid\n",
      "0.0461881160736084\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_lalnGhxT3PQ_0.mid\n",
      "0.03475356101989746\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_QBJ6YYYpIyk_0.mid\n",
      "0.013741731643676758\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_sEQf5lcnj_o_1.mid\n",
      "0.05902743339538574\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_UuLTvhERcnk_1.mid\n",
      "0.02102828025817871\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_epX33OVpkmA_0.mid\n",
      "0.04339146614074707\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_fnqb8zqHqdY_0.mid\n",
      "0.014573097229003906\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_kopwU6zpS4k_0.mid\n",
      "0.010708332061767578\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_WpZ1rWHOCAQ_2.mid\n",
      "0.025851726531982422\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_Q5b5unyP8BM_1.mid\n",
      "0.04933953285217285\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_jIKX3ShvLxo_0.mid\n",
      "0.026879310607910156\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_AWNbcxXDmt0_0.mid\n",
      "0.031073570251464844\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_gSwv8hZGM-s_0.mid\n",
      "0.03963470458984375\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_S7Bo4-NCEDk_2.mid\n",
      "0.02047586441040039\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_I2MrA-o5H8I_0.mid\n",
      "0.02521514892578125\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_GY3f6ckBVkA_2.mid\n",
      "0.0286405086517334\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_FyK_c-TIcCA_1.mid\n",
      "0.037718772888183594\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_3_MxhSS86oU_1.mid\n",
      "0.04240250587463379\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_0J7lX-Tcx-w_1.mid\n",
      "0.04040193557739258\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_2eIsQtm4YNs_0.mid\n",
      "0.017203569412231445\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_GAvPwG2wYZM_0.mid\n",
      "0.010970830917358398\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_Fc1qk52SaKY_1.mid\n",
      "0.0314936637878418\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_mX-xs3OVhTs_0.mid\n",
      "0.016392230987548828\n",
      "data/emopia/EMOPIA_2.2/midis/Q2__S991V8N-8s_1.mid\n",
      "0.03269219398498535\n",
      "data/emopia/EMOPIA_2.2/midis/Q2_--2B4d4lQhs_2.mid\n",
      "0.047823429107666016\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_yFw_kO7DF-Y_1.mid\n",
      "0.028357505798339844\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_2Z9SjI131jA_1.mid\n",
      "0.01884317398071289\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_gwmVjvR-sVs_2.mid\n",
      "0.008822202682495117\n",
      "data/emopia/EMOPIA_2.2/midis/Q4_KxlqB3j0zys_2.mid\n",
      "0.01993560791015625\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_Ie5koh4qvJc_25.mid\n",
      "0.023735523223876953\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_s5b8viFsVJE_0.mid\n",
      "0.03056025505065918\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_miqLU2739dk_1.mid\n",
      "0.03712058067321777\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_0vLPYiPN7qY_1.mid\n",
      "0.03477764129638672\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_V3Y9L4UOcpk_2.mid\n",
      "0.028827190399169922\n",
      "data/emopia/EMOPIA_2.2/midis/Q3_gvWDOIiocuE_0.mid\n",
      "0.01770472526550293\n",
      "data/emopia/EMOPIA_2.2/midis/Q1_nOIBJHkqrE8_0.mid\n",
      "0.021236181259155273\n"
     ]
    }
   ],
   "source": [
    "encoded_sequences_path = \"data/encoded_sequences.pkl\"\n",
    "encoded_sequences = midi_encoder.encode_midi_list(midi_files_list, pkl_path=encoded_sequences_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Now create the dataset and save it in a NumPy file\n",
    "dataset_path = \"data/datasets/\"\n",
    "midi_encoder_remi.save_dataset(midi_files_list, dataset_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/codespace/venv/lib/python3.8/site-packages/numpy/lib/npyio.py:696: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n",
      "  val = np.asanyarray(val)\n"
     ]
    }
   ],
   "source": [
    "# Save the dataset as a single file\n",
    "single_file_dataset_path = \"data/single_file_dataset.npz\"\n",
    "midi_encoder_remi.save_dataset_as_single_file(glob.glob(os.path.join(dataset_path, '*.npy')), single_file_dataset_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/workspaces/ece661-final-proj2/data/dataset.py:170: UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor. (Triggered internally at ../torch/csrc/utils/tensor_new.cpp:261.)\n",
      "  self.sequences = torch.Tensor(self.sequences)\n"
     ]
    }
   ],
   "source": [
    "max_seq_len = 256\n",
    "dataset = TransformerDatasetREMI(single_file_dataset_path, seq_len=max_seq_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_size = int(0.7 * len(dataset))\n",
    "valid_size = int(0.15 * len(dataset))\n",
    "test_size = len(dataset) - train_size - valid_size\n",
    "train_dataset, valid_dataset, test_dataset = torch.utils.data.random_split(dataset, [train_size, valid_size, test_size])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train dataset size:  3507\n",
      "Validation dataset size:  751\n",
      "Test dataset size:  753\n"
     ]
    }
   ],
   "source": [
    "print(\"Train dataset size: \", len(train_dataset))\n",
    "print(\"Validation dataset size: \", len(valid_dataset))\n",
    "print(\"Test dataset size: \", len(test_dataset))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train dataset size:  54\n",
      "Validation dataset size:  11\n",
      "Test dataset size:  11\n"
     ]
    }
   ],
   "source": [
    "from torch.utils.data import DataLoader\n",
    "batch_size = 64\n",
    "\n",
    "train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=False, drop_last=True)\n",
    "valid_dataloader = DataLoader(valid_dataset, batch_size=batch_size, shuffle=False, drop_last=True)\n",
    "test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False, drop_last=True)\n",
    "\n",
    "print(\"Train dataset size: \", len(train_dataloader))\n",
    "print(\"Validation dataset size: \", len(valid_dataloader))\n",
    "print(\"Test dataset size: \", len(test_dataloader))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['ids', 'input', 'target', 'input_mask', 'target_mask'])\n",
      "tensor([602])\n",
      "tensor([ 40,  94,  24,   3,  53,  76,  19,   3,  17,  84,  64,   3,   7, 112,\n",
      "         61,  49,  53,  76,  61,  81,  40,  74,  61,  81,  75, 112,  61,  81,\n",
      "         73, 113,  61,   6,  56,  94,  15,   6,  73,  76, 166,  34,   7,  82,\n",
      "        147,  34,  53,  25, 134,  34,   7, 112,  61,  20,  40, 112,  83,  20,\n",
      "          7,  74, 134,  43,  10, 112,  52,  43,   7,  27,  50,  59,   7,  88,\n",
      "        134,   2,   3,  44,  36, 127,   3,  10, 112,  61,   3,  37,  33, 127,\n",
      "         81,  10, 112,  61,  81,   7,  27,  52,  81,  44,  25,  80,  12,   7,\n",
      "        112,  64,  12,  17,  74,  52,  12,  53, 114,  64,  20,   7,  60,  72,\n",
      "         20,  10,  82,  83,  43,  40,  29,  61,  43,  10,  74, 127,  59,   7,\n",
      "         27, 137,  59,  53,  76, 127,  59,  42,  41,  61,   2,  49,  10,  84,\n",
      "        137,  49,  40,  41,  64,  81,   7,  90,  64,  12,   7,  74,  50,  12,\n",
      "          7,  28,  64,  12,   7,  41,  61,  20,  56,  41,  61,  20,  42,  27,\n",
      "         80,  39,  44,  90,  50,  43,  40,  28,   9,  43,   7,  41,  61,  43,\n",
      "         10,  84,  45,  59,  40,  74,  61,  59,  56,  41,  83,  48,  53,  28,\n",
      "         15,   2,  49,  56,  23, 123,  49,   7,  74,  50,  81,   7,  27,  61,\n",
      "         81,  42,  36,  52,  81,  44,  82,  61,  12,  40,  27, 131,  12,  56,\n",
      "         82,  61,  12,  42,  25,  61,  39,  10,  99,  32,  39,   7,  82,  61,\n",
      "         39,  17,  33, 103,  39,  44,  36,  64,  39,  53,  25,  61,  43,  10,\n",
      "         82,  72,  43,   7])\n"
     ]
    }
   ],
   "source": [
    "for test in test_dataloader:\n",
    "    print(test.keys())\n",
    "    print(test['ids'][0])\n",
    "    print(test['input'][0])\n",
    "    # print(test['input_mask'])\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "vocab_set = set()\n",
    "for dic in train_dataset:\n",
    "    ipt = dic['input'].numpy()\n",
    "    vocab_set = vocab_set.union(set(ipt))\n",
    "vocab_size = len(vocab_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from data.dataset import ClassifierDataset\n",
    "classifier_dataset = ClassifierDataset(single_file_dataset_path, seq_len=max_seq_len, labels_path=\"data/emopia/EMOPIA_2.2/label.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "Q1, Q2, Q3, Q4 = [], [], [], []\n",
    "for dic in classifier_dataset:\n",
    "    label = dic['target']\n",
    "    if label == 0:\n",
    "        Q1.append(dic)\n",
    "    elif label == 1:\n",
    "        Q2.append(dic)\n",
    "    elif label == 2:\n",
    "        Q3.append(dic)\n",
    "    elif label == 3:\n",
    "        Q4.append(dic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier_valid = {'Q1': Q1, 'Q2': Q2, 'Q3': Q3, 'Q4': Q4}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from utils.trainer import TransformerTrainer\n",
    "# from model.transformer import Generator, Discriminator, PatchDiscriminator\n",
    "# from utils.losses import MultiCrossEntropyLoss, TransfoCrossEntropyLoss,TransfoL1Loss, wgan_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# generator = Generator(vocab_size, max_seq_len, dim=256)\n",
    "# discriminator = Discriminator(vocab_size, max_seq_len, dim=256)\n",
    "# patch_discriminator = PatchDiscriminator(vocab_size, max_seq_len, dim=256)\n",
    "\n",
    "# ce_loss = TransfoCrossEntropyLoss()\n",
    "# gan_loss = wgan_loss\n",
    "\n",
    "# device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "# generator.to(device)\n",
    "# discriminator.to(device)\n",
    "# patch_discriminator.to(device)\n",
    "\n",
    "# g_lr = 1e-4\n",
    "# d_lr = 1e-4\n",
    "\n",
    "# EPOCHS = 10\n",
    "# checkpoint_dir = \"checkpoints/model.pt\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# trainer = TransformerTrainer(generator, patch_discriminator, train_dataloader, train_dataloader, test_dataloader, ce_loss,\n",
    "#                gan_loss, device, g_lr, d_lr, vocab_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# hist = trainer.train( EPOCHS, checkpoint_dir, validate = False, log_interval=20, load=False, save=True, train_gan=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "\n",
    "import torch\n",
    "from torch import nn, Tensor\n",
    "from torch.nn import TransformerEncoder, TransformerEncoderLayer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PositionalEncoding(nn.Module):\n",
    "    def __init__(self, d_model, max_seq_length, dropout=0.1):\n",
    "        super(PositionalEncoding, self).__init__()\n",
    "        self.dropout = nn.Dropout(p=dropout)\n",
    "        \n",
    "        pe = torch.zeros(max_seq_length, d_model)\n",
    "        position = torch.arange(0, max_seq_length, dtype=torch.float).unsqueeze(1)\n",
    "        div_term = torch.exp(torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model))\n",
    "        \n",
    "        pe[:, 0::2] = torch.sin(position * div_term)\n",
    "        pe[:, 1::2] = torch.cos(position * div_term)\n",
    "        \n",
    "        self.register_buffer('pe', pe.unsqueeze(0))\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = x + self.pe[:, :x.size(1)]\n",
    "        return self.dropout(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TransformerModel(nn.Module):\n",
    "\n",
    "    def __init__(self, max_seq_length: int, ntoken: int, d_model: int, nhead: int, d_hid: int,\n",
    "                 nlayers: int, dropout: float = 0.5):\n",
    "        super().__init__()\n",
    "        self.model_type = 'Transformer'\n",
    "        self.pos_encoder = PositionalEncoding(d_model, max_seq_length, dropout)\n",
    "        encoder_layers = TransformerEncoderLayer(d_model, nhead, d_hid, dropout)\n",
    "        self.transformer_encoder = TransformerEncoder(encoder_layers, nlayers)\n",
    "        self.embedding = nn.Embedding(ntoken, d_model)\n",
    "        self.d_model = d_model\n",
    "        self.linear = nn.Linear(d_model, ntoken)\n",
    "\n",
    "        self.init_weights()\n",
    "\n",
    "    def init_weights(self) -> None:\n",
    "        initrange = 0.1\n",
    "        self.embedding.weight.data.uniform_(-initrange, initrange)\n",
    "        self.linear.bias.data.zero_()\n",
    "        self.linear.weight.data.uniform_(-initrange, initrange)\n",
    "\n",
    "    def forward(self, src: Tensor, src_mask: Tensor = None) -> Tensor:\n",
    "        src = self.embedding(src) * math.sqrt(self.d_model)\n",
    "        src = self.pos_encoder(src)\n",
    "        if src_mask is None:\n",
    "            \"\"\"Generate a square causal mask for the sequence. The masked positions are filled with float('-inf').\n",
    "            Unmasked positions are filled with float(0.0).\n",
    "            \"\"\"\n",
    "            src_mask = nn.Transformer.generate_square_subsequent_mask(len(src)).to(device)\n",
    "            # Add padding mask if provided\n",
    "        output = self.transformer_encoder(src, src_mask)\n",
    "        output = self.linear(output)\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/codespace/venv/lib/python3.8/site-packages/torch/nn/modules/transformer.py:282: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.self_attn.batch_first was not True(use batch_first for better inference performance)\n",
      "  warnings.warn(f\"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}\")\n"
     ]
    }
   ],
   "source": [
    "ntokens = vocab_size  # size of vocabulary\n",
    "emsize = 256  # embedding dimension\n",
    "d_hid = 512  # dimension of the feedforward network model in ``nn.TransformerEncoder``\n",
    "nlayers = 6  # number of ``nn.TransformerEncoderLayer`` in ``nn.TransformerEncoder``\n",
    "nhead = 8  # number of heads in ``nn.MultiheadAttention``\n",
    "dropout = 0.1  # dropout probability\n",
    "max_seq_length = max_seq_len\n",
    "model = TransformerModel(max_seq_len, ntokens, emsize, nhead, d_hid, nlayers, dropout).to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TransformerModel(\n",
      "  (pos_encoder): PositionalEncoding(\n",
      "    (dropout): Dropout(p=0.1, inplace=False)\n",
      "  )\n",
      "  (transformer_encoder): TransformerEncoder(\n",
      "    (layers): ModuleList(\n",
      "      (0-5): 6 x TransformerEncoderLayer(\n",
      "        (self_attn): MultiheadAttention(\n",
      "          (out_proj): NonDynamicallyQuantizableLinear(in_features=256, out_features=256, bias=True)\n",
      "        )\n",
      "        (linear1): Linear(in_features=256, out_features=512, bias=True)\n",
      "        (dropout): Dropout(p=0.1, inplace=False)\n",
      "        (linear2): Linear(in_features=512, out_features=256, bias=True)\n",
      "        (norm1): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n",
      "        (norm2): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n",
      "        (dropout1): Dropout(p=0.1, inplace=False)\n",
      "        (dropout2): Dropout(p=0.1, inplace=False)\n",
      "      )\n",
      "    )\n",
      "  )\n",
      "  (embedding): Embedding(198, 256)\n",
      "  (linear): Linear(in_features=256, out_features=198, bias=True)\n",
      ")\n"
     ]
    }
   ],
   "source": [
    "print(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.CrossEntropyLoss()\n",
    "lr = 0.001  # learning rate\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n",
    "scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=5, gamma=0.1)\n",
    "epochs = 100  # The number of epochs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def early_stopping(val_loss, epoch,best_loss,epochs_no_improve, patience=10):\n",
    "    if epoch == 0:\n",
    "        torch.save(model.state_dict(), 'checkpoints/transformer_v3.pt')\n",
    "        best_loss = val_loss\n",
    "        epochs_no_improve = 0\n",
    "        return False, best_loss, epochs_no_improve\n",
    "    else:\n",
    "        if val_loss < best_loss:\n",
    "            torch.save(model.state_dict(), 'checkpoints/transformer_v3.pt')\n",
    "            print(f\"Saved model at epoch { epoch } with validation loss of {val_loss}\")\n",
    "            best_loss = val_loss\n",
    "            epochs_no_improve = 0\n",
    "            return False, best_loss, epochs_no_improve\n",
    "        else:\n",
    "            epochs_no_improve += 1\n",
    "            if epochs_no_improve >= patience:\n",
    "                print('Early stopping!')\n",
    "                return True, best_loss, epochs_no_improve\n",
    "            else:\n",
    "                return False, best_loss, epochs_no_improve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [ 0 / 100 ]; TrainLoss: 3.93455; ValidLoss: 2.98324\n",
      "Epoch: [ 1 / 100 ]; TrainLoss: 2.95194; ValidLoss: 2.94213\n",
      "Saved model at epoch 1 with validation loss of 2.942125688899647\n",
      "Epoch: [ 2 / 100 ]; TrainLoss: 2.92879; ValidLoss: 2.93443\n",
      "Saved model at epoch 2 with validation loss of 2.9344312277707187\n",
      "Epoch: [ 3 / 100 ]; TrainLoss: 2.92138; ValidLoss: 2.93046\n",
      "Saved model at epoch 3 with validation loss of 2.9304559230804443\n",
      "Epoch: [ 4 / 100 ]; TrainLoss: 2.91624; ValidLoss: 2.92786\n",
      "Saved model at epoch 4 with validation loss of 2.927857594056563\n",
      "Epoch: [ 5 / 100 ]; TrainLoss: 2.90325; ValidLoss: 2.91324\n",
      "Saved model at epoch 5 with validation loss of 2.9132426218553023\n",
      "Epoch: [ 6 / 100 ]; TrainLoss: 2.89989; ValidLoss: 2.91311\n",
      "Saved model at epoch 6 with validation loss of 2.9131095192649146\n",
      "Epoch: [ 7 / 100 ]; TrainLoss: 2.89920; ValidLoss: 2.91316\n",
      "Epoch: [ 8 / 100 ]; TrainLoss: 2.89897; ValidLoss: 2.91324\n",
      "Epoch: [ 9 / 100 ]; TrainLoss: 2.89802; ValidLoss: 2.91335\n",
      "Epoch: [ 10 / 100 ]; TrainLoss: 2.89600; ValidLoss: 2.91232\n",
      "Saved model at epoch 10 with validation loss of 2.9123184464194556\n",
      "Epoch: [ 11 / 100 ]; TrainLoss: 2.89583; ValidLoss: 2.91214\n",
      "Saved model at epoch 11 with validation loss of 2.9121424935080786\n",
      "Epoch: [ 12 / 100 ]; TrainLoss: 2.89553; ValidLoss: 2.91210\n",
      "Saved model at epoch 12 with validation loss of 2.9120994264429267\n",
      "Epoch: [ 13 / 100 ]; TrainLoss: 2.89550; ValidLoss: 2.91207\n",
      "Saved model at epoch 13 with validation loss of 2.9120718999342485\n",
      "Epoch: [ 14 / 100 ]; TrainLoss: 2.89533; ValidLoss: 2.91208\n",
      "Epoch: [ 15 / 100 ]; TrainLoss: 2.89493; ValidLoss: 2.91207\n",
      "Saved model at epoch 15 with validation loss of 2.912066698074341\n",
      "Epoch: [ 16 / 100 ]; TrainLoss: 2.89492; ValidLoss: 2.91205\n",
      "Saved model at epoch 16 with validation loss of 2.912054191936146\n",
      "Epoch: [ 17 / 100 ]; TrainLoss: 2.89493; ValidLoss: 2.91205\n",
      "Saved model at epoch 17 with validation loss of 2.9120459339835425\n",
      "Epoch: [ 18 / 100 ]; TrainLoss: 2.89503; ValidLoss: 2.91204\n",
      "Saved model at epoch 18 with validation loss of 2.912038651379672\n",
      "Epoch: [ 19 / 100 ]; TrainLoss: 2.89503; ValidLoss: 2.91204\n",
      "Saved model at epoch 19 with validation loss of 2.912035248496316\n",
      "Epoch: [ 20 / 100 ]; TrainLoss: 2.89477; ValidLoss: 2.91203\n",
      "Saved model at epoch 20 with validation loss of 2.9120347933335737\n",
      "Epoch: [ 21 / 100 ]; TrainLoss: 2.89509; ValidLoss: 2.91203\n",
      "Saved model at epoch 21 with validation loss of 2.912034338170832\n",
      "Epoch: [ 22 / 100 ]; TrainLoss: 2.89499; ValidLoss: 2.91203\n",
      "Saved model at epoch 22 with validation loss of 2.9120336662639272\n",
      "Epoch: [ 23 / 100 ]; TrainLoss: 2.89468; ValidLoss: 2.91203\n",
      "Saved model at epoch 23 with validation loss of 2.9120331460779365\n",
      "Epoch: [ 24 / 100 ]; TrainLoss: 2.89457; ValidLoss: 2.91203\n",
      "Saved model at epoch 24 with validation loss of 2.9120328426361084\n",
      "Epoch: [ 25 / 100 ]; TrainLoss: 2.89486; ValidLoss: 2.91203\n",
      "Saved model at epoch 25 with validation loss of 2.912032647566362\n",
      "Epoch: [ 26 / 100 ]; TrainLoss: 2.89488; ValidLoss: 2.91203\n",
      "Saved model at epoch 26 with validation loss of 2.912032625891946\n",
      "Epoch: [ 27 / 100 ]; TrainLoss: 2.89493; ValidLoss: 2.91203\n",
      "Epoch: [ 28 / 100 ]; TrainLoss: 2.89485; ValidLoss: 2.91203\n",
      "Epoch: [ 29 / 100 ]; TrainLoss: 2.89480; ValidLoss: 2.91203\n",
      "Saved model at epoch 29 with validation loss of 2.9120325391942803\n",
      "Epoch: [ 30 / 100 ]; TrainLoss: 2.89473; ValidLoss: 2.91203\n",
      "Epoch: [ 31 / 100 ]; TrainLoss: 2.89475; ValidLoss: 2.91203\n",
      "Epoch: [ 32 / 100 ]; TrainLoss: 2.89473; ValidLoss: 2.91203\n",
      "Epoch: [ 33 / 100 ]; TrainLoss: 2.89473; ValidLoss: 2.91203\n",
      "Epoch: [ 34 / 100 ]; TrainLoss: 2.89471; ValidLoss: 2.91203\n",
      "Epoch: [ 35 / 100 ]; TrainLoss: 2.89490; ValidLoss: 2.91203\n",
      "Epoch: [ 36 / 100 ]; TrainLoss: 2.89500; ValidLoss: 2.91203\n",
      "Epoch: [ 37 / 100 ]; TrainLoss: 2.89456; ValidLoss: 2.91203\n",
      "Epoch: [ 38 / 100 ]; TrainLoss: 2.89467; ValidLoss: 2.91203\n",
      "Saved model at epoch 38 with validation loss of 2.9120325175198642\n",
      "Epoch: [ 39 / 100 ]; TrainLoss: 2.89478; ValidLoss: 2.91203\n",
      "Epoch: [ 40 / 100 ]; TrainLoss: 2.89472; ValidLoss: 2.91203\n",
      "Epoch: [ 41 / 100 ]; TrainLoss: 2.89492; ValidLoss: 2.91203\n",
      "Epoch: [ 42 / 100 ]; TrainLoss: 2.89470; ValidLoss: 2.91203\n",
      "Epoch: [ 43 / 100 ]; TrainLoss: 2.89495; ValidLoss: 2.91203\n",
      "Epoch: [ 44 / 100 ]; TrainLoss: 2.89481; ValidLoss: 2.91203\n",
      "Epoch: [ 45 / 100 ]; TrainLoss: 2.89476; ValidLoss: 2.91203\n",
      "Epoch: [ 46 / 100 ]; TrainLoss: 2.89485; ValidLoss: 2.91203\n",
      "Epoch: [ 47 / 100 ]; TrainLoss: 2.89474; ValidLoss: 2.91203\n",
      "Epoch: [ 48 / 100 ]; TrainLoss: 2.89481; ValidLoss: 2.91203\n",
      "Early stopping!\n",
      "Training complete!\n",
      "Best loss:  2.9120325175198642\n"
     ]
    }
   ],
   "source": [
    "# save the train valid losses for each epoch, and print them after epoch ends\n",
    "best_loss = 50\n",
    "epochs_no_improve = 0\n",
    "train_losses = []\n",
    "valid_losses = []\n",
    "\n",
    "\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    model.train()\n",
    "    train_loss = 0\n",
    "    valid_loss = 0\n",
    "    for batch in train_dataloader:\n",
    "        src_data = batch['input'].to(device)\n",
    "        tgt_data = batch['target'].to(device)\n",
    "        optimizer.zero_grad()\n",
    "        output = model(src_data)\n",
    "        loss = criterion(output.view(-1, vocab_size), tgt_data.view(-1))\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        train_loss+=loss.item()\n",
    "    train_losses.append(train_loss/len(train_dataloader))\n",
    "\n",
    "    # evaluate on validation set\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        for batch in valid_dataloader:\n",
    "            src_data = batch['input'].to(device)\n",
    "            tgt_data = batch['target'].to(device)\n",
    "            output = model(src_data)\n",
    "            loss = criterion(output.view(-1, vocab_size), tgt_data.view(-1))\n",
    "            valid_loss+=loss.item()\n",
    "            \n",
    "    valid_losses.append(valid_loss/len(valid_dataloader))\n",
    "\n",
    "    print(\"Epoch: [ {} / {} ]; TrainLoss: {:.5f}; ValidLoss: {:.5f}\".format(\n",
    "        epoch, epochs, train_loss/len(train_dataloader), valid_loss/len(valid_dataloader)\n",
    "    ))\n",
    "\n",
    "\n",
    "    cond, best_loss, epochs_no_improve = early_stopping(valid_loss/len(valid_dataloader), epoch, best_loss,epochs_no_improve, patience=10)\n",
    "    if cond:\n",
    "        break\n",
    "    scheduler.step()\n",
    "print(\"Training complete!\")\n",
    "print(\"Best loss: \", best_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": {
      "image/png": {
       "height": 413,
       "width": 552
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot losses\n",
    "import matplotlib.pyplot as plt\n",
    "%config InlineBackend.figure_format = 'retina' # Makes the plots clear on high-res screens\n",
    "plt.plot(train_losses, label='Training loss')\n",
    "plt.plot(valid_losses, label='Validation loss')\n",
    "plt.legend(frameon=False)\n",
    "plt.tight_layout\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.nn.functional import softmax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/codespace/venv/lib/python3.8/site-packages/torch/nn/modules/transformer.py:282: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.self_attn.batch_first was not True(use batch_first for better inference performance)\n",
      "  warnings.warn(f\"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}\")\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = TransformerModel(max_seq_len, ntokens, emsize, nhead, d_hid, nlayers, dropout).to(device)\n",
    "model.load_state_dict(torch.load(\"checkpoints/transformer_v3.pt\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pad_features(input, seq_length):\n",
    "    # getting the correct rows x cols shape\n",
    "    features = np.zeros((len(input), seq_length), dtype=int)\n",
    "\n",
    "    # for each review, I grab that review and put it into features\n",
    "    for i, row in enumerate(input):\n",
    "        features[i, :len(row)] = np.array(row)[:seq_length]\n",
    "\n",
    "    return features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for batch in test_dataloader:\n",
    "    data = batch['input'].to(device)\n",
    "    print(data.shape)\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "model.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = []\n",
    "original_musics = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    for batch in test_dataloader:\n",
    "        input = batch['input'].to(device)\n",
    "        target = batch['target'].to(device)\n",
    "        for i in range(len(input)):\n",
    "            generated_tokens = [input[i][0]]\n",
    "            output = model(input[i])\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "                # current_token = torch.argmax(probabilities[j]).item()\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "                else:\n",
    "                    generated_tokens.append(current_token)\n",
    "\n",
    "            generated_musics.append(generated_tokens)\n",
    "            original_musics.append(target[i].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(generated_musics)):\n",
    "    midi_encoder_remi.words_to_midi(generated_musics[i],f'generated_musics/transformer_1/transformer{i}.mid')\n",
    "    midi_encoder_remi.words_to_midi(original_musics[i],f'generated_musics/original/original{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "model.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = {'Q1':[], 'Q2':[], 'Q3':[], 'Q4':[]}\n",
    "\n",
    "with torch.no_grad():\n",
    "    for c in classifier_valid.keys():\n",
    "        print(c)\n",
    "        for batch in classifier_valid[c]:\n",
    "            input = batch['input'].to(device)\n",
    "            target = batch['target'].to(device)\n",
    "            generated_tokens = [input[0]]\n",
    "            output = model(input)\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "                # current_token = torch.argmax(probabilities[j]).item()\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "                else:\n",
    "                    generated_tokens.append(current_token)\n",
    "\n",
    "            generated_musics[c].append(generated_tokens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in generated_musics.keys():\n",
    "    for i in range(len(generated_musics[c])):\n",
    "        midi_encoder_remi.words_to_midi(generated_musics[c][i],f'generated_musics/transformer_1_by_emotion/{c}/transformer{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.utils.data as data\n",
    "import math\n",
    "import copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "class MultiHeadAttention(nn.Module):\n",
    "    def __init__(self, d_model, num_heads):\n",
    "        super(MultiHeadAttention, self).__init__()\n",
    "        assert d_model % num_heads == 0, \"d_model must be divisible by num_heads\"\n",
    "        \n",
    "        self.d_model = d_model\n",
    "        self.num_heads = num_heads\n",
    "        self.d_k = d_model // num_heads\n",
    "        \n",
    "        self.W_q = nn.Linear(d_model, d_model)\n",
    "        self.W_k = nn.Linear(d_model, d_model)\n",
    "        self.W_v = nn.Linear(d_model, d_model)\n",
    "        self.W_o = nn.Linear(d_model, d_model)\n",
    "        \n",
    "    def scaled_dot_product_attention(self, Q, K, V, mask=None):\n",
    "        attn_scores = torch.matmul(Q, K.transpose(-2, -1)) / math.sqrt(self.d_k)\n",
    "        if mask is not None:\n",
    "            attn_scores = attn_scores.masked_fill(mask == 0, -1e9)\n",
    "        attn_probs = torch.softmax(attn_scores, dim=-1)\n",
    "        output = torch.matmul(attn_probs, V)\n",
    "        return output\n",
    "        \n",
    "    def split_heads(self, x):\n",
    "        batch_size, seq_length, d_model = x.size()\n",
    "        return x.view(batch_size, seq_length, self.num_heads, self.d_k).transpose(1, 2)\n",
    "        \n",
    "    def combine_heads(self, x):\n",
    "        batch_size, _, seq_length, d_k = x.size()\n",
    "        return x.transpose(1, 2).contiguous().view(batch_size, seq_length, self.d_model)\n",
    "        \n",
    "    def forward(self, Q, K, V, mask=None):\n",
    "        Q = self.split_heads(self.W_q(Q))\n",
    "        K = self.split_heads(self.W_k(K))\n",
    "        V = self.split_heads(self.W_v(V))\n",
    "        \n",
    "        attn_output = self.scaled_dot_product_attention(Q, K, V, mask)\n",
    "        output = self.W_o(self.combine_heads(attn_output))\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PositionWiseFeedForward(nn.Module):\n",
    "    def __init__(self, d_model, d_ff):\n",
    "        super(PositionWiseFeedForward, self).__init__()\n",
    "        self.fc1 = nn.Linear(d_model, d_ff)\n",
    "        self.fc2 = nn.Linear(d_ff, d_model)\n",
    "        self.relu = nn.ReLU()\n",
    "\n",
    "    def forward(self, x):\n",
    "        return self.fc2(self.relu(self.fc1(x)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PositionalEncoding(nn.Module):\n",
    "    def __init__(self, d_model, max_seq_length):\n",
    "        super(PositionalEncoding, self).__init__()\n",
    "        \n",
    "        pe = torch.zeros(max_seq_length, d_model)\n",
    "        position = torch.arange(0, max_seq_length, dtype=torch.float).unsqueeze(1)\n",
    "        div_term = torch.exp(torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model))\n",
    "        \n",
    "        pe[:, 0::2] = torch.sin(position * div_term)\n",
    "        pe[:, 1::2] = torch.cos(position * div_term)\n",
    "        \n",
    "        self.register_buffer('pe', pe.unsqueeze(0))\n",
    "        \n",
    "    def forward(self, x):\n",
    "        return x + self.pe[:, :x.size(1)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "class EncoderLayer(nn.Module):\n",
    "    def __init__(self, d_model, num_heads, d_ff, dropout):\n",
    "        super(EncoderLayer, self).__init__()\n",
    "        self.self_attn = MultiHeadAttention(d_model, num_heads)\n",
    "        self.feed_forward = PositionWiseFeedForward(d_model, d_ff)\n",
    "        self.norm1 = nn.LayerNorm(d_model)\n",
    "        self.norm2 = nn.LayerNorm(d_model)\n",
    "        self.dropout = nn.Dropout(dropout)\n",
    "        \n",
    "    def forward(self, x, mask):\n",
    "        attn_output = self.self_attn(x, x, x, mask)\n",
    "        x = self.norm1(x + self.dropout(attn_output))\n",
    "        ff_output = self.feed_forward(x)\n",
    "        x = self.norm2(x + self.dropout(ff_output))\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DecoderLayer(nn.Module):\n",
    "    def __init__(self, d_model, num_heads, d_ff, dropout):\n",
    "        super(DecoderLayer, self).__init__()\n",
    "        self.self_attn = MultiHeadAttention(d_model, num_heads)\n",
    "        self.cross_attn = MultiHeadAttention(d_model, num_heads)\n",
    "        self.feed_forward = PositionWiseFeedForward(d_model, d_ff)\n",
    "        self.norm1 = nn.LayerNorm(d_model)\n",
    "        self.norm2 = nn.LayerNorm(d_model)\n",
    "        self.norm3 = nn.LayerNorm(d_model)\n",
    "        self.dropout = nn.Dropout(dropout)\n",
    "        \n",
    "    def forward(self, x, enc_output, src_mask, tgt_mask):\n",
    "        attn_output = self.self_attn(x, x, x, tgt_mask)\n",
    "        x = self.norm1(x + self.dropout(attn_output))\n",
    "        attn_output = self.cross_attn(x, enc_output, enc_output, src_mask)\n",
    "        x = self.norm2(x + self.dropout(attn_output))\n",
    "        ff_output = self.feed_forward(x)\n",
    "        x = self.norm3(x + self.dropout(ff_output))\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Transformer(nn.Module):\n",
    "    def __init__(self, src_vocab_size, tgt_vocab_size, d_model, num_heads, num_layers, d_ff, max_seq_length, dropout):\n",
    "        super(Transformer, self).__init__()\n",
    "        self.encoder_embedding = nn.Embedding(src_vocab_size, d_model)\n",
    "        self.decoder_embedding = nn.Embedding(tgt_vocab_size, d_model)\n",
    "        self.positional_encoding = PositionalEncoding(d_model, max_seq_length)\n",
    "\n",
    "        self.encoder_layers = nn.ModuleList([EncoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])\n",
    "        self.decoder_layers = nn.ModuleList([DecoderLayer(d_model, num_heads, d_ff, dropout) for _ in range(num_layers)])\n",
    "\n",
    "        self.fc = nn.Linear(d_model, tgt_vocab_size)\n",
    "        self.dropout = nn.Dropout(dropout)\n",
    "\n",
    "    def generate_mask(self, src, tgt):\n",
    "        src_mask = (src != 0).unsqueeze(1).unsqueeze(2)\n",
    "        tgt_mask = (tgt != 0).unsqueeze(1).unsqueeze(3)\n",
    "        seq_length = tgt.size(1)\n",
    "        nopeak_mask = (1 - torch.triu(torch.ones(1, seq_length, seq_length), diagonal=1)).bool().to(device)\n",
    "        tgt_mask = tgt_mask & nopeak_mask\n",
    "        return src_mask, tgt_mask\n",
    "\n",
    "    def forward(self, src, tgt):\n",
    "        src_mask, tgt_mask = self.generate_mask(src, tgt)\n",
    "        src_embedded = self.dropout(self.positional_encoding(self.encoder_embedding(src)))\n",
    "        tgt_embedded = self.dropout(self.positional_encoding(self.decoder_embedding(tgt)))\n",
    "\n",
    "        enc_output = src_embedded\n",
    "        for enc_layer in self.encoder_layers:\n",
    "            enc_output = enc_layer(enc_output, src_mask)\n",
    "\n",
    "        dec_output = tgt_embedded\n",
    "        for dec_layer in self.decoder_layers:\n",
    "            dec_output = dec_layer(dec_output, enc_output, src_mask, tgt_mask)\n",
    "\n",
    "        output = self.fc(dec_output)\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Transformer(\n",
       "  (encoder_embedding): Embedding(198, 512)\n",
       "  (decoder_embedding): Embedding(198, 512)\n",
       "  (positional_encoding): PositionalEncoding()\n",
       "  (encoder_layers): ModuleList(\n",
       "    (0-5): 6 x EncoderLayer(\n",
       "      (self_attn): MultiHeadAttention(\n",
       "        (W_q): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_k): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_v): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_o): Linear(in_features=512, out_features=512, bias=True)\n",
       "      )\n",
       "      (feed_forward): PositionWiseFeedForward(\n",
       "        (fc1): Linear(in_features=512, out_features=256, bias=True)\n",
       "        (fc2): Linear(in_features=256, out_features=512, bias=True)\n",
       "        (relu): ReLU()\n",
       "      )\n",
       "      (norm1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      (norm2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      (dropout): Dropout(p=0.2, inplace=False)\n",
       "    )\n",
       "  )\n",
       "  (decoder_layers): ModuleList(\n",
       "    (0-5): 6 x DecoderLayer(\n",
       "      (self_attn): MultiHeadAttention(\n",
       "        (W_q): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_k): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_v): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_o): Linear(in_features=512, out_features=512, bias=True)\n",
       "      )\n",
       "      (cross_attn): MultiHeadAttention(\n",
       "        (W_q): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_k): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_v): Linear(in_features=512, out_features=512, bias=True)\n",
       "        (W_o): Linear(in_features=512, out_features=512, bias=True)\n",
       "      )\n",
       "      (feed_forward): PositionWiseFeedForward(\n",
       "        (fc1): Linear(in_features=512, out_features=256, bias=True)\n",
       "        (fc2): Linear(in_features=256, out_features=512, bias=True)\n",
       "        (relu): ReLU()\n",
       "      )\n",
       "      (norm1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      (norm2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      (norm3): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
       "      (dropout): Dropout(p=0.2, inplace=False)\n",
       "    )\n",
       "  )\n",
       "  (fc): Linear(in_features=512, out_features=198, bias=True)\n",
       "  (dropout): Dropout(p=0.2, inplace=False)\n",
       ")"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "src_vocab_size = vocab_size\n",
    "tgt_vocab_size = vocab_size\n",
    "d_model = 512\n",
    "num_heads = 8\n",
    "num_layers = 6\n",
    "d_ff = 256\n",
    "max_seq_length = max_seq_len\n",
    "dropout = 0.2\n",
    "epochs = 100\n",
    "\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "transformer = Transformer(src_vocab_size, tgt_vocab_size, d_model, num_heads, num_layers, d_ff, max_seq_length, dropout)\n",
    "transformer.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "def early_stopping(val_loss, epoch,best_loss,epochs_no_improve, patience=10, checkpoint_dir=\"checkpoints/transformer.pt\"):\n",
    "    if epoch == 0:\n",
    "        torch.save(transformer.state_dict(), 'checkpoints/transformer_v3.pt')\n",
    "        best_loss = val_loss\n",
    "        epochs_no_improve = 0\n",
    "        return False, best_loss, epochs_no_improve\n",
    "    else:\n",
    "        if val_loss < best_loss:\n",
    "            torch.save(transformer.state_dict(), checkpoint_dir)\n",
    "            print(f\"Saved model at epoch { epoch } with validation loss of {val_loss}\")\n",
    "            best_loss = val_loss\n",
    "            epochs_no_improve = 0\n",
    "            return False, best_loss, epochs_no_improve\n",
    "        else:\n",
    "            epochs_no_improve += 1\n",
    "            if epochs_no_improve >= patience:\n",
    "                print('Early stopping!')\n",
    "                return True, best_loss, epochs_no_improve\n",
    "            else:\n",
    "                return False, best_loss, epochs_no_improve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: [ 0 / 100 ]; TrainLoss: 4.08579; ValidLoss: 3.33072\n",
      "Epoch: [ 1 / 100 ]; TrainLoss: 2.89704; ValidLoss: 2.58892\n",
      "Saved model at epoch 1 with validation loss of 2.588922478935935\n",
      "Epoch: [ 2 / 100 ]; TrainLoss: 2.42888; ValidLoss: 2.17449\n",
      "Saved model at epoch 2 with validation loss of 2.174491145394065\n",
      "Epoch: [ 3 / 100 ]; TrainLoss: 2.12409; ValidLoss: 1.97018\n",
      "Saved model at epoch 3 with validation loss of 1.970175949009982\n",
      "Epoch: [ 4 / 100 ]; TrainLoss: 1.91861; ValidLoss: 1.75791\n",
      "Saved model at epoch 4 with validation loss of 1.7579066645015369\n",
      "Epoch: [ 5 / 100 ]; TrainLoss: 1.74842; ValidLoss: 1.60031\n",
      "Saved model at epoch 5 with validation loss of 1.6003079631111838\n",
      "Epoch: [ 6 / 100 ]; TrainLoss: 1.62358; ValidLoss: 1.45099\n",
      "Saved model at epoch 6 with validation loss of 1.4509866129268298\n",
      "Epoch: [ 7 / 100 ]; TrainLoss: 1.49694; ValidLoss: 1.30291\n",
      "Saved model at epoch 7 with validation loss of 1.3029079870744185\n",
      "Epoch: [ 8 / 100 ]; TrainLoss: 1.37773; ValidLoss: 1.16068\n",
      "Saved model at epoch 8 with validation loss of 1.1606784842231057\n",
      "Epoch: [ 9 / 100 ]; TrainLoss: 1.25525; ValidLoss: 1.00846\n",
      "Saved model at epoch 9 with validation loss of 1.0084597143259915\n",
      "Epoch: [ 10 / 100 ]; TrainLoss: 1.13818; ValidLoss: 0.86763\n",
      "Saved model at epoch 10 with validation loss of 0.8676301674409346\n",
      "Epoch: [ 11 / 100 ]; TrainLoss: 1.02881; ValidLoss: 0.82618\n",
      "Saved model at epoch 11 with validation loss of 0.8261802304874767\n",
      "Epoch: [ 12 / 100 ]; TrainLoss: 0.93344; ValidLoss: 0.63080\n",
      "Saved model at epoch 12 with validation loss of 0.6307993368668989\n",
      "Epoch: [ 13 / 100 ]; TrainLoss: 0.85800; ValidLoss: 0.62682\n",
      "Saved model at epoch 13 with validation loss of 0.6268177303400907\n",
      "Epoch: [ 14 / 100 ]; TrainLoss: 0.76948; ValidLoss: 0.47529\n",
      "Saved model at epoch 14 with validation loss of 0.475294522263787\n",
      "Epoch: [ 15 / 100 ]; TrainLoss: 0.69965; ValidLoss: 0.44456\n",
      "Saved model at epoch 15 with validation loss of 0.44455957954580133\n",
      "Epoch: [ 16 / 100 ]; TrainLoss: 0.61952; ValidLoss: 0.33085\n",
      "Saved model at epoch 16 with validation loss of 0.33085193688219244\n",
      "Epoch: [ 17 / 100 ]; TrainLoss: 0.54847; ValidLoss: 0.24457\n",
      "Saved model at epoch 17 with validation loss of 0.2445738302035765\n",
      "Epoch: [ 18 / 100 ]; TrainLoss: 0.47288; ValidLoss: 0.21932\n",
      "Saved model at epoch 18 with validation loss of 0.2193207253109325\n",
      "Epoch: [ 19 / 100 ]; TrainLoss: 0.40905; ValidLoss: 0.16409\n",
      "Saved model at epoch 19 with validation loss of 0.1640898504040458\n",
      "Epoch: [ 20 / 100 ]; TrainLoss: 0.32879; ValidLoss: 0.10376\n",
      "Saved model at epoch 20 with validation loss of 0.10375928472388875\n",
      "Epoch: [ 21 / 100 ]; TrainLoss: 0.28330; ValidLoss: 0.06328\n",
      "Saved model at epoch 21 with validation loss of 0.06327936832200397\n",
      "Epoch: [ 22 / 100 ]; TrainLoss: 0.21732; ValidLoss: 0.05018\n",
      "Saved model at epoch 22 with validation loss of 0.050176918506622314\n",
      "Epoch: [ 23 / 100 ]; TrainLoss: 0.17288; ValidLoss: 0.04020\n",
      "Saved model at epoch 23 with validation loss of 0.040197144855152474\n",
      "Epoch: [ 24 / 100 ]; TrainLoss: 0.13753; ValidLoss: 0.02550\n",
      "Saved model at epoch 24 with validation loss of 0.025498811494220387\n",
      "Epoch: [ 25 / 100 ]; TrainLoss: 0.10589; ValidLoss: 0.02002\n",
      "Saved model at epoch 25 with validation loss of 0.020019519904797726\n",
      "Epoch: [ 26 / 100 ]; TrainLoss: 0.08628; ValidLoss: 0.01635\n",
      "Saved model at epoch 26 with validation loss of 0.016354528001763603\n",
      "Epoch: [ 27 / 100 ]; TrainLoss: 0.06992; ValidLoss: 0.01516\n",
      "Saved model at epoch 27 with validation loss of 0.015159385448152369\n",
      "Epoch: [ 28 / 100 ]; TrainLoss: 0.05236; ValidLoss: 0.01319\n",
      "Saved model at epoch 28 with validation loss of 0.013186928900805387\n",
      "Epoch: [ 29 / 100 ]; TrainLoss: 0.05059; ValidLoss: 0.01209\n",
      "Saved model at epoch 29 with validation loss of 0.012093778106976639\n",
      "Epoch: [ 30 / 100 ]; TrainLoss: 0.03528; ValidLoss: 0.01136\n",
      "Saved model at epoch 30 with validation loss of 0.011362806724553759\n",
      "Epoch: [ 31 / 100 ]; TrainLoss: 0.03071; ValidLoss: 0.01098\n",
      "Saved model at epoch 31 with validation loss of 0.010979198342697187\n",
      "Epoch: [ 32 / 100 ]; TrainLoss: 0.02749; ValidLoss: 0.01025\n",
      "Saved model at epoch 32 with validation loss of 0.010246573651040142\n",
      "Epoch: [ 33 / 100 ]; TrainLoss: 0.02426; ValidLoss: 0.01173\n",
      "Epoch: [ 34 / 100 ]; TrainLoss: 0.04960; ValidLoss: 0.00959\n",
      "Saved model at epoch 34 with validation loss of 0.009592007379978895\n",
      "Epoch: [ 35 / 100 ]; TrainLoss: 0.01651; ValidLoss: 0.00963\n",
      "Epoch: [ 36 / 100 ]; TrainLoss: 0.01581; ValidLoss: 0.00945\n",
      "Saved model at epoch 36 with validation loss of 0.009446269620887258\n",
      "Epoch: [ 37 / 100 ]; TrainLoss: 0.01628; ValidLoss: 0.00949\n",
      "Epoch: [ 38 / 100 ]; TrainLoss: 0.01621; ValidLoss: 0.00948\n",
      "Epoch: [ 39 / 100 ]; TrainLoss: 0.02606; ValidLoss: 0.00972\n",
      "Epoch: [ 40 / 100 ]; TrainLoss: 0.01499; ValidLoss: 0.00933\n",
      "Saved model at epoch 40 with validation loss of 0.009332646488804709\n",
      "Epoch: [ 41 / 100 ]; TrainLoss: 0.01217; ValidLoss: 0.00934\n",
      "Epoch: [ 42 / 100 ]; TrainLoss: 0.06626; ValidLoss: 0.00957\n",
      "Epoch: [ 43 / 100 ]; TrainLoss: 0.01373; ValidLoss: 0.00889\n",
      "Saved model at epoch 43 with validation loss of 0.008891902271319519\n",
      "Epoch: [ 44 / 100 ]; TrainLoss: 0.01035; ValidLoss: 0.00905\n",
      "Epoch: [ 45 / 100 ]; TrainLoss: 0.01000; ValidLoss: 0.00913\n",
      "Epoch: [ 46 / 100 ]; TrainLoss: 0.01029; ValidLoss: 0.00920\n",
      "Epoch: [ 47 / 100 ]; TrainLoss: 0.01039; ValidLoss: 0.00936\n",
      "Epoch: [ 48 / 100 ]; TrainLoss: 0.01108; ValidLoss: 0.00935\n",
      "Epoch: [ 49 / 100 ]; TrainLoss: 0.01119; ValidLoss: 0.00937\n",
      "Epoch: [ 50 / 100 ]; TrainLoss: 0.01074; ValidLoss: 0.00979\n",
      "Epoch: [ 51 / 100 ]; TrainLoss: 0.01369; ValidLoss: 0.00936\n",
      "Epoch: [ 52 / 100 ]; TrainLoss: 0.00834; ValidLoss: 0.00964\n",
      "Epoch: [ 53 / 100 ]; TrainLoss: 0.00835; ValidLoss: 0.00955\n",
      "Early stopping!\n",
      "Training complete!\n",
      "Best loss:  0.008891902271319519\n"
     ]
    }
   ],
   "source": [
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.Adam(transformer.parameters(), lr=0.0001, betas=(0.9, 0.98), eps=1e-9)\n",
    "\n",
    "best_loss = 50\n",
    "epochs_no_improve = 0\n",
    "\n",
    "# save the train valid losses for each epoch, and print them after epoch ends\n",
    "\n",
    "train_losses = []\n",
    "valid_losses = []\n",
    "\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    transformer.train()\n",
    "    train_loss = 0\n",
    "    valid_loss = 0\n",
    "    for batch in train_dataloader:\n",
    "        src_data = batch['input'].to(device)\n",
    "        tgt_data = batch['target'].to(device)\n",
    "        optimizer.zero_grad()\n",
    "        output = transformer(src_data, tgt_data[:, :-1])\n",
    "        loss = criterion(output.contiguous().view(-1, tgt_vocab_size), tgt_data[:, 1:].contiguous().view(-1))\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        train_loss+=loss.item()\n",
    "    train_losses.append(train_loss/len(train_dataloader))\n",
    "\n",
    "    # evaluate on validation set\n",
    "    transformer.eval()\n",
    "    with torch.no_grad():\n",
    "        for batch in valid_dataloader:\n",
    "            src_data = batch['input'].to(device)\n",
    "            tgt_data = batch['target'].to(device)\n",
    "            output = transformer(src_data, tgt_data[:, :-1])\n",
    "            loss = criterion(output.contiguous().view(-1, tgt_vocab_size), tgt_data[:, 1:].contiguous().view(-1))\n",
    "            valid_loss+=loss.item()\n",
    "            \n",
    "    valid_losses.append(valid_loss/len(valid_dataloader))\n",
    "\n",
    "    print(\"Epoch: [ {} / {} ]; TrainLoss: {:.5f}; ValidLoss: {:.5f}\".format(\n",
    "        epoch, epochs, train_loss/len(train_dataloader), valid_loss/len(valid_dataloader)\n",
    "    ))\n",
    "\n",
    "    # if epoch>=10:\n",
    "    #     if valid_loss < best_loss:\n",
    "    #         best_loss = valid_loss\n",
    "    #         print(\"Saving...\")\n",
    "    #         torch.save(transformer.state_dict(), \"checkpoints/transformer.pt\")\n",
    "    #         print(\"Model saved!\")\n",
    "    # early stopping\n",
    "    cond, best_loss, epochs_no_improve = early_stopping(valid_loss/len(valid_dataloader), epoch, best_loss,epochs_no_improve, patience=10)\n",
    "    if cond:\n",
    "        break\n",
    "\n",
    "print(\"Training complete!\")\n",
    "print(\"Best loss: \", best_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 413,
       "width": 547
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot losses\n",
    "import matplotlib.pyplot as plt\n",
    "%config InlineBackend.figure_format = 'retina' # Makes the plots clear on high-res screens\n",
    "plt.plot(train_losses, label='Training loss')\n",
    "plt.plot(valid_losses, label='Validation loss')\n",
    "plt.legend(frameon=False)\n",
    "plt.tight_layout\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "198"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# generate a sequence from the transformer model using test dataloader from the best model\n",
    "transformer = Transformer(src_vocab_size, tgt_vocab_size, d_model, num_heads, num_layers, d_ff, max_seq_length, dropout)\n",
    "transformer.to(device)\n",
    "transformer.load_state_dict(torch.load(\"checkpoints/transformer.pt\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.nn.functional import softmax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "transformer.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = []\n",
    "original_musics = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    for batch in test_dataloader:\n",
    "        input = batch['input'].to(device)\n",
    "        target = batch['target'].to(device)\n",
    "        for i in range(len(input)):\n",
    "            generated_tokens = []\n",
    "            output = transformer(input[i].unsqueeze(0), target[i].unsqueeze(0))\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "                generated_tokens.append(current_token)\n",
    "\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "                if len(generated_tokens) == max_length:\n",
    "                    break\n",
    "            generated_musics.append(generated_tokens)\n",
    "            original_musics.append(target[i].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(generated_musics)):\n",
    "    midi_encoder_remi.words_to_midi(generated_musics[i],f'generated_musics/transformer_2/transformer{i}.mid')\n",
    "    # midi_encoder_remi.words_to_midi(original_musics[i],f'generated_musics/original/original{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Q1\n",
      "Q2\n",
      "Q3\n",
      "Q4\n"
     ]
    }
   ],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "transformer.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = {'Q1':[], 'Q2':[], 'Q3':[], 'Q4':[]}\n",
    "\n",
    "with torch.no_grad():\n",
    "    for c in classifier_valid.keys():\n",
    "        print(c)\n",
    "        for batch in classifier_valid[c]:\n",
    "            input = batch['input'].to(device)\n",
    "            target = torch.cat((input[1:], torch.tensor([0],dtype=torch.long).to(device)))\n",
    "            generated_tokens = [input[0]]\n",
    "\n",
    "            # for i in range(len(input)):\n",
    "            generated_tokens = []\n",
    "            output = transformer(input.unsqueeze(0), target.unsqueeze(0))\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "                else:\n",
    "                    generated_tokens.append(current_token)\n",
    "            generated_musics[c].append(generated_tokens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in generated_musics.keys():\n",
    "    for i in range(len(generated_musics[c])):\n",
    "        midi_encoder_remi.words_to_midi(generated_musics[c][i],f'generated_musics/transformer_2_by_emotion/{c}/transformer{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "dir_name = \"generated_musics/transformer\"\n",
    "test = os.listdir(dir_name)\n",
    "\n",
    "for item in test:\n",
    "    if item.endswith(\".mid\"):\n",
    "        os.remove(os.path.join(dir_name, item))\n",
    "\n",
    "dir_name = \"generated_musics/original\"\n",
    "test = os.listdir(dir_name)\n",
    "\n",
    "for item in test:\n",
    "    if item.endswith(\".mid\"):\n",
    "        os.remove(os.path.join(dir_name, item))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GAN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# build a generator\n",
    "class Generator(nn.Module):\n",
    "    def __init__(self, vocab_size, max_seq_len, emsize, d_hid, nhead, nlayers, dropout):\n",
    "        super(Generator, self).__init__()\n",
    "        # TransformerModel(max_seq_len, ntokens, emsize, nhead, d_hid, nlayers, dropout).to(device)\n",
    "        self.transformer = TransformerModel(max_seq_len, vocab_size, emsize, nhead, d_hid, nlayers, dropout)\n",
    "        # self.fc = nn.Linear(dim, vocab_size)\n",
    "        \n",
    "    def forward(self, src):\n",
    "        output = self.transformer(src)\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TransformerModel_classifier(nn.Module):\n",
    "\n",
    "    def __init__(self, max_seq_length: int, ntoken: int, d_model: int, nhead: int, d_hid: int,\n",
    "                 nlayers: int, dropout: float = 0.5, num_classes=1):\n",
    "        super().__init__()\n",
    "        self.pos_encoder = PositionalEncoding(d_model, max_seq_length, dropout)\n",
    "        encoder_layers = TransformerEncoderLayer(d_model, nhead, d_hid, dropout)\n",
    "        self.transformer_encoder = TransformerEncoder(encoder_layers, nlayers)\n",
    "        self.embedding = nn.Embedding(ntoken, d_model)\n",
    "        self.d_model = d_model\n",
    "        self.linear = nn.Linear(d_model, ntoken)\n",
    "        self.fc = nn.Linear(d_model, num_classes)\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "        self.init_weights()\n",
    "\n",
    "    def init_weights(self) -> None:\n",
    "        initrange = 0.1\n",
    "        self.embedding.weight.data.uniform_(-initrange, initrange)\n",
    "        self.linear.bias.data.zero_()\n",
    "        self.linear.weight.data.uniform_(-initrange, initrange)\n",
    "\n",
    "    def forward(self, src: Tensor, src_mask: Tensor = None) -> Tensor:\n",
    "        src = self.embedding(src) * math.sqrt(self.d_model)\n",
    "        src = self.pos_encoder(src)\n",
    "        # if src_mask is None:\n",
    "        #     \"\"\"Generate a square causal mask for the sequence. The masked positions are filled with float('-inf').\n",
    "        #     Unmasked positions are filled with float(0.0).\n",
    "        #     \"\"\"\n",
    "        #     src_mask = nn.Transformer.generate_square_subsequent_mask(len(src)).to(device)\n",
    "        output = self.transformer_encoder(src)\n",
    "        output,_ = torch.max(output, dim=1)              # x has shape (Batch, Embd_dim)\n",
    "        output = self.fc(output)\n",
    "        output = self.sigmoid(output)\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# build a discriminator\n",
    "class Discriminator(nn.Module):\n",
    "    def __init__(self, vocab_size, max_seq_len, emsize, d_hid, nhead, nlayers, dropout):\n",
    "        super(Discriminator, self).__init__()\n",
    "        self.transformer = TransformerModel_classifier(max_seq_len, vocab_size, emsize, nhead, d_hid, nlayers, dropout)\n",
    "\n",
    "    def forward(self, src):\n",
    "        output = self.transformer(src)\n",
    "        return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the generator\n",
    "\n",
    "netG = Generator(vocab_size, max_seq_len, emsize, d_hid, nhead, nlayers, dropout).to(device)\n",
    "# Create the Discriminator\n",
    "netD = Discriminator(vocab_size, max_seq_len, emsize, d_hid, nhead, nlayers, dropout).to(device)\n",
    "\n",
    "# Initialize the ``BCELoss`` function\n",
    "criterion_BCE = nn.BCELoss()\n",
    "\n",
    "criterion_ce = nn.CrossEntropyLoss()\n",
    "\n",
    "# Create batch of latent vectors that we will use to visualize\n",
    "#  the progression of the generator\n",
    "nz = 100\n",
    "fixed_noise = torch.randn(64, nz, 1, 1, device=device)\n",
    "\n",
    "# Establish convention for real and fake labels during training\n",
    "real_label = 1.\n",
    "fake_label = 0.\n",
    "\n",
    "lr = 0.0002\n",
    "beta1 = 0.5\n",
    "\n",
    "# Setup Adam optimizers for both G and D\n",
    "optimizerD = torch.optim.Adam(netD.parameters(), lr=lr, betas=(beta1, 0.999))\n",
    "optimizerG = torch.optim.Adam(netG.parameters(), lr=lr, betas=(beta1, 0.999))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_epochs = 5\n",
    "temperature = 1.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting Training Loop...\n",
      "[0/5][63/54]\tLoss_D: 1.5515\tLoss_G: 5.6418\tD(x): 0.7517\tD(G(z)): 0.7159 / 0.7159\n",
      "[0/5][63/54]\tLoss_D: 0.0364\tLoss_G: 3.1777\tD(x): 0.9807\tD(G(z)): 0.0168 / 0.0168\n",
      "[1/5][63/54]\tLoss_D: 0.0229\tLoss_G: 3.1077\tD(x): 0.9906\tD(G(z)): 0.0134 / 0.0134\n",
      "[1/5][63/54]\tLoss_D: 0.0824\tLoss_G: 2.9925\tD(x): 0.9519\tD(G(z)): 0.0326 / 0.0326\n",
      "[2/5][63/54]\tLoss_D: 0.0695\tLoss_G: 2.9593\tD(x): 0.9766\tD(G(z)): 0.0447 / 0.0447\n",
      "[2/5][63/54]\tLoss_D: 0.0316\tLoss_G: 2.9662\tD(x): 0.9797\tD(G(z)): 0.0110 / 0.0110\n",
      "[3/5][63/54]\tLoss_D: 0.0208\tLoss_G: 2.9278\tD(x): 0.9939\tD(G(z)): 0.0146 / 0.0146\n",
      "[3/5][63/54]\tLoss_D: 0.0103\tLoss_G: 2.9537\tD(x): 0.9937\tD(G(z)): 0.0040 / 0.0040\n",
      "[4/5][63/54]\tLoss_D: 0.0084\tLoss_G: 2.9120\tD(x): 0.9977\tD(G(z)): 0.0060 / 0.0060\n",
      "[4/5][63/54]\tLoss_D: 0.0059\tLoss_G: 2.9447\tD(x): 0.9966\tD(G(z)): 0.0025 / 0.0025\n"
     ]
    }
   ],
   "source": [
    "# Training Loop\n",
    "# Lists to keep track of progress\n",
    "txt_list = []\n",
    "G_losses = []\n",
    "D_losses = []\n",
    "iters = 0\n",
    "\n",
    "\n",
    "print(\"Starting Training Loop...\")\n",
    "# For each epoch\n",
    "for epoch in range(num_epochs):\n",
    "    # For each batch in the dataloader\n",
    "    for k, data in enumerate(train_dataloader):\n",
    "        input = data['input']\n",
    "        target = data['target'].to(device)\n",
    "        ############################\n",
    "        # (1) Update D network: maximize log(D(x)) + log(1 - D(G(z)))\n",
    "        ###########################\n",
    "        ## Train with all-real batch\n",
    "        netD.zero_grad()\n",
    "        # Format batch\n",
    "        real_cpu = input.to(device)\n",
    "        b_size = real_cpu.size(0)\n",
    "        label = torch.full((b_size,), real_label, dtype=torch.float, device=device)\n",
    "        # Forward pass real batch through D\n",
    "        output = netD(real_cpu).view(-1)\n",
    "        # Calculate loss on all-real batch\n",
    "        errD_real = criterion_BCE(output, label)\n",
    "        # Calculate gradients for D in backward pass\n",
    "        errD_real.backward()\n",
    "        D_x = output.mean().item()\n",
    "\n",
    "        ## Train with all-fake batch\n",
    "        # Generate batch of latent vectors\n",
    "        # noise = torch.randn(b_size, nz, 1, 1, device=device)\n",
    "        # Generate fake image batch with G\n",
    "        fake = netG(real_cpu)\n",
    "        fake_probabilities = softmax(fake.squeeze() / temperature, dim=-1)\n",
    "        fake_mat = torch.zeros((b_size, max_seq_length), dtype=torch.long).to(device)\n",
    "        for i in range(b_size):\n",
    "            for j in range(max_seq_length):\n",
    "                fake_mat[i][j] = torch.multinomial(fake_probabilities[i][j], 1).item()\n",
    "        \n",
    "        label.fill_(fake_label)\n",
    "        # Classify all fake batch with D\n",
    "        output = netD(fake_mat).view(-1)\n",
    "        # Calculate D's loss on the all-fake batch\n",
    "        errD_fake = criterion_BCE(output, label)\n",
    "        # Calculate the gradients for this batch, accumulated (summed) with previous gradients\n",
    "        errD_fake.backward()\n",
    "        D_G_z1 = output.mean().item()\n",
    "        # Compute error of D as sum over the fake and the real batches\n",
    "        errD = errD_real + errD_fake\n",
    "        # Update D\n",
    "        optimizerD.step()\n",
    "\n",
    "        ############################\n",
    "        # (2) Update G network: maximize log(D(G(z)))\n",
    "        ###########################\n",
    "        netG.zero_grad()\n",
    "        label.fill_(real_label)  # fake labels are real for generator cost\n",
    "        # Since we just updated D, perform another forward pass of all-fake batch through D\n",
    "        # output = netD(fake_mat).view(-1)\n",
    "        # Calculate G's loss based on this output\n",
    "        # errG = criterion(output, label)\n",
    "        errG = criterion_ce(fake.view(-1, vocab_size), target.view(-1))\n",
    "        # Calculate gradients for G\n",
    "        errG.backward()\n",
    "        D_G_z2 = output.mean().item()\n",
    "        # Update G\n",
    "        optimizerG.step()\n",
    "\n",
    "        # Output training stats\n",
    "        if k % 50 == 0:\n",
    "            print('[%d/%d][%d/%d]\\tLoss_D: %.4f\\tLoss_G: %.4f\\tD(x): %.4f\\tD(G(z)): %.4f / %.4f'\n",
    "                  % (epoch, num_epochs, i, len(train_dataloader),\n",
    "                     errD.item(), errG.item(), D_x, D_G_z1, D_G_z2))\n",
    "\n",
    "        # Save Losses for plotting later\n",
    "        G_losses.append(errG.item())\n",
    "        D_losses.append(errD.item())\n",
    "\n",
    "        # Check how the generator is doing by saving G's output on fixed_noise\n",
    "        if (iters % 500 == 0) or ((epoch == num_epochs-1) and (i == len(train_dataloader)-1)):\n",
    "            with torch.no_grad():\n",
    "                fake = netG(real_cpu).detach().cpu()\n",
    "            fake_probabilities = softmax(fake.squeeze() / temperature, dim=-1)\n",
    "            fake_mat = torch.zeros((b_size, max_seq_length), dtype=torch.long).to(device)\n",
    "            for i in range(b_size):\n",
    "                for j in range(max_seq_length):\n",
    "                    fake_mat[i][j] = torch.multinomial(fake_probabilities[i][j], 1).item()\n",
    "            txt_list.append(fake_mat)\n",
    "\n",
    "        iters += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABC0AAAM6CAYAAACl+LmmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy89olMNAAAACXBIWXMAAB7CAAAewgFu0HU+AADzE0lEQVR4nOzdd3zTdf4H8FdG0703bSmlbIrsKYgyVFDcGwd66rnP8bvz9E5F8dS783DhXuBWBEUQRYbsvaEIpbSle4+kK2nG74+QkO/Ity1N27R9PR8PHo/mm2+SD+23ab7v73uobDabDUREREREREREXkbd2QsgIiIiIiIiIpLDoAUREREREREReSUGLYiIiIiIiIjIKzFoQUREREREREReiUELIiIiIiIiIvJKDFoQERERERERkVdi0IKIiIiIiIiIvBKDFkRERERERETklRi0ICIiIiIiIiKvxKAFEREREREREXklBi2IiIiIiIiIyCsxaEFEREREREREXolBCyIiIiIiIiLySgxaEBEREREREZFXYtCCiIiIiIiIiLwSgxZERERERERE5JW0nb2A9tTY2IgjR44AAKKjo6HVduv/LhEREREREVGnMJvNKCsrAwAMGzYMfn5+Hnnebn0Wf+TIEYwbN66zl0FERERERETUY+zevRtjx471yHOxPISIiIiIiIiIvFK3zrSIjo52fr17927Ex8d34mqIiIiIiIiIuqeioiJnpYPruXhbdeughWsPi/j4eCQmJnbiaoiIiIiIiIi6P0/2k2R5CBERERERERF5JQYtiIiIiIiIiMgrMWhBRERERERERF6JQQsiIiIiIiIi8koMWhARERERERGRV2LQgoiIiIiIiIi8EoMWREREREREROSVGLQgIiIiIiIiIq/EoAUREREREREReSUGLYiIiIiIiIjIKzFoQUREREREREReiUELIiIiIiIiIvJKDFoQERERERERkVdi0IKIiIiIiIiIvBKDFkRERERERETklRi0ICIiIiIiIiKvxKAFEREREREREXklBi2IiIiIiIiIyCsxaEFEREREREREXolBCyIiIiIiIiLySgxaEBEREREREZFXYtCCiIiIiIiIiLwSgxZERERERERE5JUYtCAiIiIiIiKvlJOTA5VKBZVKhcWLF3f2cqgTaDt7AUREREREROTdzGYzVqxYgTVr1mDHjh0oKSlBVVUVAgICEB0djWHDhmHSpEm47rrrkJKS0tnLpW6EQQsiIiIiIiJy66effsITTzyBzMxMyX16vR56vR6nTp3Cjz/+iL/97W+47LLL8MorryAtLa0TVkvdDYMWREREREREJOvFF1/Es88+C5vNBgC48MILcfnll+O8885DZGQk6uvrUVRUhM2bN2PVqlXIycnBzz//jMTERLz33nudvHrqDhi0ICIiIiIiIolPPvkEzzzzDAAgNjYW33zzDS688ELZfa+//nq8/vrr+Oabb/D000934Cqpu2PQwsvYbDboG8woq21Eqd6ItMRQhPj5dPayiIiIiIioB8nLy8ODDz4IAAgJCcHWrVvRr18/xcdoNBrMnTsXl112GbZs2dIRy6QegNNDvMyYF9dh+Au/YcbCzbjlo104XmTo7CUREREREVEPs3DhQjQ2NgIA/vWvfzUbsHAVFhaGOXPmtNfSZJlMJrzzzju46KKLEB0dDZ1Oh7i4OMyePRtffPEFrFar4uMzMjLw8MMPIy0tDcHBwdDpdOjVqxdGjBiBu+66C99++y2MRqPkcRaLBYsXL8Yll1yCuLg46HQ6hIaGon///pg+fTpeeuklHDt2TPG1f/zxR1x//fXo3bs3/Pz8EBYWhjFjxuD5559HVVVVu6y7K1HZHMVJ3VB+fj6SkpIA2COFiYmJnbyi5l306kZkl9c5by+6ZSQuP69XJ66IiIiIiIh6EpvNhujoaFRUVCA4OBhFRUUIDAzslLXk5OQ4p5F8+umnmDdvnuw+s2bNwvHjx90+z+TJk7FixQpERERI7lu6dCluvfVWmEwmxbUcOXJE0Fy0trYWs2fPbjar5Nprr8X3338v2V5VVYXrrrsOGzZscPvYmJgYrFixAhMmTPDYuttLe51/szzEy0QH+QqCFmWGrh0VIyIiIiKiruXo0aOoqKgAAEyZMqXTAhYtUVtbi+nTpyMrKwsAcNVVV+Guu+5Cr169kJ2djUWLFmHTpk3YunUr5syZg82bN0Oj0TgfX1JSgjvvvBMmkwkxMTF46KGHMGHCBERFRaGhoQGZmZnYtGkTfvzxR8lrz58/3xmwuPzyyzF37lxntkRpaSkOHDiAVatWQaVSSR5rNBoxY8YM7N+/HxqNBrfccgtmz56NlJQUNDU1YfPmzVi4cCFKS0sxe/ZsHDhwAMnJyR5Zd1fDoIWXiQ72Fdxm0IKIiIiIqGWsVhuq6pWvOnc34QE6qNXSk+K2OHz4sPPrUaNGefS5Pe355593Biz++c9/YsGCBc77Ro8ejWuvvRa33XYbvvzyS2zfvh0ffPAB7r//fuc+P//8M+rq7BeN169fL8lImDRpEm6//XYsWrRI8trfffcdAOC6667D0qVLJfdfeumleOqpp1BZWSm574UXXsD+/fsRFhaGdevWYfTo0YL7J0+ejLlz52LixIkoKirC008/jS+//NIj6+5qGLTwMgxaEBERERGdm6p6E0a/uK6zl9Gh9v1zBiKDfJvfsRXKy8udX0dHR7vdz2q1KvZrGDhwIHx82m+ogNFoxEcffQQAGDp0KObPny/ZR6VS4Z133sGvv/6KiooKLFq0SBC0KC4uBgCEh4crllD4+/tLtjkeO2XKFMV1iktSamtr8fbbbwMAFixYIAlYOCQnJ+OZZ57BAw88gKVLl+KDDz5wZr20Zd1dDRtxehlJ0KKWQQsiIiIiIuo4BsPZYQBKpSF6vR7Dhg1z+6+goKBd17lv3z5UV1cDAObNmyco+3AVEhKCG264AQBw7NgxFBUVOe+Lj48HYO8vsWLFila9vuOx3377Lerr61v8uE2bNqGmpgaAPUtDyQUXXAAAaGpqwr59+zyy7q6GQQsvIw5alOoZtCAiIiIioo4THBzs/NpRguCNjh496vx6/Pjxivu63u/6uCuuuAJhYWEAgKuvvhrTpk3Da6+9hn379sFisSg+5x133AEA2L59O1JSUvDQQw/hhx9+QFlZmeLj9u7d6/w6Pj4eKpXK7T/XLApHdkVb193VMGjhZZhpQUREREREnSkyMtL5tdIJeFhYGGw2m+Cf40S+I7j2ioiJiVHcNy4uTvZxkZGR+Omnn5CQkACbzYbff/8djz/+OMaMGYOIiAhcc801WLVqlexzPvPMM7jrrrugUqlQWlqKt99+G9dccw1iYmKQlpaG5557DiUlJZLHlZaWtva/CgCCbI62rLurYU8LLxMtqkerqDXCYrVB4+HmOkRERERE3U14gA77/jmjs5fRocIDdB5/zuHDhzu/PnDggMefvz3ITehoqSlTpiAzMxPLli3D6tWrsXnzZuTn50Ov1+OHH37ADz/8gEsuuQTLly9HQECA83E+Pj74+OOP8cQTT+Drr7/Ghg0bsHfvXphMJqSnpyM9PR0LFy7EF198gSuvvNL5ONdMiP3797e474d4hOi5rrurYdDCy8SECIMWVhtQWWeSZGAQEREREZGQWq3yeFPKnigtLQ2RkZGoqKjAli1bUF9f75Unva4NLktKSjBgwAC3+7qWVogbYwKAn58f5s6di7lz5wIAsrOz8fPPP+Ott95CRkYG1qxZg3/84x947bXXJI8dMmQIFixYgAULFqCxsRFbt27FV199hc8++wy1tbW4+eabcerUKWcfCtdMlujoaEkwojXasu6uguUhXiYy0BfipApOECEiIiIioo6iUqlw6623ArA321yyZEknr0iea7+HXbt2Ke67e/du2ce54+hRsWfPHmdQwTHiVImfnx9mzJiBTz75BP/9738BAA0NDYJSjZEjRzq/3rZtW7PP2Rrnum5vxqCFl9GoVYgIFDXjNDR20mqIiIiIiKgnevzxx+Hn5wcAeOqpp5Cdnd3JK5IaPXq0sxnlkiVLYLVaZfczGAzOE/chQ4Y4Mx5aIiQkBGPHjgUgHAXbEtOnT3d+7frYGTNmODNX3nzzTdhstlY9b0u0Zd3ehkELLyRpxslMCyIiIiIi6kC9e/fGm2++CQCoqanB5MmTsXXrVsXH2Gw25wjSjuDr64u7774bgH0iyIIFC2TX9NBDDzlP3B966CHB/WvWrBGMQBWrqalxZmmkpKQ4t1dWVmLlypWKAYfffvvN+bXrY8PCwpzr2L59Ox577DG3ARfAXvry0UcfeWTdXRF7Wnih6GBf/OFy/HGCCBERERERdbR77rkHBQUFeP7551FYWIgpU6Zg2rRpmDNnDoYNG4aIiAhYLBYUFxdj//79+O6775Ceng4A0Gg00Ok83yRU7Nlnn8Xy5cuRlZWF+fPn48iRI7jzzjsRHx+P7OxsLFq0CBs3bgQATJw4Effee6/g8V9//TXmzJmDmTNn4uKLL0ZaWhoiIiJgMBhw9OhRLFq0CAUFBQCA++67z/k4vV6PK664An369ME111yD8ePHIzk5GVqtFkVFRVi5cqUz0JCQkIDLL79c8LovvPACNm3ahF27duGNN97Axo0bcc8992DEiBEIDAxEVVUV0tPTsW7dOvzyyy8YNmyYM0DTlnV3RQxaeKEYZloQEREREZEXmD9/PoYPH47/+7//Q1ZWFjZs2IANGza43V+lUuGSSy7Bf//7X/Tq1avd1xccHIz169dj1qxZOH78OJYtW4Zly5ZJ9jv//PPx008/QaPRSO5ramrC6tWrsXr1arevc9999+GRRx6RbM/JycHChQvdPi4+Ph4rVqxAUFCQYLuvry/Wrl2LefPmYfny5Th06JAkC8RVSEiIR9fdlTBo4YVYHkJERERERN7i6quvxpw5c/DDDz9gzZo12LFjB0pLS1FdXY2AgABERkZi2LBhmDhxIm688cYOL0fo06cPDh06hA8//BBLly7F0aNHodfrERERgZEjR2Lu3Lm45ZZboFZLuyO89tprmDlzJjZs2IDDhw+jqKgIZWVl0Gg0SEpKwsSJE3H33Xdj8uTJgsclJydj9+7dWL16NbZv347Tp0+jpKQEtbW1CAsLw5AhQzBnzhzce++9sgEHwB5wWbZsGbZu3YolS5Zgy5YtKCwsRENDA0JCQpCamopx48bhsssuw8UXX+yRdXdFKlt7dP3wEvn5+UhKSgIA5OXltWmUTEf6ZGs2Xlh1zHl7XEoEvvvzxE5cEREREREREZF77XX+zUacXkicaVHOTAsiIiIiIiLqgRi08EIsDyEiIiIiIiJi0MIriRtxGoxmNJgsnbQaIiIiIiIios7BRpxeSJxpAQDltUYkRQR0wmqIiIiIiIjOTUZGBkwmU6sfFxMTg5iYmHZYEXU1DFp4oSBfLfx81Ghssjq3lRoYtCAiIiIioq7l4osvxunTp1v9uOeeew7z58/3/IKoy2F5iBdSqVQyfS0aO2k1RERERERERJ2DmRZeKjrIF3mVDc7bbMZJRERERERdTU5OTmcvgbo4Zlp4qZhgP8FtBi2IiIiIiIiop2HQwktJykNqGbQgIiIiIiKinoVBCy8l7WnBoAURERERERH1LAxaeClx0KKUQQsiIiIiIiLqYRi08FIxzLQgIiIiIiKiHo5BCy8lzrQorzXCarV10mqIiIiIiIiIOh6DFl5KHLRosthQ09DUSashIiIiIiIi6ngMWnipyEBfyTZOECEiIiIiIqKehEELL6XTqhEe4CPYVqpn0IKIiIiIiIh6DgYtvFhMsJ/gdlltYyethIiIiIiIiKjjMWjhxcR9LThBhIiIiIiIiHoSBi28GIMWRERERERE1JMxaOHFGLQgIiIiIiKinoxBCy8WIwpaFFazpwUREREREfUsOTk5UKlUUKlUWLx4cWcvhzqYtrMXQO4lRQQIbudU1HXSSoiIiIiIqKfZuHEjLrroIsl2jUaDkJAQhIaGIikpCaNHj8bkyZMxZ84c6HS6TlgpdWfMtPBiKVGBgtulBiNqjeZOWg0RERERERFgsVhQVVWFnJwcbNmyBa+//jquu+46JCYm4sUXX4TZzHMW8hxmWnix3hEBUKkAm+3stpzyOqQlhHbeooiIiIiIqMe5//778cADDzhv19bWoqqqCocPH8b69euxbt06lJWV4ZlnnsHKlSuxatUqREdHd+KKqbtg0MKL+flo0CvUHwXVDc5tORUMWhARERERUceKiYlBWlqaZPusWbPw5JNP4tixY7j11ltx4MAB7N69G1dffTU2bNjAchFqM5aHeDlxiUhOOftaEBERERGRdxkyZAi2bduGkSNHAgC2bduGt99+u5NXRd0BgxZerk+UsBlndnl9J62EiIiIiIjIPX9/f3z++edQqVQAgFdffRVNTU0d9vomkwnvvPMOLrroIkRHR0On0yEuLg6zZ8/GF198AavVqvj4jIwMPPzww0hLS0NwcDB0Oh169eqFESNG4K677sK3334Lo9EoeZzFYsHixYtxySWXIC4uDjqdDqGhoejfvz+mT5+Ol156CceOHVN87R9//BHXX389evfuDT8/P4SFhWHMmDF4/vnnUVVV1S7r7ipYHuLl+kQKMy2yy2s7aSVERERERETKhg4dipkzZ+K3335DYWEh9uzZg0mTJrX76+bk5GDWrFk4fvy4YHtJSQl++eUX/PLLL3j//fexYsUKRERESB6/dOlS3HrrrTCZTILtRUVFKCoqwqFDh/Dpp5/iyJEjgjKZ2tpazJ49G1u2bBE8rqmpCXq9HpmZmdiwYQP279+P77//XvK6VVVVuO6667BhwwbBdqPRiH379mHfvn145513sGLFCkyYMMFj6+5KGLTwcpLykApmWhARERERkfeaMWMGfvvtNwDAli1b2j1oUVtbi+nTpyMrKwsAcNVVV+Guu+5Cr169kJ2djUWLFmHTpk3YunUr5syZg82bN0Oj0TgfX1JSgjvvvBMmkwkxMTF46KGHMGHCBERFRaGhoQGZmZnYtGkTfvzxR8lrz58/3xmwuPzyyzF37lxntkRpaSkOHDiAVatWObNPXBmNRsyYMQP79++HRqPBLbfcgtmzZyMlJQVNTU3YvHkzFi5ciNLSUsyePRsHDhxAcnKyR9bdlTBo4eX6iIIWlXUm1DQ0IdTfp5NWRERERETkpaxWoKGys1fRsfwjALV3Vf2PGjXK+XVGRka7v97zzz/vDFj885//xIIFC5z3jR49Gtdeey1uu+02fPnll9i+fTs++OAD3H///c59fv75Z9TV2XsHrl+/XpKRMGnSJNx+++1YtGiR5LW/++47AMB1112HpUuXSu6/9NJL8dRTT6GyUnpcvvDCC9i/fz/CwsKwbt06jB49WnD/5MmTMXfuXEycOBFFRUV4+umn8eWXX3pk3V0JgxZeLik8ABq1Chbr2bmnOeV1GJ4U1nmLIiIiIiLyRg2VwH9TO3sVHeuvp4DAqM5ehUBkZKTz6+b6MbSV0WjERx99BMBemjJ//nzJPiqVCu+88w5+/fVXVFRUYNGiRYKgRXFxMQAgPDxcsYTC399fss3x2ClTpiiuU1ySUltb62xUumDBAknAwiE5ORnPPPMMHnjgASxduhQffPABAgMD27zursS7QnIkodOqkRguPMhyKjhBhIiIiIiIvFNQUJDza4PB0K6vtW/fPlRXVwMA5s2bJyj7cBUSEoIbbrgBAHDs2DEUFRU574uPjwdgD7CsWLGiVa/veOy3336L+vqWl/Jv2rQJNTU1AOxZGkouuOACAPY+Gfv27fPIursSBi26AGkzTgYtiIiIiIjIO7kGKkJCQtr1tY4ePer8evz48Yr7ut7v+rgrrrgCYWFhAICrr74a06ZNw2uvvYZ9+/bBYrEoPucdd9wBANi+fTtSUlLw0EMP4YcffkBZWZni4/bu3ev8Oj4+HiqVyu0/1ywKR3ZFW9fdlTBo0QWIm3EyaEFERERERN6qvLzc+bXcpA5Pcu0VERMTo7hvXFyc7OMiIyPx008/ISEhATabDb///jsef/xxjBkzBhEREbjmmmuwatUq2ed85plncNddd0GlUqG0tBRvv/02rrnmGsTExCAtLQ3PPfccSkpKJI8rLS1t7X8VAATZHG1Zd1fCnhZdQJ/IAMHtHAYtiIiIiIik/CPsPR56Ev/2DQqciwMHDji/HjhwYIe9rtyEjpaaMmUKMjMzsWzZMqxevRqbN29Gfn4+9Ho9fvjhB/zwww+45JJLsHz5cgQEnD0/8/Hxwccff4wnnngCX3/9NTZs2IC9e/fCZDIhPT0d6enpWLhwIb744gtceeWVzse5ZkLs378fPj4tG7SQmJjokXV3JQxadAHiCSLZ5XWw2Wxt+qUkIiIiIup21Gqva0rZE61du9b59eTJk9v1tVwzOUpKSjBgwAC3+7qWVshlgPj5+WHu3LmYO3cuACA7Oxs///wz3nrrLWRkZGDNmjX4xz/+gddee03y2CFDhmDBggVYsGABGhsbsXXrVnz11Vf47LPPUFtbi5tvvhmnTp1y9qFwbVYaHR0tCUa0RlvW3RWwPKQLEJeH6BvNqKpv6qTVEBERERERyTt69CjWr18PAEhKSsKYMWPa9fVc+z3s2rVLcd/du3fLPs4dR4+KPXv2OIMKjhGnSvz8/DBjxgx88skn+O9//wsAaGhoEJRqjBw50vn1tm3bmn3O1jjXdXsrBi26gIQwf/hohFkV7GtBRERERETepKGhAbfffjtsNhsA4P/+7/+g1bZvcv/o0aOdzSiXLFkCq9Uqu5/BYHCeuA8ZMsSZ8dASISEhGDt2LABhv46WmD59uvNr18fOmDHDWa7x5ptvOr9nntSWdXsTBi26AK1GjaQIYf0RgxZEREREROQtjh07hsmTJzv7WUydOhX3339/u7+ur68v7r77bgD2LI8FCxZI9rHZbHjooYecJ+4PPfSQ4P41a9YIRqCK1dTUOLM0UlJSnNsrKyuxcuVKxYDDb7/95vza9bFhYWHOdWzfvh2PPfaY24ALYC99+eijjzyy7q6GPS26iJTIQGSVnQ1UsBknERERERF1lNLSUsGY0Lq6OlRVVeHw4cNYv3491q5d6zx5nzBhAr7//vsWN5dsq2effRbLly9HVlYW5s+fjyNHjuDOO+9EfHw8srOzsWjRImzcuBEAMHHiRNx7772Cx3/99deYM2cOZs6ciYsvvhhpaWmIiIiAwWDA0aNHsWjRIhQUFAAA7rvvPufj9Ho9rrjiCvTp0wfXXHMNxo8fj+TkZGi1WhQVFWHlypXOQENCQgIuv/xyweu+8MIL2LRpE3bt2oU33ngDGzduxD333IMRI0YgMDAQVVVVSE9Px7p16/DLL79g2LBhzgBNW9bd1TBo0UVImnFWMGhBREREREQd491338W7776ruE90dDQeffRR/O1vf2v3shBXwcHBWL9+PWbNmoXjx49j2bJlWLZsmWS/888/Hz/99BM0Go3kvqamJqxevRqrV692+zr33XcfHnnkEcn2nJwcLFy40O3j4uPjsWLFCgQFBQm2+/r6Yu3atZg3bx6WL1+OQ4cOSbJAXIWEhHh03V0FgxZdhDhowUwLIiIiIiLqDGq1GsHBwQgNDUVycjJGjx6NKVOm4PLLL4dOp+uUNfXp0weHDh3Chx9+iKVLl+Lo0aPQ6/WIiIjAyJEjMXfuXNxyyy1Qq6UdEl577TXMnDkTGzZswOHDh1FUVISysjJoNBokJSVh4sSJuPvuuyWTUJKTk7F7926sXr0a27dvx+nTp1FSUoLa2lqEhYVhyJAhmDNnDu69917ZgANgD7gsW7YMW7duxZIlS7BlyxYUFhaioaEBISEhSE1Nxbhx43DZZZfh4osv9si6uxqVrT06fniJ/Px8JCUlAQDy8vLaNEams209WY5bPz7bDTdQp8HR5y/h2FMiIiIiIiLqdO11/s1GnF1EnyhhI846kwUVdaZOWg0RERERERFR+2PQoouIC/GDWpRUUVzT2DmLISIiIiIiIuoADFp0EVqNGlFBvoJtpQYGLYiIiIiIiKj7YiPOLiQu1A+lBqPzdnGNUWFvIiIiIiKizpeRkQGTqfWl7TExMYiJiWmHFVFXwqBFFxIT7Aegxnm7WM9MCyIiIiIi8m4XX3wxTp8+3erHPffcc5g/f77nF0RdCstDupC4UGF5SAl7WhAREREREVE3xkyLLiQuxE9wm5kWRERERETk7XJycjp7CdSFMdOiC4kVBS1KGLQgIiIiIiKiboxBiy4kLpRBCyIiIiIiIuo52jVooVKpWvTvwgsvbM9ldBvi8pCq+iY0Nlk6aTVERERERERE7YuZFl1IjChoAQCleo49JSIiIiIiou6pQxpx3n///XjggQfc3h8YGNgRy+jyQvy08PfRoMElu6JY34jekQGduCoiIiIiIiKi9tEhQYuYmBikpaV1xEt1ayqVCnGhfsgur3Nu4wQRIiIiIiIi6q5YHtLFxIb4Cm6XMmhBRERERERE3RSDFl2MuBlncQ2DFkRERERERNQ9MWjRxcSKgxbMtCAiIiIiIqJuqkN6WixduhTfffcdcnJyoNFoEBcXh0mTJmHevHm46KKLzvl58/PzFe8vKio65+f2VuKgRQmDFkRERERERNRNdUjQ4tixY4LbmZmZyMzMxGeffYarrroKixcvRmhoaKufNykpyVNL7DLiQplpQURERERERD1DuwYtAgICcMUVV2D69OkYNGgQgoKCUFZWhk2bNuG9995DRUUFfvzxR1x55ZVYu3YtfHx82nM53YI008IIm80GlUrVSSsiIiIiIiIiah/tGrQoKChAWFiYZPvMmTPx8MMPY9asWThw4AA2bdqEd999F4888kirnj8vL0/x/qKiIowbN65Vz+ntxJkWJrMV1fVNCA/UddKKiIiIiIiIiNpHuwYt5AIWDrGxsfj+++8xaNAgNDU14a233mp10CIxMbGNK+x6YoJ9JduK9Y0MWhAREREREVG306nTQ/r27YuZM2cCsPe5KCws7MzldAk+GjWigoQBCva1ICIiIiIiou6o00eeDhkyxPl1QUFBJ66k65D0tahh0IKIiIiIiIi6n04PWrCBZOvFhXCCCBEREREREXV/nR60cB2H2qtXr05cSdcRGyqdIEJERERERETU3XRq0CI7Oxtr164FAKSmpiIhIaEzl9NliDMtSphpQURERERERN1QuwUtVq5cCbPZ7Pb+kpISXHvttTCZTACABx54oL2W0u1IykPY04KIiIiIiIi6oXYbefrwww+jqakJ1157LSZOnIg+ffrA398f5eXl2LhxI95//32Ul5cDACZPnowHH3ywvZbS7cSECMeeMtOCiIiIiIiIuqN2C1oAQGFhId566y289dZbbve59tpr8dFHH8HX19ftPiQUJ+ppUVFngtFsga9W00krIiIiIiIiIvK8dgtaLFmyBJs2bcKOHTuQlZWF8vJy6PV6BAUFISkpCZMmTcIdd9yBiRMnttcSui1xeQgAlBmMSAwP6ITVEBEREREREbWPdgtaTJ06FVOnTm2vp+/RQv194KtVw2i2OreV6BsZtCAiIiIiIqJupdNHnlLrqVQqSYlIcY1w7GluRT2yymo7cllEREREREREHsWgRRcVKyoRKaiud379/qZTmPrq75j2v014afUfHb00IiIiIiIiIo9g0KKLSokMFNzOLLVnVTQ2WfDG+pOw2ezbP96ajco6U0cvj4iIiIiIiKjNGLToovrFBAluO4IWWWV1qDdZnNstVhvLRIiIiIiIiKhLYtCii5ILWthsNmSUGCT75lc1dNSyiIiIiIiIiDyGQYsuShy00DeaUVZrxAmZoEVeZb1kGxEREREREZG3Y9Cii0oI84efj/DHl1lai4xiZloQERERERFR98CgRRelVqvQN0paIiKXaZFfzUwLIiIiIiIi6noYtOjC+scKgxaH8mpksyqYaUFERERERERdEYMWXVi/aGHQYu2xYtn9CqsbYLHaOmJJRERERERERB7DoEUXJteMU06TxYZSQ2NHLImIiIiIiIjIYxi06MLEQQsleZUsESEiIiIiIqKuhUGLLiw5MhAatapF++ZXsRknERERERERdS0MWnRhOq0ayZEBLdqXzTiJiIiIiIioq2HQoovr38ISEWZaEBERERERUVfDoEUX19K+Fsy0ICIiIiIioq6GQYsurqVBizxmWhAREREREVEXw6BFF9cvOlh2e9+oQMHtoupGmC3WjlgSERERERERkUcwaNHFpcYEym6fNihGcNtstaHEYOyIJRERERERERF5BIMWXVyATouEMH/J9gl9I+HvoxFsy69kiQgRERERERF1HQxadANyfS0GxgUjKUIYzGAzTiIiIiIiIupKGLToBsRBi0CdBglh/kgMDxBsZzNOIiIiIiIi6koYtOgGBsYJm3EOig+BWq1CYjgzLYiIiIiIiKjrYtCiG7hsWDziQ/0AACoVcNf5KQAgE7RgpgURERERERF1HdrOXgC1XaCvFr8+egE2ZZShb1Qg0hJCAUBSHsJMCyIiIiIiIupKGLToJkL9fXDF8F6CbeJMi6KaRpgtVmg1TLAhIiIiIiIi78ez125MnGlhsdpQVNPYSashIiIiIiIiah0GLbqx8AAfBOg0gm0sESEiIiIiIqKugkGLbkylUiFJPPa0ks04iYiIiIiIqGtg0KKbS4oQ9rX4ek8ubDZbJ62GiIiIiIiIqOUYtOjmpg+OFdw+kFuNNenFnbQaIiIiIiIiopZj0KKbu250IvpECktE/vPrCTRZrJJ9G5ssOJhXjcYmS0ctj4iIiIiIiMgtBi26OR+NGv93yUDBtqzyOny7J0+wLa+yHue/sgFXvb0NMxZuQpnB2JHLJCIiIiIiIpJg0KIHuGxYPIYnhgq2vb7uJOqMZuftD7dkoaLOBMA+YeT1dRkdukYiIiIiIiIiMQYtegCVSoW/zxos2FZea8QnW7Odtw/lVQvu/+FAAQyNTR2xPCIiIiIiIiJZDFr0EBNTI3HhwGjBth8OFgAALFYbTpQYBPfVmyz44UBBh62PiIiIiIiISIxBix7koYv6CW5nldXB0NiEnIo6NDZJG3N+sfO07HjUbZnleOWX49icUdZuayUiIiIiIiJi0KIHSUsIhVatEmxLL9TjeJFBdv+Mklrszq4UbNuVVYG5H+3Ce5tO4fZPdmPryfJ2Wy8RERERERH1bAxa9CB+PhoMiA0WbDtaUIPjxXq3j/liV67g9i9HiwW3VxxkCQkRERERERG1DwYtepjzRFNEjhTU4A83mRYA8OvRIpQaGp23S/SNgvtzK+s9u0AiIiIiIiKiMxi06GHSEqRBC6VMiyaLDd/tyXPeLq81Cu7Pr2rw7AKJiIiIiIiIzmDQoocZJgpaZJXVSQIP4myMlYeKnF+XGYRBi6KaBjRZpE08iYiIiIiIiNqKQYseZmBcsKQZpyutWoXHZw4QbDtdWeecIlJeaxLcZ7UBhdXMtiAiIiIiIiLPY9Cih/Hz0WBgXLDb+1Ojg5AaHSTY1thkhb7RjMYmC2qNZslj8ioZtCAiIiIiIiLPY9CiBxKXiLgaFB+M6GBfyfYyQ6OkNMQhr4rNOImIiIiIiMjzGLTogcTNOF0NiguBn48Gof4+gu0leqOkCadDHieIEBERERERUTtg0KIHai7TAgBiRNkWpYZGST8LhzxOECEiIiIiIqJ2wKBFDzQoPhg+GvlmnIPjQgAAsSF+gu3MtCAiIiIiIqKOxqBFD+Sr1WBArLQZZ1iAD2JD7BkWkkwLvdFtT4t89rQgIiIiIiKidsCgRQ8lVyIyKC4YKpU9AyNGnGlhaHSbaVFea0K9STpVhIiIiIiIiKgtGLTooeSacQ6OD3F+Lc60KFMoDwGAfPa1ICIiIiIiIg9j0KKHksu0cPSzAGR6WhgaUW6Qb8QJsK8FEREREREReZ62sxdAnWNQfDB0GjVMFqtzm2umhaO3hUOp3giNSr55J8CgBREREREREXkeMy16KF+tBjePS3LeHpEUhrQE1/IQYaZFQ5MFeQoNNzn2lIiIiIiIiDyNmRY92HNzhmJC30gYGs2YM7yXswknAMSIMi0AoMlic/tczLQgIiIiIiIiT2PQogdTq1WYNSxe9j4/Hw1C/LTQN7ZsKggzLYiIiIiIiMjTWB5CbombcSrJr6yHzeY+E4OIiIiIiIiotRi0ILfkSkQctGphU06D0Yyahqb2XhIRERERERH1IAxakFuxwe4zLfrHBkMUt0BeJUtEiIiIiIiIyHMYtCC3ohUyLeJD/RAf6i/YpjRdhIiIiIiIiKi1GLQgt5QyLaKCdEiKEAUtOEGEiIiIiIiIPIhBC3JLqadFVJAvksIDBNuYaUFERERERESexJGn5JbS9JCoIF/4+WgE29jTgoiIiIiIiDyJQQtyKyZYIdMi2BfhgT6Cbcy0ICIiIiIiIk9i0ILcimmmp4VOI6wuyq9sgMVqg0Y8VoSIiIiIiIjoHLCnBbnlr9Mg2E8+rhUd5IuUqEDBNpPFioIqlogQERERERGRZzBoQYrc9bWICvJFRKAOof7CEpFTZbUdsSwiIiIiIiLqARi0IEWxMhNEfDQqhPr7QKVSoW+0MNuCQQsiIiIiIiLyFAYtSJFcX4vIQF+oz/St6BsVJLgvq7yuQ9ZFRERERERE3R+DFqQoRibTIipY5/w6NUaYaZHFTAsiIiIiIiLyEAYtSJFcpkVU0NlAhjjT4lTZ2UyL0xV1+GxHDg7nV7fb+oiIiIiIiKj74shTUiTX08I1aJEq6mlRZjDC0NiE8loTLnl9M0xmK9Qq4JVrz8MNY5JkX+NYoR6/HSvGsIRQTB8c69n/ABEREREREXVZDFqQouYyLXpHBkCtAqy2s/dnldXhp0OFMJmtAOz3Pb38COJD/TClf7RzP5vNhnc3ncKra044H//ajcNx9cjE9vnPEBERERERUZfC8hBSJJ9pcbanha9Wg94RAYL7s8prsTu7UrDNbLXh/i/243ixHgBgaGzCfV/sw39+PSEIeKw6VOTB1RMREREREVFXxkwLUiSXaREdLAxk9I0OQk5FvfP2obwapBfWSB5XazRj7oe7kBgRgNyKOlTVN0n2Ka81emDVRERERERE1B0w04IU+es0CPX3EWwTBzL6Rgn7Wvx4sECQPeGqos6EQ3nVsgELAKhukN9OREREREREPQ+DFtSsy86Ld34dE+yL0cnhgvv7RgsniFS7CUi0RFseS0RERERERN0Ly0OoWfPnDEVqdBAqao24dUIydFphrKuvaIKI2C3je2NnVgWyXMahOgTqNKgzWZy39Y1NsFht0KhVyKusx0Nf7cexIj16RwRgTHIERieHY9rgGEEzUCIiIiIiIuqeGLSgZum0avxpcorb+1NFmRZilw6Nw99nDcL3e/NhaDQjOtgXUUE6JEcGItTfBxNeXu/c12YD9A1NCA/U4Z2NmTiUb++NcaqsDqfK6vDt3jwE+WrxybyxGJcS4Zn/IBEREREREXklBi2ozaKCdAj208LQaJbcp1GrMDo5HIG+WtwlE/hobLJItlXVmxAeqMPRAr3s69UazXj790yMSxnX9sUTERERERGR12JPC2ozlUol6WvhkJYQikBf97ExPx8N/HyEh6GjGWdlncnt444WnJ1O0thkwb9+PoZr3tmGxduyW7N0IiIiIiIi8mLMtCCPSI0KxKG8asn2CS0o4QgP0KGoptF5u+ZMM86KOvfjTyvqTKioNSIyyBefbsvBh1vswYr9udVIiQ7C1AHRrfwfEBERERERkbdhpgV5RGqMfKZFS/pOiEeqVtWbUG8yo7HJqvi4k6W1AIBf04sF25fty2/2NYmIiIiIiMj7MWhBHtE3SjpBRKUCxvRpPmgRFiAMWlTXN8mWhiRF+AtunywxwGyx4niRsPfFxhOlMFuUAx6tZbPZPPp8RERERERE1DwGLcgj5HpaDI4LkWRRyAkP0AluV9ebJEELH40KI5PCBdtOltYip6IORrMwQKFvNGPf6aqWLr1Zn2zNxvDnf8Pkf2/Ae5tOoc4obThKREREREREnsegBXlEcmQAVCrhtvF9WzaSVJJp0dCEClHQIjxAhwGxwsBIRokB6YXyE0Y2nCht0Ws3Z+vJcryw6hj0jWbkVzXglV+OY/K/N+Dt3zMZvCAiIiIiImpnDFqQR/j5aJAiKhEZnxLZoseGSTItmlBZKwxaRATq0D82WLDtZEktjrkLWvzR9qBFg8mCp384ItleVd+E/645gWvf3S47spWIiIiIiIg8g0EL8pg/TU5xfj0oLhjTB8e06HFhMo04xeUhkUE69Bc1+6yoM2HLyXLZ5zxZWovcivoWvb47r6/LQG6l++c4XmzATwcL2/QaRERERERE5B5HnpLHzB2fjLReoSiqacCFA2Pgo2lZTExcHlIjUx4SEeiL3hEB0GnVMLn0sDhWJJ9pAQAbjpdg3vkpbu9XcrSgBh9uyRJsCw/wQdWZcawOn+3MwfVjEqES18YQERERERFRmzHTgjxqeFIYLk2Lh5+PpsWPEZeH2DMtjIJtkYE6aDVq2Skl7qw/fm4lIk0WK/72/WFYXQaG6DRqLL1vIt68eaRg36MFehzKrzmn1yEiIiIiIiJlDFpQpxOXh9hHngozGiIC7YGNAaK+Fkp2ZVUqNsu0WG04kl+DguoGwfa31p+UZHA8NK0f+sUE47Jh8UgIE45e/WxHTovXRERERERERC3HoAV1uvBAYaaFodGMslqj7D7ivhauBseHwEdztkzDZLFia6Z8zwubzYY7F+/BnEVbccF/fsfr6zJgs9mwKaMMb/2eKdh3QGwQ7puaCgDQqFWYO6G34P5Vh4skPTiIiIiIiIio7Ri0oE4nzrQAgOyyWsHtSEfQQiHTYnxKBMalCMesrv+jRHbfg3nV2JxRBsCecfH6upN45JuDeOzbg7C5lIVo1Sr8+9rzoNOe/VW5YUwSdC79OkxmK5buzXO7LiIiIiIiIjo3DFpQpwsNkAYt9I3Cso6z5SHuMy2GxIfgooHCiSW/Hi2WHUuaXV4n2bbyUKEkY+LvswZhZO9wwbaoIF/MHhYn2PbFrtOwujbBICIiIiIiojZj0II6na9WgwCdcuNOR6ZF74gAQZaDqyG9QnDJUGEwQd9oxm/HpNkWJXqjZJvYzCGxgjGurm6b2EdwO6+yAZvOZG4QERERERGRZzBoQV5BrkTElSPTQqtRo2+0dIKIVq1C/9ggJEUE4Px+kYL75Eo3Sg2Niq+XFOGPV68b7naU6ajeYRgSHyLY9vORIsXnJCIiIiIiotZh0IK8gnjsqSuVSni/XF+LfjFB8NXaszVuGJMkuG9rZjnyq+oF20oVMi10GjXevmWUbNnK2TWpcM2oBMG2Y4V6N3sTERERERHRuei0oMWTTz4JlUrl/Ldx48bOWgp5gTCFAEF4gA4a9dmMhwEyE0SG9gp1fn3J0DgE+2mdt202YNm+AsH+4kyLey/oixvGJGLG4Fh8dc94nJcY1uyaXV8TADLLamG2WJt9HBEREREREbVMpwQtDh48iIULF3bGS5OXUg5aCO/rL9OMc0ivs6Uafj4aXDmil+D+pfvyBI0yxT0tRiaF4T/XDcdHd4zBmD7CCSTuDIoTZnyYzFbkVEgbfBIREREREdG56fCghdVqxb333guz2YyYmJjmH0A9glJ5SGSgr+C2XHnI0F7C/hLiEpH8qgbszKoAANhsNpTohZkWMSHC12iJ8EAdYkWP+6PI0OrnISIiIiIiInkdHrR48803sWfPHgwaNAh/+tOfOvrlyUspNeJ0NOF0SI4IQHTw2WBBkK8WaQnCUo1hCaGSTIjvzjTk1DeaYTQLyzhigv3Oad0D44TBkhPFDFoQERERERF5SocGLXJzc/HMM88AAN577z3odO6vrlPPEq6QaRERJLxPq1HjpauHITJQhxA/LRZcNRRBvlrBPiqVCteLsi1+OVqMWqMZpXrp5BDXIEhriAMjxxm0ICIiIiIi8hht87t4zoMPPoja2lrccccdmDp1Kn7//feOfHnyYkqTOiIDpQGNmUNiMXPITMXnvGpEL7y8+g+Yz/SyMJqtOF6kl2RZhAX4wM9Hcw6rlgYtTpRwgggREREREZGndFimxXfffYdVq1YhIiICr776ake9LHURipkWMkGLlogM8kVyZIBg2+mKekk/i9hzLA0BgIGioEVeZQNqjeZzfj4iIiIiIiI6q0OCFtXV1fjLX/4CAPj3v/+NqKiojnhZ6kKUpoeca9ACAJIjAwW3T1fUodQgnBxyLk04HfrFBAnGsQLsa0FEREREROQpHVIe8re//Q3FxcU4//zzPdp8Mz8/X/H+oqIij70Wta/WNOJsDUmmRWW95PnOtQknAPhqNUiJCkRmaa1z24liA0Ynh5/zcxIREREREZFduwcttmzZgo8++gharRbvvfceVCpV8w9qoaSkpOZ3oi5BaeRpW4IWfUSZFjkV9c4eFw5tybQA7CUiwqAF+1oQERERERF5QruWh5hMJtx7772w2Wx47LHHkJaW1p4vR11YqEKmRWTguQcVekt6WtRJpofEnuPkEIfBnCBCRERERETULto10+Kll17C8ePH0bt3bzz33HMef/68vDzF+4uKijBu3DiPvy55nk6rRpCvVraJZXig+4BGc8SZFtX1TYKsCACICTn38hAAGBgXIrh9vNgAm83mNquout6EJdtPQ6MG5p2fIhnXSkRERERERHbtdrZ0/PhxvPzyywCAt956C4GBgc08ovUSExM9/pzUeUL9fSRBi2BfLXy15zaOFAASwvyhVgGuFSFV9U2CfWLbWB4iHnta09CEEr0RcaHSYIjNZsO8T/fgYF41AGBnViW+uHt8m16fiIiIiIiou2q3oMVrr70Gk8mEvn37or6+Ht98841kn6NHjzq/3rBhA4qLiwEAc+bMaZcgB3m3sAAfFFQ3CLZFBJ17PwvAnsGREO6PvMoGt/u0pREnYA+MBOo0qDNZnNuOF+vR0GRBTkUdzk+Ngk5rr8TKqah3BiwAYGtmOWrqmxCqMD2FiIiIiIiop2q3oIXRaB8rmZWVhZtvvrnZ/RcsWOD8Ojs7m0GLHihcphlnW5pwOvSJDFQMWkS3saeFWq3CwLhg7M+tdm77549HkV9lf82BscFY9chk+GjUOFki7XdRYmhk0IKIiIiIiEhGuzbiJGoNuRP3CIWpIi3VOyLA7X2h/j7w8zn38hMHcV8LR8ACAE6UGPBbegkAILNM2E8DAEr1xja/PhERERERUXfUbkGLxYsXw2azKf5zbc75+++/O7f36dOnvZZFXixcLmjhoUwLd9raz8JB3NdC7EhBDQAgs0QatCgRTTMhIiIiIiIiO2ZakNcI85cpD2ljTwsASI50n2nR1n4WDgObCVr8UaQH4CbTwsBMCyIiIiIiIjkMWpDXCJPJtIj0QKZFskKmRYyHMi1GJIUhTmF06vFiPaxWm2TcKgCUGphpQUREREREJIdBC/IaYbKNONseVFDqaeGpTAs/Hw3euXUUpg2KwZUjeuGVa4YJ7i/RG3GsSI96lwkjDsy0ICIiIiIiktdu00OIWivMv30yLfx1GsSG+KJEpuGlp3paAMCo3uH4ZN5YAIDZYsWzP6XDZLY67195qFD2cWVsxElERERERCSrUzMt5s+f72y+eeGFF3bmUsgLhAe2TyNOwH2JiKcyLcS0GjUGxgr7XKw6XCS7L8tDiIiIiIiI5LE8hLxGqFwjTg8FLfq4acbpyUwLMfFEkYLqBtn9WB5CREREREQkj0EL8hrxoX7w1Z49JAN1GkQHeyao0NGZFgAwOD6kRfvVmyyoNZrbbR1ERERERERdFYMW5DUCfbW4b2oqAEClAh6a1h9+PhqPPLe7saeemh4iZ1C88hhUVyV6logQERERERGJsREneZXHZg7AdaMTAQBJClM/Wis5QpppEerv47GgiJwhLcy0AIBSvRGp0UHtthYiIiIiIqKuiJkW5HWSIgI8GrAAgN4ymRYxHio9cScsQIf40JaVn5QaGmGz2fDMj0cx4J+/4IpFW5FbUd+u6yMiIiIiIvJ2DFpQjxDq7yNp6hkb0n79LBzEzTjdKTMYseVkOT7feRomsxWH82vw+vqMdl4dERERERGRd2PQgnqM3qLsjfbOtABa3oyz1GDE7uxKwTbxbSIiIiIiop6GQQvqMcQ9IxLC/dv9NQe5CVpo1SrB7VJ9I06UGATb8qsaUMepIkRERERE1IMxaEE9xs3jkpzBAn8fDa4ckdDurznEzQSR4UlhgtulBiMyREELADhZWtseyyIiIiIiIuoSOD2EeowxfSKw6pHJOJhbjQl9I9EnSjpRxNP6RAbCV6uG0WwVbJ+UGol9p6uct09X1KOwpkHy+IwSA0aIAhxEREREREQ9BYMW1KMMigvBoLiWjyJtK61GjQGxwThSUHN2m1qFcSkRgv0KqqUBCwDIKJZmXxAREREREfUULA8hameDRSUifaIC0SusZf00MlgeQkREREREPRiDFkTtbERSuOD2sITQFk8uYaYFERERERH1ZAxaELWzq0cmYFhCKAAgOtgXf57aF0G+Wvj7aJp9bLG+ETUNTe29RCIiIiIiIq/EnhZE7cxfp8GPD56PnIo6xIb4IcjX/msXE+KL0xX1zT7+ZIkBY/pENLsfERERERFRd8NMC6IOoFGrkBod5AxYAGh5iUgJ+1oQEREREVHPxKAFUSeJCfZr0X4ZJexrQUREREREPRODFkSdJLrFmRYMWhARERERUc/EoAVRJ4kJkQ9aBOqEDToZtCAiIiIiop6KQQuiTuKuPOTioXGC2+W1JlTUGjtiSURERERERF6FQQuiThLrJtNi5pBY6LTCX0024yQiIiIiop6IQQuiTuIu02JwfAhSo4ME206WskSEiIiIiIh6HgYtiDqJ3MhTX60avSMCMDBWGLRgXwsiIiIiIuqJGLQg6iRhAT7QaYS/gv1jg6BRq9A/NliwPaOY5SFERERERNTzMGhB1ElUKpVk7OmAM8GKgeKgRakBNputw9ZGRERERETkDRi0IOpE4macjqDFAFHQorq+CSV6ThAhIiIiIqKehUELok40Y0is82uNWoVZafZxp4nh/gjy1Qr2PVJQ06FrIyIiIiIi6mza5nchovby5wtS4avV4ESxHteMSkRyZCAAQK1WIS0hBDuzKp37HsmvxkyXIAcREREREVF3x6AFUSfSqFX40+QU2fvOSwwTBC0OM9OCiIiIiIh6GJaHEHmpYQmhgttH8mvYjJOIiIiIiHoUBi2IvNR5icKgRUWdCQXVDZ20GiIiIiIioo7HoAWRl+odEYBQfx/BtiP5LBEhIiIiIqKeg0ELIi+lUqkk2Rbsa0FERERERD0JgxZEXkyurwUREREREVFPwaAFkReTZFrkV7MZJxERERER9RgMWhB5sfMSwwS39Y1mnK6o75zFEBERERERdTAGLYi8WHyoH6KCdIJt7GtBREREREQ9BYMWRF5MpVLJ9LWo7pzFEBERERERdTAGLYi83DBRichhNuMkIiIiIqIegkELIi83XNSM82hBDaxWNuMkIiIiIqLuj0ELIi8nLg+pM1mQVV7bSashIiIiIiLqOAxaEHm5mBA/xIX4CbYdymOJCBERERERdX8MWhB1AWmibIuMUkObn9NktqKgugFNFmubn4uIiIiIiKg9MGhB1AX0jw0S3D5VWtem5yvVN+Lyt7bg/Fc24LI3t6DMYGzT8xEREREREbUHBi2IuoB+0cKgRVZZ23pafLsnDxkl9ufIKKnFV7ty2/R8RERERERE7YFBC6IuIDVGGLQ4XVkPk/ncyzoyRUGPY0XskUFERERERN6HQQuiLqBvdKDgtsVqQ27luZeIVNSaBLfzKhvO+bmIiIiIiIjaC4MWRF1AiJ8PYoJ9BdsyW9DX4mhBDZ78/jBeX5eBxiaLc3t5rbCHRV5lPWw2m2cWS0RERERE5CHazl4AEbVManQQSl0aZp5qpq9FTX0Tbv5wJwyNZgBAmcGIf109DABQLsq0MBjNqGloQliAzsOrJiIiIiIiOnfMtCDqIlJjhCUip0qVgxa/nyh1BiwAYN0fJQAAq9WGyjrptBCWiBARERERkbdh0IKoi0gVTRBpLtPiaIGwuWaZwQiL1YbqhiZYZSpB8qrq27xGIiIiIiIiT2J5CFEXIQ1a1MFms0GlUsnuf7RQGLSw2oCKOiNq6ptk98+rZNCCiIiIiIi8CzMtiLoI8djTWqNZ0OPCldVqQ3qBXrK9VG+U9LNwYKYFERERERF5GwYtiLqI+BA/BOg0gm3u+lrkVdXDYDRLtpcZjKiQ6WcBsKcFERERERF5HwYtiLoItVqFvtGiZpxu+loclcmyAIBSQyMqmGlBRERERERdBIMWRF2IXF8LOeJ+Fg728hD5TIv8qgZY5Tp0EhERERERdRIGLYi6kJZOEBFPDnEoNbjvaWEyW932yCAiIiIiIuoMDFoQdSGSoIVMTwubzYb0QqXyEPeBCZaIEBERERGRN2HQgqgLSY0R9rQorGlEnajhZlFNIyrr5LMpSg1GVLi5D+DYUyIiIiIi8i7azl4AEbVcn8hAqFSAzaX1RFZZHTRqFepNZozsHe62NASw97Tw0ajc3s8JIkRERERE5E0YtCDqQvx8NEgKD0CuS0bEnz/fi8KaRgDAJUNj0T8m2O3jywxG6LTuE6xYHkJERERERN6EQQuiLiY1OlAQtHAELABgTXoJtpwsd/tYk8UKk8Xq9n6WhxARERERkTdhTwuiLkbcjFOs3mQ55+fOr7KXh5gtVlTXu+99QURERERE1BGYaUHUxfSLUQ5aiIl7YCgpqmnAtsxy3Pf5PhiMZsyb1Afzrxh6DqskIiIiIiJqO2ZaEHUxM4bEIkCncd4O9tVicr8o2X0DdBrFzIxAl+cBAKsNuHvJXhjOTCRZvD0HJ4oNHlg1ERERERFR6zHTgqiLiQryxVf3TMDXu3KRFOGPm8f1RrCfD65+ZxvSC/WCfYfEh8DPR4PM0lrZ50qODEReVT0MjWfHpjY0CctLjhXVYGCc++aeRERERERE7YVBC6IuaERSGEYkhQm2vXHTCFz25lYYzWcbbaYlhELf0OT2eSKDdFCpIAl2uCrRG9u8XiIiIiIionPB8hCibqJfTDCenTNEsO260YmIDvF1+5ioIF8khQcoPm+JvlHxfiIiIiIiovbCTAuibmTu+GTEh/phT04VZgyORVpCKHZnV7rdPzJQB1UzfT1LmWlBRERERESdhEELom5m2qBYTBsU67wdo5BpERnki0Bfjdv7AWZaEBERERFR52HQgqibiwn2c3tfZJAO0UHugxoAUMygBRERERERdRL2tCDq5mKClXpa6DAqORxBvu7jl6V6I2w2W3ssjYiIiIiISBGDFkTdnFJ5SFSQL0L9ffDO3FEYlxKBS4fG4ccHzxfsY7JYUV3vfgIJERERERFRe2F5CFE3F6DTIshXi1qjWXJf5JnSkAsGROOCAdEAgCaLVbJfiaER4YG69l0oERERERGRCDMtiHoAdyUikTKBCB+NGlFBwu0l7TRBpKLWiMXbsvHh5ixU1Zna5TWIiIiIiKjrYqYFUQ8QHeyLrPI6wbYgXy38fOQnh8QE+6G89mwQoaTG8804t5wsw2PfHkJ5rT0gsjGjFF/8aTxUKpXHX4uIiIiIiLomZloQ9QAxIdIJIpFB7ss94kKF+3ty7GmTxYp//3oct3+y2xmwAIBtmRU4VVbr9nF5lfXILHV/PxERERERdT/MtCDqAeTKQ+RKQxxiRc07SwyeCVoYzRbM+2QPdmRVyN5/ILca/WKCJduXbM/B8yvTYbUBd0xMxvNXpnlkPURERERE5N2YaUHUA8gGLYLcTxWJCRZnWnimp8WvR4vdBiwA4EBetWSbxWrDwrUZsJ6ZuvrZztMoqG7wyHqIiIiIiMi7MWhB1APIjT0VN9t0FRvSPuUhu7IrFe8/kFst2VZQ1YCahrMjV2024Fih3iPrISIiIiIi78agBVEPIM6cAIDIQPeZFnGhovIQDwUt9p+uEtyeNihGcPtEsR71JuFo1lPl0j4WGSUGj6yHiIiIiIi8G4MWRD2AXHmIUqaFOMhRZjDC4qjPOEe1RrMk2DBvUh9o1GenhVhtwOH8GsE+p2Sab570cNAis9SAJdtzkF5Y0/zORERERETUYRi0IOoBZDMtFHpaiMtDrDagorZtfS0O5VXDNe6hVaswLiUCg+KEjTcPivpaiEe1AsBJD04RySgxYPabW/HcT+m4/K2t2KXQc4OIiIiIiDoWgxZEPUCIvxa+WuGvu9LI08hAHbQuGRCAvRlnY5MF646VYMepClhbmXkhLg0Z2isEfj4ajOwdJth+IFe4X5bMGNTM0to2Z344/HCgACazFYC9X8aSHTkeeV4iIiIiImo7Bi2IegCVSoWpA6Kdt4P9tBiZFO52f7VaJSkpKdY34oEv9+Puz/bi5g934tXfTrh9/Jr0Ykz+9wZM+99G7DyTubBfFIwY2dv++uJ1HMiths12NiBxqkyaaWE0W5FfVe/29e37WDD/p3RM/99GPPDlPmw5WSYbaDldIXz+vTlVgtenrqOgugE/HSpEboXysUFEREREXYe2sxdARB3jX1cPQ0SgDpV1Jtx/YSr8dRrF/WNC/FBYc7YB569Hi7HheKnz9kdbsvHI9P7w8xE+T4PJgqeWH0FlnQkA8OfP92HrkxdJxpmOSrYHK0aIMi1KDUYU1TSiV5g/DI1NKDPIl6VklNQiOTLQ7fq/35ePxdtzANgDH6uPFCMlKhDzJvXBrROSnb00CqqE41NLDUbkVzUgKSLA7XOT98kqq8WVi7bBYDTDV6vGsvsnIS0htLOXRURERERtxEwLoh4iOtgXr1x7Hj64fYwzy0FJrGhM6uojRYLbJotVtnHliRKDM2ABADUNTVi4NgPV9U2C/UadCVakRAYi1N9HcJ9j9GmWTJaFw8lS5WacWzLKJduyy+vw3E/pePHnY85t+aKgBSDNCiHvt3x/AQxG++QZo9mKL3fldvKKiIiIiMgTGLQgIlniZpwNTRbJPo7ggiu5aR+OjAeHmGBfJIT5A7CXogxPChPcfzDPHjQ4JdPPwuFkiXIzTqWgxvf78mGz2dBgsqDCJcDisDeHQYuuJrdSWBJyoljfSSshIiIiIk9i0IKIZImDFnLEJR8AkCkTaBC3iBjVOxwq1dlGnyNFQYu2ZlqYzFbkKPQ1MDSaUVFnQkG1/D77TjNo0RomsxXf7cnDt3tyYTRLg1sdodTQKLh9qqyOvUmIiIiIugH2tCAiWS0JWhxsYaaF2KjkMMFt8QSRIwU1aLJYkVXu/rkyS2thtdqgFk05AYCcirpmp4vklNeh9kw5gdjxYj1qjWYE+fItsiUe/Go/1h4rAQD8ll6Cj+eN7fA1lIp6n9Q0NKGizoQohdG+RN7MarXh463ZWHusBKOSw/HoDGkPISIiop6AmRZEJEvc00JOQXUDSvXiK9wtCFqIemqMEGVaGM1WHC8yKGZaNDZZZftRANLSkbgQP/SLCRJsyy6vc/t4q00+IENSpYZGZ8ACANYfL0WJ6JjokHXopQ1bWxJAI/JWmzLK8K/Vf2B3TiXe23QK3+xmnxYiIuqZGLQgIlktybQAhCUiTRYrTjczbtJHo5JMdQgL0KFvlHASyJbMMmSVuw9aAEBGiXyJiLh0pH9sEPqIJo3kVNShoFo+aAEAe09XKr422WUUSwMDHR0sqDOaZbNm5EqVvMXJEgPmfrQTc97aig3HS5p/QCuV1xqxJr242dHA5L22ZgqbCW84UdZJKyHynKKaBizdm4c/ith3iIhajkELIpIVG9zCoIVLRsLpinqYmynLGNorVDbFeWyfCMHtj7dkw2S2CraJAxsn3Zwci7f3iwlCSpRwhGlOeb3bTAuAfS1a6oRM4Mg12LTxRCnu/Wwv/v3rcY/0u1i+Px/TXt2IG97bgcwzP2dxaYjDqVLloFd7yS6vw4li5ek2Ty47jG2ZFThSUINHvj6ImoYmxf1bo6imAZe8thl//nwfpr26CQc4DadLEh/XzByirq6wugEXv7YZf/3+MC57cwu2npRO+SLqak6WGLBow0n8fry0s5fSrTFoQUSyQvy18PNp/i3CMekDkJaG+MsEJ8SlIQ5Xj0oQ3BZP9Qjx02JCaqRg28kSA9akF+PG93fg0W8OOJsxZorKQ/rHBKOPKOBhLw9xfxX6QG51s30xurKKWqPbnh6tcVImaJF9JmiRV1mPu5fsxW/HSvDuxlN48vvDbXqtopoGPLnsMLLK67A7pxLzf0oHAEmJkkNLSpU8yWS24qnlR3DRqxtxyeub8eyKo7L71RnN2O8S7Ks1mj06ZvfHA4XO3x+TxYr3Np3y2HP3JKX6RhTVuA9stjdxmVVBdQPqTW3/nSXqLCsPFcLQaD+GrTbg8505nbsgojbKr6rH5W9txau/ZeDOxXvw44GCzl5St8WgBRHJUqlUsiUijlGlDofza5wn95miK4HDk0JxwYBowbYLBkTJvt74lAgkRwbI3gcAfaODMEDUl2LtHyX48+f7sCu7Ej8eLMRzK9Jhlmng2T82CCky5SFKmRa1RrPb8pOu7qXVf2D0i+sw6oW1eGHlsTYFL+S+R1lnggW/HSsRZN78eLAQO05VnPNr7cyqQJPl7PPtzqmE1WpDiZtMC8fxeLLEgEtf34y059Zg4dqMc359JdX1Jtz+yS587dJ34LMdp5FXKQ2MZcuUPTWXmdEaWaJgza5s+/eJWm7xtmxMfGUDJr68Aa+10zHTnDKZ41qpz483yC6vw6G8ak7uIVl5VeLR1N3zbyz1HL8eLYbRJSv4u715nbia7o1BCyJyS65E5MGL+glu15sszhNX8ZXt1Ogg/Pe685CWEAJfrRp3TEzGVFEQw0GlUuH60Ylu19I3OhD9Y4MF2xxXbBzWHivB0UK94MQWAPpFB0kyLepNFslJgY9GOIlkbzcsEdlysgwfbM4CYL8K/8m2bMxcuAlr0otb/Vw2m03S9BQ4e1J+XKZmef5P9sDSuTheJPyAazJbUVZrdJtpUVDdgAaTBfNXpuN4sQG1RjPeXH/S47XU2eV1uOad7diZJe2DsjNLGqSRC1pktPDD++/HS/HciqP4ZGu224an4l4t1fVNyFAYEUxCRrMFr/6W4QzGvrMxE5WizK+OIPfz7ejsodb4encupv9vI658exvu+2KfR5/bYrXhuRVHMXrBWsz7dHen/Dyo7cQNk3Mr69HY1Dljsok8obBa+D6dIfOZiDyDQQsicitGNEFEq1ZhzvB4JIYLsy0cfS1Oia4CpkYHITbED6senoJjL1yK569Mg0olHVHqcO3oRMhMMHU+V//YIPk7zzBbbfh0W7ZgW1SQL8IDdYgL8YOvVvktTxxQ2d/NghYWqw0vrT4u2V5U04g/f74Pj393sFUlMUU1jTDIZGnkVTXAZLbiuMyJ+IkSAz7febp1Cz/jmEywIa+y3m1PCwA4kFslye444MHJMPUmM279aJfbprG7sqWBDLmghdz3SmxnVgXuXLwHS3acxgurjmHCy+tx0wc7sGxfviCTQq7B7M42ZLj0NFllwnHITRZbh2dd1RrNqDdJT+bE2WwdoclixdK9eViyPQeGRve9V95YdxKOw3BNegmOFtR4bA0bjpdiyY7TqKgzYeOJMry14aTHnrun2HKyDBNeWo/hz//WaVeDxVlxVpt3Zw9ZrTasOFiAj7ZktUug7Ovdubjh/R2Y/1M66jxQrkkdr8QgDFqU1xpRxaBqu2i3oIVer8c333yDJ554AlOnTkW/fv0QGhoKnU6HmJgYXHjhhfjPf/6Digp+kCLyVvGhwkyLUcnhCPbzwUhRX4oDuVWw2WySRnGuY0Y17qIRgtfzl5STOKRGByI6yBeh/j6Kz7HqcJHgdv8za1CrVZIJIq5C/X0wdWCMYFt3myDyw4ECxSyD5fsL8OpvJ1r8fO5O5CxWm70ZpZv7F67NQHmt+0CDO38USZ8vr6rebaYFACzZkQNxHOZ0pec+JP94oFBxCs3uFgYtMstqm81AWXGwUHDbZgN2ZlXiiaWH8OLPfwCwf8guqpZ+P+SCJyRP7rgW/8waTBaPlUAs25eP81/ZgMvf2uI80XeXReOaaWG12tq97KfBZMGtH+3CX78/jOd+SsctH+6Sfc3KOhOKRWv2ZNBC/F7sOmaZmmez2fDU8iMo1jeipqEJz644ipp6zzX/bSm592rxtC9v8upvJ/CXbw7ixZ//wBWLtkqag7fFwbxqPLX8CHZnV2Lx9hy8/Xumx567p9h3uhLT/7cR4/61Dt/vy++UNcgf08y2aA/tFrTYvXs3br75ZixcuBCbN2/GqVOnoNfr0dTUhLKyMmzatAlPPvkkBg0ahDVr1rTXMoioDaYPjhXcvmlsEgBgZFKYYPvBvGqUGqSNHVNjlDMj5NwwJkl2e9/oIKhUKmcQwh1xpoBrdkafKPc9MxLC/DFaFIzJq2xo9cm12WLF8yvTMey5Nbju3e2KJ7QdqbHJgv+JAhI6mcyTdzeewqrDhZLtcuRKQxw2HC91+wHP0GjGf36VZnwoKTMYZX8W+ZUNKNG7/xn9JnNyI9dn4lztyhYG3uNEfWByK+slzRzlsjJMZitymhkXrHS1/5s9ubBabSirNcIkE/zYlV3JPgMtpFTyBADPrjiKIc/9iokvb2jziXmpoRFPLT+CguoGHC3Q458/2pu3itPoHRyZFhklBsx4bRP6/WM1nl1xtF1+to1NFtz7+V5BwOtIQQ0O5Ekz0MR9VICWZQ8BwKaMMvzjhyP4ZGs2quvlr1AWiPoP5Vc1ePT3uLsrqmkU9HBqbLIivchzQSU51fUmQfaA1WqT7dPSGdlD2zPLMfW/v2PEC7/hC4XMv2/3nM1Iya9qwLZMz0072ZIhHGEsvuBCzXt6+VGcKqtDqcGIf/xwpFPKxuQyPb05ENeVtWt5SFJSEm6//Xa88cYbWL58OXbs2IFt27bh22+/xfXXXw+NRoPy8nJcccUVOHToUHsuhYjOwYS+kXh37ihcMyoBr1wzDFePtE/4GNE7TLBfZlmtJOXe30eDeJlGns2ZMTgWEYE6wTa1Cs4mnWNEo1GHxIcoPp9rkEPc18JVYrg/BsQGSSaeHM6vdn69fH8+5n26G2+uP+k2lfN/azPw6bYcGIxm7D1dhXe85OrJx1uzUVQjvCLw/q2j8dqNw6EVZcH8delhSUaGzWZPk311zQln8zR3mRQAsPqI8gewpfvyW9WEzV2GSF5VvXNqjBy5c7lcD57s7BFlMDw4rR9C/LSCbbtcel3YbDZku+lLoPT9sNlsin0v6k0WFOkb3TaXrawzdeurP7uyKnDj+ztw+ye7sTenbVkl8s1l7UGLvTmV+GzHadhsQLG+Efd9sQ8NMmUcLbU7u1IQZDqUXw2j2eL2mM4pr4fZYsWLP/+BrLI6WG32hq+enD4D2INoD365H1tkRlLK9W6R67XRkpKaA7lVuOOT3fhyVy5eWHUM419aj78uPSQJBskFf9vS1LenkQsMyGV8ecrCtRkYtWAtxry4Dj+fORmvrDfJjkRXCn63h32nq3DXkj04XVGP6vomPL8yHYUyx1dlnUkyxaylgbjyWiOW7ctXDGoWioLZuZX1XnORoyuoqjMJPoMYzVbB57WOYLPZZLPiOvqY7inaLWhx0UUXITc3F0uWLMEjjzyCq6++GhMmTMCkSZNwww034LvvvsP3338PADCZTHj++efbaylE1AazhsVj4Q0jcNO43s5+FEN7hUCnOfv2YbMBS7bnCB6XGhMIdQtKQsR0WjWuGiEcf5oUEQBfrT2Y8KfJKRh1JmgyfVAMvr5ngqAMRaxfzNnmneIJIq4Swv2h1agxLDFUsP3gmWDMnpxKPP7dIWw8UYaFazNw7bvbkSu6Mv778VK8u1E4XvLQOf4R3XKyDP/77QQOeOBkpKLWKFnXpNRIXDgwGlePTMRzVwwV3Ndw5gqra13mog2Z+Ms3B7Ho90xc99525FfVy447dTgi+rCWlhCCQN3ZgJDNhlZNZXAbtKhscHtV2h3xz+1c5VfVo1AUCJqQEoFxKcLAmuuV6so6E/SN8gEvpSBQsV6+f4ir7LI6xQ+9ck1Bu4NSQyP+tMSeEbA5oww3frATH27OOufsA/kTPPu23aKASH5VAxb9fu79FY4VioOD9sZu7o5pk8XeK2bHKWEwwZN9WgDgr98fwvrjpbL3yZUayfUlaElQUlzyZDRbsXRfPuYs2ir4myLOtACAHd30eG4PcgFL15/Z78dLcdvHu/Dk94dR09C2spGC6gYs2mDvb9LQZMG/fj7m9uTOvjb7cZJXWY8r396GtOfWYMGqY+2SPZRZWos/LdmDxqazgcImiw1bZYJzctlDLQnEleobceF/N+KJpYdw+VtbMfejndgn0x9LLsC8i8d0i2XK/Hzasz9KemENHvjS3vvL8XdW32gWHEvOtZ35fXP0RHnll+MeLZfrqdotaKHRaJrd56qrrsLAgQMBAFu2bGmvpRCRh/lqNThPdHIv/gCZGt360hCH2yYmC05wrxzey/l1dLAvlt0/Cdkvz8ZHd4xBaIAPpg2KkXsaAOLyEKVMC3smxwhx6Uu+/Q/NMlG95PFiA654e6szXbSwugGPf3dQ8rzik5KqOhM2nijF6iNF+OFAPr7bm4f0QuEfM/sHyN14a0MmbvxgZ5v/2L362wlJ6c7Tswc7g1C3ju/tLP1xyKtswPyV6We+rsdbG85mjBgazfhsx+lWXbk/v18U7rmgr2Dbr+nFLf6/uQtanCytbfZkXkzfaPZIPfce0QlseIAP+sUEYXxKpGC7awmJ0tXNE8Xu+42IO5IH6jSSLKPsijrZK4bOdchcIe8O3t+UJTi+LVYb/rX6D9z3xT7oFRpHymlssiCnQvozyq20ZzjI9VX5YHMWMs8xHTi9UPozz62sd3uCBwBf7c6VTEjyZPbQ7uxKSTDB1b6cSkn/FblMi4o6U7Plde5OAm02+9V6q9WGxiaLbAr2jlMV3brkqbrehI+2ZGHxtuw2BxKUMi2Kaxrx5y/2YcvJcny7Nw9/WrynVQ2ZxfadrhL0ESqsaUR1fZPbhsk5FfUwma145ZfjOJRXjVqjGR9vzZbN6GmLEn0j7vhkN6pl3vvlArpyx3RLMi2+3ZMneD/allmBa9/djrsW7xH8XjN7qG2ayx4qM9gv1izfn9/m3j9GswV3L9mL1UeKsXx/Af7y9YEzryH/Pu14X/ty12n85ZuDeG/TKVz9zjbkV7GkrS06fXpIcLD9Kmhjo/s/0ETkfW6bmKx4f1uCFilRgfj0znG4YUwi/j5rEO6/UDhmVaVSOf8BcBu0CA/wQaRLqUmKQtAiIcw+EUUctDiUVw2bzYatMrWs1fVNmPvRLpz/ygZc9+52VLk5EXZ8ANyfW4Xz/70B8z7dgwe+3I/Hvj2Ev31/GJe/tRUrDhY49/9sR47za5PZinc2nnuJyYHcKnzjUpcLAFePTEBawtmgk0qlwvNXDsVIUdnPioOF2JRRhv+sOSHpk/DlztOy0w3cGRwXgrsmpyAsQNhIdWELsy3kThYBnFNDT8AzJ3niJptj+kRApVJhfF9hpkVWWZ0z3d/dlBFAeVSaOKulf2ww+kYLj+fssjrZK9IOu7K730leqaHRbU36mvQSXPPO9lbVOTtKLsSaLDYUVDfIBs+aLDY882P6OX1v3U3EEU9ZcLV8v7ThXHP9UFrj9xPCDAtxyVydyYKjomCLeHKUQ3PZFkqBz5qGJhTpGyVlbQ7F+kaP/r+9idlixbxP9+DFn//A/JXHMPuNLW0qAZILqjkyCTZnlAn6D+09XSX4G9Ra6TKBaKWGyRarDSeKDdggyuzxZMmTxWrDPZ/tdZuJtjNL+t4od9X+VGnzDZPdBTY2HC91jgK22WyyAead2d0/aFFc06gYXG8puaBF1pmMOKPZgqve3oZ//3ocj393CH9ffrhNr3U4v0bwPrT3dBX0jU1u+2mVGoyoqW8STEprstjw69HWj5anszo1aHHixAkcPHgQADBo0KDOXAoRtdKc83op9pNQKtloiXEpEfjPdcNx39RU+OuUM7dGJ4dL+ggAQP+YYMGI1ZhgXwS4eS7HGNfhoqBFTUMTNp8sd9srALBfMRGXCThYbUBFnf0P2ydbs2VP9G02+9Vi+9c2HMoXfuhbe6wEFedwcm6x2vDsinRBX4cAnQZ/u3SgZF9frQbvzh0t+T4+8d0hrDwkvepa18o6/kHxwQjx88G9omyLDcdLZVNnXRnNFtmrXnICdJpmJ8wA7RO0GHem38qQ+BAE+Qq/j3uy7f9HpUyLnIo6t/0RxCd/A2KDJEG47PJaxfKQ8lpTi7+PXcX7m7JgVOjon1lai3s+24vGppYdr0oN1P4o0sumjAP2TDOl7AQ5pYZG2caEzU3EkUtHzpXJDjlX4qvOd03ug1RRgMw1jd1ktrr9fVIKWlTXm2T//66aC8RtP+W5xoje5OcjRTiYV+28XVDdgBve24GPtrS+7Mlms8kGhxyjqeUCZ//59cQ5l9EdLZQJWjTTMPmr3afRIPodPe3BY3pzRhkO5yv1l5D2A5J7rzRZrLKZWK6USkgO5FY7e2XI/R7nVTZ066vxb64/iQkvr8f5/96Ax787iHrTuY95lTums88EmraeLBf8Lfxubz5+kvkc01JHZI6dvGYy4radKpdciGju2CFlHR60qK+vx8mTJ7Fw4UJMnToVZrP9gH300Uc7eilE1AZqtUr25NehLZkWreWjUcuOSu0XK1yDSqVCspu+FklnykN6hfohKshXcF9bR5E56tOV6i2PF+thaGxCflWD5Mpwk8WGHw4UuHnkWWaLFVtOlmH7qXLUm8z4Zk+upLfEI9P7Iz7UX/bxcaF+eGr2YMG2lmYyDE8MhbsWJj4alfN4mDepD6KChI1Wm+ttkVlaK9vATU5siF+LAmbnMvY0s7TW+UG2otYoubrs6GWh1agxOlk4icZRIpKtcAzYbO5PmjNEH9AGxAZLghY5FfWKJ3gAsKMblYjIZVlM6BshyebZd7oKj393sEUpwkonHGvSS2SzMBxe/PlYq9L4xaVjDvmVDW5T6d3Jr2po9gpwS9QZzZKTu4l9ozC+r7jk6exxlFtZ57acQCloIf5A76NRYVBcsGBbdkWd4klcd0ynt1ptsn9zzFYbXvz5Dzz41f5W/awr6kyyJREWqw25lfWyx2FDkwVPLjt8TgESuZKnvCrlE7zl+6V/3zyZRSMObiWG+yNc9D4hLnF1nz3kPvBrMlubbXCaXa4ciPN0WYy3+KNI78ystNnsP/Or3t7mNhDcnFMyQYvCmkbUm8yyx+A/fzgimeTVUvKBuHrF9+nF23Ik205308ywjtIhQYvFixc7U7kDAwMxYMAAPPHEEygpsY+i+/vf/45bbrml1c+bn5+v+K+oiOODiNrT1AHRmCBKhQeE0z46yvTB0hIRufGoKTJjT4N8tQjxt18ZV6lUkhIR8RX1y8+Lx18vGYhgX2l2h5+P9G3VcTWxWOFDm9UGHMqrcXs16Lu9eYofIG02G277eDdu+3g3bvlwF4Y//xteWHlMsE9qdCDuOj/F7XMAwI1jkjC2T7jiPnKG9ApFUoT8z7xfTDB8zjRuDdBpcd/UVMH9WzPLJf0hXLkrDZETHewruSosxzEu0WazYd/pKhwr1Ct+fxf+dgIzFm7C9P9twsur/8CeHGF2SIBOg6G9zmYeiUtEHP0kmvtAe6LYgMzSWryw8hje2ZiJOqMZNpsNmSXiTAtp0CK3sh55ohO8+FDhBJ+2NHozW6xtqnX3NHGWhVatwn+vG46VD01GL9H/e/WRYrz8yx+CbSazFf/59TjuWrzHOeFAqURnnWh8rnhkcHmtCa+uEY4VViL3wRqw/xyVMi3kmK02t2UUrbH3dJXgZ+yjUWFUchjGi5rL7smudO6XWaoQjFUIAokDdH2jgtA/VhS0aLa5bPcb5fvbsRLF43D1kWL8pxXHmdIkg1NltW77Be3IqsDXu/Nk73OnoLpBNkDS3AmeXLaUpxomA9JAwHWjEzExNVK0TwuzhxSO6dMVdZIAu/giSE55z2yY/OZ6acPijJJaXLFoG3492rrztTqj2e33MKdcPhCnbzTj/5YeOqf+FukFcmV8DYqBOHHTZsCzvYd6ok4tDxkxYgR2796Nl19+WZDC3VJJSUmK/8aNG9cOqyYiB5VKhScvlZZ2JUUEwM9HuaTD06YOiJFc6e8fEyzZr49MpkViuL/gPWhEUqhkH1dT+kfhwYv6Yc8/Z+DnRybjrZtH4i/T++Phaf2w4sHJgpNXwH5FuLHJIsmgEJcQ7Dtd5XbaSEZJrSBdWOxwfo3gSlGTxSb5IPjClWmSEy0xtVqFl68ZBh+N9D05IcxfMDXG1UCZcgWHwaKrp7dOSEZsiPCDnFwJioO7D9VyYkP8ZLN8xNkdjg8Pf/v+MK59dztmv7kFr62TnwJR09CE986U7wDA+5uz8Po6YXbIqN7h0Lp8b8QneSdKDCivNSJblB4qXteG46W45p1t+GRbNv7z6wk8vzIdBdUNknIcuaCFxWqTlB85xhQ77Mpu/UmeyWzF/y09hH7/+AUzFm5qUQf99iaXZXH9mEQkRQQgKSIAn945ThJU/HBLNr7Zneu8/fflh/HOxlPYcLwUD3+9H/tzq2TrpB3EDV+nDYzBDFGw9Itdp3FI4ffUlVxaPmDP6Glt+RXgmat44hOm4YlhCNBpMUGUaWEwmp2/l0olRydLDG5PEsQn0/3dlTwpXJUurzUq/sy6GptNmmUhzhwC7M1fmxsr7SA3ZcFhy8kyxUbGL63+o1VjOI/KnNwB9lKU1gbiivWNbRon7KBvbJI0ux6fEik5pl0bFStnD7W8YXJsiK/k80RzmRbdMXvoWKEev7jp51BrNOP+L/fjlxYez4Dye05WeS3Si+Qv/mzLrMCnokl3zWkwWWQzIHObCcTJKfBQRlxP1SFBi6uuugpHjhzBkSNHsHv3bnz99de4+uqrcfDgQdx8881YtWpVRyyDiNrByN7huGRorGDbwFhpsKC9RQTqcJXLCVrviADJ6ElAfoKIowmnw4gk5UyDSalRAAA/Hw2G9grFnOG98NjMAXji4oEYGBeMmGDhCXmp3igbkb9sWLzg9r7cKsUTnu/2ur/q1VyvgsvPi8f5/aIU93HoFxOM+0XZEIB94shM0c/aYUBsMPpGyZdlDIoXHg9+PhrcPrGPYNsWmZFzDuKghdIk3ZhgX9mgxY2i6Si5lfXIKqvFUpepMO/8ngmDzLSJkyUGSSNScbO1sX2Ex9qwhDBJ1s2PBwoEDe8AYOaQOMHtX44WC0aiLt9fIMn0CfbTIjbEF2EBOkmKsyuVShq0KDMYkVd59gNzdb0Jqw4XKl7VfOWX4/j+zPcpu7xOMEmmpaxWG97fdAq3f7Ibn7ehyZ/D6+tOSrIsHnBp2DswLhjv3zZaEnx7fuUxZJXVYt2xEkFKutUGvL/pVKvq6AfHh+C5OUMFP2ebDfjnj0dblJHyh5tMC3FtPwC3wUJXnqiXFgctHCd2sSF+6CPKnnPsq1T2Vm+yuO0HJA5+9Y8JlmTCZZfXIb+Zk+bt3egkb/PJcklJ3/NXDMWnd46VHAN/XXpItsHm9lPl+GDzKWdWlzhLy5W4MaD4NWqNZjy9/EiLA53i4IBDfmW9Yk8LdzxxZXpvTqWgrEunVWNk7zBJ0KKgusGZgaeUPaSUBSM+puWCy9nNZFq4rqO7eGO9cgmozQb85duDihmXrpSyhw7lVQv+xon9+9fjLRrH7PBHsV62LLC53kNyzFYbCqs5eOJcdUjQIiwsDGlpaUhLS8PYsWNx0003Yfny5fjss8+QlZWFK6+8EosXL2718+bl5Sn+2717t+f/M0Qk8fTswYg+c6KuVtn7FnSGF65Mw9OzB+HuySn47K5xslkFctkAjiacDsMS3WdaJEcGuC2DcIgJFqamlxqMktTtIF8tpg4U9uE4kFulOAJ05aEit42rlBqFRgbq8I/LBru9X84DF/XDYJdGq1P6R2H2sDhcPzpRdv/+scFIcVOWMShO2rB1qqgHSXZ5newHNZvNJglajOkjDUY5xIb44rykUGhdIhu9IwJwsSg4UFjdKAmUmK022Q9NSh9SHcQBMp1W2tfinY2nBLeDfbWY3EwgySxT3z4w9myDWaUxvrHB9v4erhN0AGBfrv3/WFPfhEte34yHvjqAC/77O15bmyE5OVmTXoxPtmULtolPTGqNZvx8uAgv//IH5n60E+e/sgG3frRL8PNcfqAAL/9yHJszyvDMinRnOca5yCgxCDImAHu6t/j3clK/KPz72vME2xqaLHjkmwN4+ocjkudtrmeF2OD4YCRFBOCR6f0F248U1ODLXfITTRzqjGZJ1o07wb7aFvVpcT3BKzMYm52aUmYw4qXVf+CFlcdQqm+U7WfhemInHeVrP46aC5ieKDHAZrMhu/zsFB1A2kjP3lxW+P/Mq2pAjqikSpx50J2uTL8tCgj2iQzAZcPicdHAGMy/YqjgvjqTBfd+vk8QaP10WzZu+XAXXlp9HJe/uQW5FfWKmRbltcJjZEyfcNwwRvgevymjDMtkek7Icff3K7+q4ZwmPXkmECd8Tx+RFAY/Hw36xwQhQvTe6AjEKR3TSg2TxVk/9kCc8JjOKq9T/Hvtuo7uIL2wBmvShaV1D1yYimtGCQPqJrMVdy/ZK5s5ZWhswrFCvTNLQemYXn1EGIgTX+Qwma149NuDMJpblsXj7pjOPcdAHJtxnrtOLQ+57bbbcP3118NqteKhhx5CZWXrms8kJiYq/ouPj2/+SYiozZIjA/HTQ+fjjZtGYO3jUzGphVf0PS3IV4t7L0jFPy8f4vZkTr48RHiyE+rv47YvgiPLQkmMqPSh1NAoybSIC/WTnNQaGs2StHDXyjnHyaEccbO6Kf2j8MCFqbj/wlT88MD5bptvuuPno8HX94zHU7MG4dnLh+C9W0dDpVJhSv9oSWlHRKAOUUE69HXzPRdnWgD2CRvik+nNJ8sk+5XojZJRshcPkc/2AOwBo5hgPzw7ZwgCdRrEh/rhlWuHSXqsWKw2LJdpbip3AqQ0UQKw1/2Lx8UCwKVpwr9B4pPIlOhADIxrPitJ3BDOte5fcYzvmbKnUaLjzDGt5YtdpwUfut5YfxKPuXyYy6usx1+XHpI8r+uV9ZMlBlz434148Kv9eH9TFrZlVqCgugFbM8vxr5/P9pAQl/98uCUL5+rl1X8Iggt+Pmo8OmOA7L7XjErErRN6C7YdLdC3KK1XXIsu5gjq3T25rySo8N9fTwhO0MWOF+vR0iqdmBDfljWXPfNh+N2NpzD+pXUY9691WCwKOLl6+Ov9+GBzFj7Zlo2r39mO30+UyvazcBD3admTUwmr1SY5wRNnt/xRpMdj3x7ERa9uxKSXN+CHA/myk0P6xwYjJVJa8iT+WV01QniyszO74pzq1H85UoRbPtyJp5YfQY2bcdUdaWdWhaQO/v4LU51lZzePS5IEjbPK6vDnz/ehscmC7afK8aLL71ydyYJPtmUrXpUWGxIfgn9cNkSSLfjCyvQWXVUWj8J1MFmskl4PzZUqAuc2QaTWaBYEFcR9fByBOJVKJSnjcwQ4lIIWNtvZ4ITNZhMEeiXZQzIlTznl0uayWtGZtbgpaFf2hqjsMtTfB/dfmIr/XT8cN48TZkDWNDThjk92C461/blVuOjVjZj95hbMemML9I1NiiVh4iyWQXEh+NNkYS+vP4r0eG2tfDmomFIgTuk93p3T3SyLpiN1atACAK688koAQF1dHX799ddOXg0Rnav4UH9cOSKhQ6eGnIuoIB0CRWNPE8KlJ/Ti0acOzV0ZByAtD5HJtIgP9UNsiJ+kNMVVr1A/SUbCu5tOyV7lEV+5mdwvCn+7dBCevHQQep9jU9SwAB3+PDUVd01OQeCZ/gAatQpXjxR+cB4UZ7/y31cm0BMVpJNkngD23hmT+wu/l5szzgYt8irr8fPhIry/WZidEOQrra935QgY3T6xD9JfuBTb/z4Nk1KjEOrvg2DROFe5Uhy5D4vNfegflhAq28PlulGJkit5rvpGBaJPZECLPry7GuAyFUd8kufKcWyJg2P7T1cDEH6/HX48WIjr39uBvy87jHmf7haUqriqO1MH/8m2bLdXUDefLIPFav9QL055P5hX3aIUXZvNhqMFNTh55mr91pPl+P2EcN33TumLuFDpMebwj9lDWtScVWxYQoikoadDsK/WmaGl06qx4Mo0wf0Goxlf73JfzuWuCaecmGD5Pi3inh2nK+pRXmvEwrUnYLXZs3Re+fU4amV6FpTqGwVXoAuqG/D3ZcLsE0c/CwfxBJHq+iZsO1UOg+gYEb9Hfrw1Gz+eGQdrttrwwspjkkZ5PhoV+kQGIDTARxLMFLtOdOJeXd8kufJaU9+kWNLw44EC3P/lfmw/VYGvd+fif2vdN7ZsbLLgjyK97HjW1UeK8OT3h1vcX8Idq9WGl1cLG8X2CvUTvNeqVCosuCoNaQnCzLXtpypwz2d78fBXByRlScv357eq7n5wfAhC/X3w0tXDBNv1jWb848ejit/TUr38CF85ahUw0s3fWFeOCSK5FfV4dsVR/OOHI5LMG1crDhZg/L/WYfCzv+KdjfZyP/F7j2vzcElfi+zmS54Ae9Dx6925GP3iOlz46kbsyamUnRwyIDZI8nexockiyTKaNkjYG2fXOTaYNZmt2JxRhpNe0HcIsJ/w/yZqYHzPlBQE+/nYj+cr0zBd9H8vqG7A7Z/sRk19E/Iq63HPkr3OjKCTpbX4aEt2q/rYDOkVgr9eMlBStvz+5lMtakztrk+LyWyVjK0Vf7aU09rx1C+sPIb7Pt+H//12Aj8dKjznCSjdQacHLaKjz34gP31aOZ2SiKitVCoVLhx49o+kTquW7X3h7gOVuOO4nGhxeYjeiGJR0CIuxL6P+Cq4q+FJYbhJ1Ishq6wOL4k+3AKQTI0QZ4940l2T+zjXD8B5FSM22A/+opN3udIQhwv6CwMy2zMr0GSxYsn2HEz5z+948Kv9+FQ0NmxgXLBiEEYcIHGUUdjH3Tb/PUkv1EuuujaXaTFW5vgBAH+dRrFUKiUqCFqNGv1aGehz/fDlriQHOBuMG9VbeIwdL9aj1NCI/blVcg/D4fwafLMnz+3IPwDOEyGl0pl6kwUZJQYUVEvH+ALAt3uan0zw6LcHcflbWzHztc2Y8p/f8eSyw4L7o4N98WeZ/iuu/HUavHHTSNnmsuIrnK4GKJU8xQcLmvdOTI3ElSN6Cfb5/USp2+d2N+5UTqybTItrRSfvuZX12JVViSbL2ZOdxiYrDsj8nMUnTQAkwQ3xCV1CmL+klO6DzcKMGV+tWnICJh4DW1XfJCk56nvmdwFQLnny97FP6REHkxzZQ1arDQ9+tR8jFvyGyf/+XdKzAbCX4T367UHBNnGpWH5VPd5cfxI3f7AT5z3/G2a9sQWT/70Bv6Wffb5NGWV44Mv9+HZvHh74cr/gvtZadaQIh0SlOfdfmCoJaPr5aPDu3NGSYOiWk+WokPkdEwcdtWoVNArH/JAzTaRnDInFVaLjee2xEmw47v6YlhsL6U5UkK+g/NCd0xV1sFptePjr/fhsx2l8uSsXD361X7ZnjMVqw4JVx5zZiv/59QTeWHdS2M9Coxa8H4qzh/Kr7P0kxJkW4oD39/vy8dTyI6isM+F0RT0e+/Ygssqlo7n7xdh7XAWITmbF6xf/LhdUNzRbQiJWZjBizltbcfsnu3HJ65vxo8KYdJvNBn1jk2yGUnmtET8dKvRIX403RBNDwgJ8cIfL30StRo23bhkpuUh0vNiAOxfvxp+W7JEc19/vzWtVBs6Q+BD4+Wjw2o0jBD1bbDbg8e8OQS/Tx8qhscnSqsbTcp8lxRwNk8trjXhr/Un8d81xFCr0ONmUUYpf04vx1oZMPPL1AWzJcN//q7vr9KBFQcHZX6qgIO++QktE3cNzc4bg6pEJmNg3Eu/fOlo2DVwu02JIfIjilXMHcXlImcEo+aPkGEU5WqaswOG8xDDMGByLYQnCHhuf7zyN9X+cvXphtlhRJGruJD658KSYYD/8+OD5ePX64fj10SmYPtherqFWqyQnHIMUyh+miDItDEYzVh4qFJQViA2JD0GInw9C/eUbUIq/9656N9OLBLB/kHFN0a5paJLUrU50OZlTqYArhgs/3Lu6fWKy5AOrg+OEWPzhPSHMX3L1yVWLy0POZFqclyjs8WG1AR9uzhKc3LaWo9xJHIwTO5hXjSNuxvguP5CvWFecWVqLFQfPlpXkVzVIUn+fmDnAmQWkJC0hFI/PHCjZvuCqNLfZTv1iFCbiyJxwiTMADudXo7pevq+EONNCbkKEQ0yIHwbGCT8fadQqSXPZepMFPx+RTuHZky0tvW3JlVi5jKaJom3ik/2+0UEtOhld94fw5Le/a/aQwjHtmPQ0WtTXZu+ZEcSrjhTh58NFsNnsJ373fbEP//zxCBrPNDctrmnEnz/fJ3ne7PI655Xtw/nVmLFwExauzcCOrApn81yj2YpXfzubkSFujPzWhswWXx3Prah3BvKMZgv+8+txwf19owJx07jecg9FUkQAPrtrHEL8mj/uxfpEBbp9H/TRqAQZPc/NGSqZbrRcobeFuyvScmJD/GQDcUkRwt/F0xX1OFFiEAR00gv1OC4zwSOvsl7So+OjrcLgmKOfhcOAmGBJM+OfDhVKsodmDhaWJO4S/U7lVzXgy53CPjuxIb4I9bdnFCgd04C9x5P4s4UjEAfYMxYWrDqGL3aels20rGlowm0f73KOY7XagI+2ng0o2mw2bDxRige/3I/Zb2xB2nNrcN783zD6xbWCksi8ynpM+ffveOTrA7jk9c1uG6u2xNGCGqyVZFn0RbCf8PsdoNPikzvGSBr97s+tlg2KF9Y0SnoPyQWkHRyBuCG9QvDExcIywoLqBnymME0ko8QgCUS5ExbggzTRZzU5jt5DT3x3CP9bm4G3fz+F+7/YJxuIM5mtzmwjh36xPfdcudODFkuXLnV+PWzYMIU9iYg8IybED6/dOAJf3zsBF7k5ORwUFyK5yiUuZ3D7/KLyEJPFKpntHhfqSN13H5kfnhQKrUaN128aIZlE8bfvDztTcUsMRskf1vYMWgD2nhzXjU6UZFKIa4Qn9VMq5fCTBDWeWn5EMqnDQa2yj0sFpB9uAftVWHHKvKvmGqg6uH6IE3fn16pV+PTOsfi/iwdgVloc3r91NIb2cv9BJSxAh5vGyp98OHqAzJ3Q29kszM9Hjf/dMBw3iE5IHcIDfAQnEnI9WhwcmRZ+PhoMFX2Y+kL0ATs1OlCQPeNKPJoXsGdaWKw2Sa8WcTbLgdwqHHZTE1xd3yT5UOuquRPrgbHBuH6M/PdJzr0X9MVMl34oVwzvhZvGJklO/h0UJ+LIZBCN7RMh+D212oCtmdKrYk0y7wczBiv1abFPxHENGNw8LgkDYoMlH9blvp97clqWaeFK3M/CYe6Z3z93+kYHCoJqLTWgFX1aAGmwd99p+0mk3P//i525uOT1zbjlw5245p1tbsslHKM/v9yZK0n7dsgoqUXVmWCDuLzsSEGN4lhqhwWrjuGC//6OUQvW4u4le/Dy6uOSq+pPzhoEH4WJMWkJoVh81zjZdHSlSTP9ot0H4vrHBAv+5oUH6iS9YjafLEOTm/dnpSbSYjHBvugvE7S4dbzw+CqsbsBWmclSrif0Di25Ii7OrFCrVZIGs++JGib7+ahxoUIQ2eGbPcL31JYe01FBvvDz0UjK+BzB86KaBtz8wU58vDUb//zxKGa+tklw0aLBZMHdS/ZIJlrllJ892d2WWYF5n+7Bz0eKcKxI78xGqapvwvMr0537fbIt2zm9qN5kwSu/CINpSswWqyBoJx4LHi7KsnAVGeSLz/803u3fICVxIX5u36eBs0ELALh7Sl9JNoQ4gOpKXFqkxNH4WmyM6OeaW1mPvMp6bHIpzzyUXyP7+5NTIR2925L+Rt1VuwUtFi9ejMZG5Sswr732GlavXg0ASElJwZQpU9prOUREraLTqiVXGpWufruKDpZe7T8tipY7Mi0GxQdLSioA+xV8R4ZFanQQnr1c2Dm+os6Ev59Jlc8XpXH6+2halBHSHh64MBWT+0UhPMAH8yb1wUUDlb9nF4h6dhhFI0EHxgbjiuG9cP+FqVjz6AXOxpWJYdIAREyIryBlX6wlmRaAsK+F+EpPSlQg/Hw0eGhaf7x762hcPDRO/HCJu6ekyJYhOLJSRvUOx+q/TMELVw7FTw9NxoS+kZg6IFo2ADMgVliWEOirlTRGdXDNIBglOskTj9W8YngCVj0yGX+Z3h+3jO+NeZP64M8X9MVfLxmInx46X3KFvVTfiIpaabBslqj56MG8ahzOr5ZdH6BcIiIueRJ7+rLBiqnuYhq1Cu/fOhqfzBuDT+8cizduGgGVSoXrRifKjtHtFxPktjxksExzWT8fjeQESK5vyKmyWsno20sUjqPYED+oVCosvmssXr9xBD64bTReuCINGrVKEoiTy545kFclOdFsrk+LuJ+Fw4ikMMwe5n6tqdFBCPX3cdsLxB3XE1h3DX2Bs8e0eIJQTkU9Sg2N2CrTzBewv/9uP1WBQoXMIEfzv+amuhzMr0Z5rVE2ff/zHcplzkU1DfjUpTRm3R+lWCy60ju2T7his2GHUb3D8fG8sfB1CTT4+2jw+Z/kJ2cB9owWd99f15M7h5midRgazbIBA0CaPaQ0+jwmxA+De4UIgnyh/j6SbCWrDc5xy672nkMgDpDPHpolOp4NojKplKggDG5Bw2Tx717/mLOPUTymzwTixvYRntzuPRO0WLYvX7Cm/KoG/GnJXsz9aCce/Go/rn5nm2xgstZodpZ8/aBQKnK82OAMxO3PrRbct+VkebPTgQD7z2jsv9bh/Fc24N2Np3Agt0oSDLjngr6yAXCHpIgAfP6ncYpjvOXINTs9+5z+CHHJ7NCoVXhkmnDak1JGXGuyh2JCfAU/c4e5okbQ9SYLVsk0VJebXCZ+n44L8RP8f3qadgtazJ8/HwkJCbj33nvx2WefYdu2bTh06BC2bt2Kd999F5MnT8bjjz8OANDpdPjggw+g0TTfwISIqKP887LBSI0OhK9WjXumpEga0bnjq9UopnsDcDYO9NGocZ7MiNXU6CBBGuXN45IkHyDXHy9FZqlB8uHZkULdGWJC/PDF3eNx4NmLMf+Koc2uQ9zXwlV4gA++u28i3rx5JJ68dJDgCq5cpkWsTMNPVy0NWvxRpHd+iBN/aBhwDleRe4X54wpRfXhMsK/gA9yguBDcPrGP8/n9fDSYOVR64iL3+u6yLVyDFuKreGKT+0chKsgXj80cgJeuHob5VwzFU7MH48GL+qFvdJCk7KZE3yhpLqtVqzBjsDBIdbK0FgdFH4RdbTlZ7rZ2Oq9SeFz3jQ5Ev5ggRAXp8NSsQZImtS2hVqswbVAsLhoY4zw2e4X5S4JnieH+CPTVyp5sqFRwO/VF/DybMsqcVx+LaxqxeFs2nhQ1vEwI85c0V3TlyNzy1Wpw1cgEXDw0DuozUZbkFhzTjU1WwVU8m82GjGb6tCg1u/3rJYPc9gJxNDwdIPP9UZrG4vq7rdTTwnGCNyguWFJ2tXhbjmTSUGsU19gzMJprdHcgt1q2iS8ArDpchAqF0Z7Hiw3NjtV9evbgFr9/T+gbia/umYBxKREYkxyOL+4eh/F9IyXlDA79YoLQ100PHbmyntgQPwwVBTPkerVU1pkkpVuXpikF4nwR4ueDZy4fAp1WjRA/LV6+Zhgig3wljVjFWUnAuWVa+GhUkv4+AHDZsHjFoEJqdCD6RAUqZrDIEZQ8KfQeSnQTiMsoqUV1vcltH5FtmRX4+XCRJMPClSMTTqlnAmAPbBrNFvwh02unuUCcobEJz/+Ujqr6JhTWNOLfvx7Hje/vFOwTEajDHRP7KD4PYH8fWHynMINIp1HjvVtHuw1Qp0ZLm506DJE5psf0CRcE+qw2ezNbOeLyGHEWrfA+e6aF60Wr2BBfzEqLlwQRl+6VButlgxai9+n+Pbg0BGjn8pDKykp8+OGHuOOOOzB58mSMGDECU6ZMwQMPPIBt27YBsI8tXbVqFWbMmNGeSyEiarUBscFY9/hUHHruYvzjsiGteqzSHzfgbKYFIH9CKQ5kqFQq/Pva8yRXIQ7kVssGLbqKMX3CJaUvDo9fPNBt7wq5Uo9ohX4WgPugRVSQTrIGRxd58YeGc03NfODCVMEHJaUP9A5zzpP2yhgg86FF7gNbWICPoNeDUtAi2E+L4TKBM1exorTdEr10Ik5siB/SEkIFJQs2GyRjfMXfa7kPcIA00+LK4QlY9/hU7P3nzGabb7bWbaKyB8cEjIQwf0kJRp/IQNksBACSQEqJ3oiMklos35+PKf/ZgPkrj0lOeIf0CkFssJ/bkyLx995VskJ5kCvXK9MVdSZUi07u/zK9v3O8sr+PBreMly9pAuzZRje76bfg6IkwVnQCdvGQWPyfqJ7cQadRC+rZlUqeHA2GtRq1ZMywOGMhIczfbU+dqCCdJPBSom+E1WpDSY0w6CC+An4gt8ptGYjJYsW3bo5nQJoVJ3b5efEYKXNirWR0cji++/NEfH//JGe54TWjEmT3VerTIneCB0CSLbfxuDSbZYsow8VXq8aFA90HFR3H9NzxyfjjhUux958zMXuYPUurJQ2TC6obJP10lJoCA/bsIX+ZchqtRo2/zOgv8wi71Ogg+GjUbk+M3RFMeVIoX3AE4tJ6hQr+RgD2cqcDLSg5cqfkzPeoJYG49EK9bGnm9/vyZScQOWSUGCTZKeLnufeCvi3qPQTY+4l9fe8ETOwbidHJ4Vhy1zhcmhaHC90EqZUyLYbES/+u+floJBegxMcvYO8ncbxI+Pd/VjOBOJ1WjddvHIFBccEYEh+CN24aCT8fDZJEn8myZCbg7M2pkvTEEV80kcvk6EnaLWixZs0a/O9//8M111yD8847D7GxsdBqtQgODkZqaiquvfZafPrppzhx4gRmzpzZXssgImoTlUolO86yOXJjPh38fNSCk3G5E8oRMo1AIwJ1knrMY0V6ycz39pwc4mlyKfWAPbX4Zje9BgAgSeb/2FymRa8wf9mrNRNTozBG1FvEMQ7SE5kWgL2L/CfzxuKyYfG494K+eOJiaVNIsfP7RUkyduQafcl9YBM3l4wP9Xebsj+xb6RzcoM70jG+jSgWfRCOD/WDn49GsRFjVJAvbhD1ofhoa7ZsCrI4A0Muu8ZTpg+OxaMz+iM2xBcXDIh21vJrNWpJsEuuNMQhNTpQ8r3/fGcOnv7hiNvGp1MHREOtVrkNNra1uSwgbC4rPqZ9tWo8Mr0/vr9vEubPGYK1j1+AXgqjmAHgken9ZRvMOk7sbp2Q7KzlHp0cjpevGYZL0+JkG+b1jQ4UHH/+Oo3bY1WQPSQ6ua8XBccuTYvDTw9NxuI7x+Lp2YPw7OVDsODKoXjtxuH49dELMEk0mrVY34iKOpPkhMtxMu1wMK8aBxSyh77cmSvbVA8A8hQmQoT4afHkpYPc3t8aFwyIlmQsqFT2E3B343/dBi1EZZEnzkwEAux9LP60eA/+8s1BwT6D40MU+zi4vp9o1CrBlWiloJWrvafPHtMWq03yHiIOAChN/rr8vF6yPTYAIPXMdnEATKNWKQb3+rmcYLZkNLVOq5b83X9zw0m4nsf6atVuRwLLlV4U6xths9kkAWZpIK7abUZcrdGMZTIlOg65zQTiIgJ1kqBwc85LtAcult0/yflzE09YcegX7T57SK7kCQCmiH73N2eUSwIG6YU1kveCS0Xlj64cgbjz+0Xh10cvwOq/THFmrLUkuFxRZ5JM7GKmhVDrWw+30MCBAzFw4EBnCQgRUU+ilGkRHyos35BLWT0vMUz2sUN7hWJN+tkmXOmFemhEqcRdKdMCsE8R2SSq/X/m8iGKJ9Jy/0elkzvAXorTK8xPUnYwPiUCNQ1NgqaJO05VoKahCcWiRpNt+dBwfr8onN+vZc1cAfuH2OfmDMFj3x4CAEwbFCMbzJL7kC83EWNkcjgKZWppxVNc5MSIrvaX6o0oEn1vHCVPI5LCcNjNxJDzEkNx09je+Mwl5bjeZMEjXx/A8gcmwVdrPxG22WySDKKWNlI9V4/OGCBpPAjYA1WuHyaVGq+qVCpcMCAKX+8+e7Vd3PTUwc9HjVvGJeP6MfYP44kRAZIrcMG+WrdZHUDLrkoD9vp4m80GlUolaS6bGh0EjVqF0cnhzZYROUQH++KeKX0FIw2TIvydaw3198HS+yaiockiWP8F/aOxXpTuLpe9lBIdKNt/wvX3XjxBRGzqgGjotGpcODBGMObaIU70flGqb5RckdaoVbh4aByeX3nMuc3QaBb0vRErqG7AhuOlknI+QBqIu3VCb/SPCUZhdQOuHpXgsWPcR6PGFSN6CUZGJ4UHwM9HA1+tGoE6jSADKiHMH6FuShpHJIUhLMBHkJ2z8UyJyDM/HpUtdxnVOxyh/j4I9tVKrsIDnsseuvxMNlpuZb2kT8zrN47AE0sPod5kQXiAD25XKE/QqFV4dMYAPPjVfsl9jtKRqQOj8aPLNKPbJybj/qmp+Hp3LsRDYxyTQxxCA3wQGaiTHUvr+l49tk+EYDKJ+G/VlP7RePX68/DzkSJkldUhQKdBgE6LUH8fTB0Yjb8vOyyY6FOsb0RVfZOkV9SstHhBH4yDedWKJa1LduTg9onJsmVLuRXKWRx/bkWWhZLpg2MQ6u8jGaHcPzYY7oqp3AYtBkQBq8/eLqhuQE5FPVKiAvFHkR4fb83GT4eEk5h6RwRgaAvK+OS0NLi8N6fS+X7YZLEiW/T3wF1grafo9OkhRETdkVKpgrhDdnigTjA9YEBskGTMqYO4vviPQr3kSkdXyrQA7Fe5XK/azkqLa3ZSi9z/0V1DSldyHx4m9I2U1PCfKDFgt2isnVatavFVQE+5emQitv19GpbdPwkf3zFG9kOjXNpygkxQR3xl2mGyQl8Rh1jRB7ISfaMkPdtxdV4usOIwLCEUQ3qFSK5Qphfq8Z9fz46TLDMYJR+05bJrOsJtE5KdpQSh/j7OIIM7Sn1aAHvpyXu3jsaBZy7Gs3OGOAM14hRioPmSJ3cneOKMoqr6JueVaHEavVzJUUvcc0FfwdX5P52fIrhfpVJJAi7i3i72129ZnxadRo1ol74YI3uHwV3rBz8ftSQrTUx84lysb0ShaHR0bLAvEsL8BeV8ACSZFOIsoEUbTkpOogHpVenB8SG4Y1IfPDV7sOxEmra4dUKyoBRrznD7VWKVSiW5Mq2UHaVRqyRlTx9vzcazK9JlAxa9IwJw1+Q+UKlUsu9DgHKAuU9Uy37P9+eePekW97OIDNRh1rB4rHn0Anx4+xhs+ttFsg2yXc1Ki5MtJ3K8v14xPAH3XtAXfaMCccv43nhq1mDEhPhhgkymoNwx7S7zxPV7NKaPctBw2qAYhAXoMHd8Mp65fAieuHgg7r8wFbeM742EMH9pGV+NNBCnVgGXiMocao1mrHOZSCKWVVYnOwkJkB7TKVGBzsyZcSkRbieGtJavVoMrRe8fEYE6RATqEB6okwRdlBoCD4wNlhwPW06WYenePFz+1lZ8vy9f8vs7LDEUIX4+boM74sC+q5YGl10z4k5X1Euy83ry5BCgHTMtiIh6MqXyEPEHYAB446YRWLIjB41NVtw+Mdlt0ynxlQOD0Sy5ktWeafTtIS7UD5/MG4vPd55GYrg/HpO52i3mr9MgPtRPkPbakmBN74gAbMPZq6RRQb5IjQ6E2WpDgE4jSDF/Y71wZFsflw9jHSkhzF82c8IhKSIAahUEJxBy+8tdRU8I8xf0E3BH/GG4zmRBZqm0szmgHLQYnmQPxj1z2RDsya4UdPz/eGs2JveLwkWDYiT9LHQadbN9YtrLpH5R+OUvU3CsSI/z+0UpNpR07K9Rq2RLBBLC/PH+baNlrzzKBdSaK3lKivCHSgXJld5xfSKQU1En+P3Yk1OFfjHBMinH51byFOSrxTd/noBtJ8sRHewraSQoZ8bgWPj5qAUjRVt6gtcrzM/ZgBQAQvx8MDA2WLYR4fiUyGbL+uT6tEhKns78Ho3sHYaiI8WyzxMT7IuHp/XH374/7Nx2KL8Gz/10FC9dPUwQaJSUPLVjIC41Oggf3T4WS/floV90EO6afDaoNDAuWDDO0V2Q3GHaoBiscMkyyCqT1uRHBupw39RU3Doh2dk7IikiQPLz0ahViAx0/zvU0kyL9EI96k1mBOi0kvHIjoy4pIiAFmevqNUqPDZzAP78+T7ntgGxQc7Am0atwtOzB+Pp2YMFj7tiRC9J5o1c74E+UYHYK9NA1DVoMSo5XPb32eGiQcoBUfEFEfsxLQzERZ8JxCWE+Quap4rH/IqzGt75/RTOT40S/A4C0t5Dc8f3xk3jeqOgqgH9Y4Ik+7fFzeN646tduc6pVa4T3fpGBQqmnwyJD3Hb0FalUmFK/ygs3392qspXu3KRVSYdMQrY//7cfqbEpXdEAKrrpZmEShdNWp4Rd/b4EGfERQf7Iiygc6bCeQtmWhARtQOlE6w4maBFoK8WD1zYD4/PHKB4UhQX4tfsONOulmkB2LMd3r5lFJ6aNbjFPURc030HxQW7zSRwJR7Dd36/SKhUKvho1JgkqnkWjzs71yvS7c1Xq5Gc8Mp98B8cHyKp857cL6pFkwrkroyeEJ2MOIJxKVGBbhuoOnpy+Os0eOuWkZIg0F+/P4zGJoskLToh3N+jH35bq39sMK4ckdBswAKwf9gf6SZw8+9rz3ObKi13ctVcyZOvVoN4mSt84/tGSIIIju704mBTW67ehfj5YNaw+BYFLAD7+9zdk/s6b/cK9ZNt2Nji7CE35SwtmSwjDVpIJ+I4jumRSe7fW4YnheGK4b0kv4Nf787DF7vOlgbVNDRB3ygOMLfve/Xk/lF446aReHh6f8Fxd8fEPs4JDVFBOtw8zn3/IMCePaT0NnHHxGRs/ttFuOeCvoJml3JBmaggneKoYndB1IhA4eMsVpuzIao0e+jcAnEXD4nFrWdGVAb7tqzHyKy0OElTV7kyQrlAXLCfVjDCMsTPx23GzeD4EMSHKl+QiA2VyR6qEZfxnQ3EuePvo8HjM4UXD3ZkVWDh2gzJvuJAXGJ4AIJ8tRgYF+zx9+zB8SF47cYRGNsnHNeOSsRfLz3bG0r8M1eayARIM+KOFxskPSx0GjWuGZmAnx4+39m8093vrFImT++IlgXicivrnRNfpE04vfPzR0di0IKIqB0o97RQvnqqRKVSSUpEXAXoNK2ec95V3X9hKr66ezzeuGkEfnjg/BZ9QLp6ZKLzj39koA4PT+vnvO++ZiZS9PPizt1zx59tdNYr1E+2T4VOq5Z0TZ82WFrrLydAp0Ww6GTbLLoi5QjGqVQq2WyL+FA/QQbSoLgQ/PMy4VXL8lojtp8ql2ku27Wyh8SjTwH7VUKlsifZ5rIKKccOvWVO8sanREqa7e3JqURlnQnltcK6+o7+MPzYzAH43/XD8fjMAfj+/kmyQUq5aQty2UPu0umnKkyucBBflS41GJEvGg3pDFoonOCNSAqDn48G7946SjIZ5/mf0rHzzFV48cmdSiX/f+oIwxJDse6Jqfh03lisf/xCxdR2wF7C6C4QN31QDJ6bM1Q2GCc7mrqZ1woL0MkGPSelRkqahe47c2VaXB5yrse0SqXCi1cNw4FnZmLbU9Mw3c3oWPF6XcuefLVqycQVALJjVeV+/uLfW4dpzWRZAHKZFtKGyb2cx7T7QNywxFBcNzpRciK+6PdMrDp8NuOmscki6fvU0v4N52rO8F5Yet8k/O+G4YK/J3PHJzuD8gE6jeBvopzm+ktdOyoR25+ahoU3jhAEkuTep8MDfJxlfnIcGXFiof4+CPYT/t44gssZpQxaiDFoQUTUDpQ+BMY1c7WkOe6aSwH2k7uWXDnvLib1i8KVIxJkR9nJCQ3wwcqHJ+OXv0zBpr9dJAhEjOkToTjSzFszLQDg7ikp+PTOsXjxqjSs/ssUt9kqf790kDOgdvl58bhYplmgO81d9Xe9CigXtBCP8QXs/SLE2w/n10gyLdr7irSnXTw0VvAhtVeoH56erXzVVu7DfktKYpJFV/F0Wvs4UPHY0bzKBklduk5mOkp706hVuHZ0Ih6Z3t/tlJLEcOmkn4Qw6TpH95ZmeCSG+8ueIIqJ07ktVhvSC4Rp345jOi0hVHI13cFxrA/tFYpXrx8uuM9steHhrw+gsckiCcTFh/h1SrmZ8/VD/XHRoBi3DTjF5E7EU6ICsfDGEW4DxnIneEqlkw5y6fTjz4zAdLUvtwpmi1XSwPZcS54cwgN1ggyI5rxwZRpun5iMCwZE471bR8tmU6bIZA/JBWPdZS1NG9R8gFkuEFcgamgc14JA3MikMAT6avHWzSMlx/1flx5GeqH996SgukFSytJZ5anDEkPx22MX4M2bR2L9E1PRp5n3gOhgX7e9XNISQvDyNcNkM+tky/iaCcS5y4gblxIhOab3ZDsml4nGrbfxmO4OGLQgImoH7ZVpAShPL+iKpSEdzTGWU25E3JOXDpIdywice8pxR1CpVLhoYAxunZCsWPc6pFcItv99Gvb8YwYW3TKqVQEupQ9mGrVKcFVuhMwHYrmJOCqVSnJyfSS/RlIn3VlNOM/VoLgQPDVrEIL9tOgfE4SP7hiL4GZOgkIDpFfdmrsCDkhLKRxX/gfEBkue792NpySPbW7cbWfw0aiRLDo5kDsZSorwl1wNnjogukXHdWSQryQwklMhPO56hdm//34+GtlgsUplP1lyuPy8XnjwImHGVpnBiI0nyiSBuMQuF4gTBnQDdBq8f9tot6VggHywsSUNk+XK2yakREgya/afrkJOhXRySEe/Vwf5avHClWn47K5xkhGxDi2d8iSXaREe4IMRCiVK/9/encdHVd/7H39PMtlXIGwhYRGIgLIVUBABUcFStIp7W6vi2mpb29prq7Z6a1u19udte3tbW3e9vRaXWhesVdsqKqII4grIjgkESdgSsmdyfn98mSRzZskkmcmck7yejwcPMmdmznwTTg5z3vP5fr5+g/OCg7iPgoI4c0wfU5ir1DC/+/4gbuZRA3TrGRMC7qtr8ukbf16rhmZfUPVQ/6zUDs9z8TRiQJa+PLmww2k0fnNDVL5lpCTrvy8MnrroF+o8FM15OlRF3MyjBgT9/7d6R5ggjkoLQgsAiIesNG/rnGG7UJ/CdIa9RLY9t5XRO83Igix9febIoO2JWDkkXrzJSR120g8lUhA3KCfwAnBKiIAiXLO/oEqLXSFCC5c1l5Wkq+aO1oe3LtQr358XsTqqvfar2Hg80hcifBrqd/rkQmW0q6y5bPZISSZIOt62ssGG8sA+Ld39RDqe2pfcZ6Ymh/yk2ePx6LRjAquF/MtgdiQ5yROwGkko7aviQk2PGD0wO+gT+esXHK3JQdVDB4NWWXBbEHf0kBxdd8pYeZM86peZoj9eNK3DcCDk0tRRVFrY+1oMyErVmEHZQZ9KV9U368WPApdxLshO7bDvUyKkpyQHhRSh+rQMzQtuvDyvZGDEPiB+A7KCg7ittqap/gv6NG9y2CU824fOF80cEdTzpHR/nV7duDdEY1l3nafnhFjp6adfPiZodZ32uloRF+r9w8yj+geFFhv3VGl9eVVQEEdoweohABA3g3LTg9bZTk1OUv9udoAeVZCljJRk1TX5gu4jtOi+75wyRk+tLQ1ompeolUOcJFKlhT2I65eVqhNGD9BbW818/oLs1LD9B+wVGBXVDUGPcdsFnl9np2r9ePF41TY2q/xgva6ae1RUlVPD8jP04nVz9Oqne3VMYV7AUp9XzT0q4lKGTn4j/O2Tx2pQTrp27qvRedOLwlYQXb/gaO2vadQnu6v0leOGa9bo4CUowxmclx40J7+99ksmThmer0dW7Qy4f3KIcC4pyaMTxhTog7K2T7g/KDuoFNun2m4M4r63oETfPGm00rxJUR3bWWleDchK1b6atj4q0VRafMEWTsw72lTP+C/o26968cDK7QGPdfKykNNG9AsYe7jqiYXHDNZDK3e03j5zyrCo9p+c5NGgnLSghrLtta/0nFrcT+varbghmX+f9pUKHo9HP/3ysXq/9FBA6Lnus4NqsexL/7rrPD1r9AAdOyy3ten2OV8o6nBJ68L8jKDVuqJabt0WxOWmezVuSK6afC1KTU5qbQJqWWYlk/b6Z6VqQBRNoHs7QgsAiJOBOWlBocWQvPRud9ROTvJo/NCcgOW9/Jge0n35man6zilj9fMXNrRua38h2FdFKoENNeXp7vMn69evbFJ1fbOuOWlM69KBdiP6Zyon3atq28oK7bntzXBXjRiQpf+7YmannzeyIEtLC0YFbT9uVH+dOaUwYLnK9pwcWiQnefTV44d3+Lh+Wan6w9emdek1Bkf4hNSb5AmY0x5qBZFQ06Akhai0OBRU3dTTvURiJdrVnfymDu8XEJxNDNHbxm7e2IH6ynHD9fR7ZRo3JEfXL2xbJWL2mAF6Yk1Z6+2DtU0Bz3XyNL7vLyjRtsrD2rq3Rl89fnjYppvfPbVEew7V68OyQzp3WlHYKSehDM5NjxhatA+Ypw7Pl1YG3h+qH1GqN0mnjh8UEFq8X3pQ+bZ+KG47ppOTPFp21Sw9/8Fu5Wek6IvHDukwjEtJTtLQvMDgLJqGyfaKipPHDVJykkfJScmaVJQXsBzu42tKAx7r5PN0TyK0AIA4CVUy2N2pIX7HFOaFCS3c9+mdEy2dPUrbKmv0zLpdmjA0V9+aP6bjJ/VyEZfxzQ1d5nzXuZNDPDpQUpJHE4fltVZl2PWlFXHi4cZF4/XK+s9V2xhcmeXk6SE9IdL5eHBuYMA8YkCmBuem6fOqtkqgcBed9uqh6vrmoFCurwRxN3zxaJXur1XpgVpdNntUxJ5MfklJHt1x9kTdcfbEoPuuOWmMnvtgt+qbWkI809nH9MiCLC3/9pwOH5eXkaJ7LupaEGdvxtmexxN4gR2qGWe46g97VdFHuw4FhRRuCy0k04/kK8d1HI62Z6+YCVVxZTd9RD99a/4YLXv3M40emK0b2i2nu2ji0IDQwt7cNNQSun0RoQUAxEmovgHdbcLpF27ZU7eW0TtNcpJHty+ZqF+cdaykzpf590aRPk3q7nE9qSg/bGhR3C+Tn383DMlL13dOGas7X9wYsD0l2RNylYa+JNIx7W/C6efxePSjReN0w1Mfqsln6bxpRQFLIbY3NC9dBdlpqjwcPNXJr6+cq0sG5+il782N2f5GFmTpP04bp58tXx/69fr4p9KRgriB2WkB05SG5WdoaF5gZUa4aXyTbRUYtY0+bdwTuMJFXwnirl9Yos/212p7ZY0umjk85MpYdh6PRz847Wj94LSjg+77+swR+svqz7TFtsyp31gHL7fekwgtACBOQjUci1WlRajGflmpyUHlmugeLpbbRJq3OzS/u6FF+Dd9bpz77zSXzR6lJ94tDehIP6ogK6jPQl8TOYgLPu6WTC3S7NEFOtzQHLFZn8fj0eSiPP1r496Q96d6k6Jq3ofQlp4wUv/4uFzv7jgQdJ+TKy16QsRj2tbg0+Px6PsLSvTDv36oFks66eiBmj4idGgxMCctqJ+InRsrLbpixIAsPXPt7JjtL9WbpDvOnqjz/rgq5P1MDzH69v9WABBHod6UhlqruytKBucEdQkv4hNpxFGkrv/dr7RgGd94SvUm6adnHqP2p4fTbEtY9kURg7gwx/Sg3PSIgYVfqCV+/YryM7rd26gvS0ry6K5zJyvN1hzZqSuH9KSIx3SI9x/nTS/Wv68/SX+75gQ9eMmMiO8hwvVwkUx1YqwqSfuiGSP7B63S4jeG6SGSCC0AIG4GhXjzMCTK9cM7kp6SHJS+088C8ZSRmqyc9NAFmt09roflZ4S92OgrJcfxNmfsQP3hq1/QqeMH6RvzRusb80YnekgJF2n+f3cvwCYXRwjiOKa7bVRBlv7DVmpvX+K3L4p0TIer9BxZkKWpw/t1GKSFWsrab1h+hrx9vHKru370xfEBzX8l09+ko6WZ+wqOLgCIk1CfTMfykwh7QzP7klpArIUqPfZ4olunPhKPxxO22qKYMC5mFk0cqvsvmaEfLRqnrDRmCEdcESe/e8ddpEqL4Ux5iomls0fpihNHKTM1WROG5ur7C0sSPaSEGxzhPYa9T0tn2ftatMc0vu7Ly0zRLWdMCNh2wugBVNAeQWgBAHEyND9d9g8uhsXwAuyrxxe3lnt7kzxaMjW6tdyBrgoVTtibu3XVpGFhQgs+lUac5KZ7lRFmCc/uBsz9s1LDXsj1lSac8Zac5NGPT5+g9bd9US9850SNjmLaTm8XudKie+8/jh2WGzQt1a+v9LOItzMmDdWNi8ZpWH6GjhvZXz9aNK7jJ/URxOwAECe56Sk6fVKhnvtgtyTptGMGB5X+dce0Ef313LUn6p3t+zRr9IColpIDuiNUpUWsqofCfTLNtCfEi8fj0eDcNO3YVxt0X6hGnJ01qShfpfuDGxcSxMUen0YbWWle5aR5Vd3QHHRfd8/VmalelQzO0YbyqqD7OKZjw+Px6Op5o3U10/eCEFoAQBz9+oIpWnjMYPlaLC2eODTm+59YlKeJUSy3BcRC6D4tsQktQh3H+ZkpyklnRRzEz+Dc9KDQIjU5SQNi0NBxclGeXviwPGg7lRaIp8F56aoOsXxmLALmKcX5IUMLKi0Qb0wPAYA4Sk7y6PRJhTpzyjCaVMH1QvdpiU0lxODc9KDO91zcId5CVQ8NzkuLyeoe4aqHmP+PeAq1gojpPRSL0CJc7yHO1Ygv3kEDAICohHozHKtKCyn4Io+LO8RbqOM3VkHcscPyZJ+1kJPuVV4G1UOIn1BBXEF2mlK93b/sm1LcL+R2Ki0Qb4QWAAAgKvHsaSFJJx09MOD2cSP7x2zfQCihmssWxuiYzk7zaoytOWRxv0z6LyCuQjXjjNUxPWZQtrJSA5vX5qR5lZ9JEIf4IrQAAABRCXWBF6lbfWedN61YXz1+uIblZ+i8aUW6YMbwmO0bCCVUpUV3V1loj+oh9LTQx3RsztPJSZ6g/kPF/QniEH804gQAAFEZnJuulGSPmnxW67ZYdo1P9Sbp9iUTY7Y/oCOhqocK82MXxH1p4hD99b2y1tsnHT0oZvsGQgldERe7sGxKcT+9vW1/6+2RBUwNQfxRaQEAAKKSnpKsc6cVt94+dfwgFebzyTHcK1SlUCwv8E4eN0i3nXmM5owt0A8WlujcaUUx2zcQSuhjOnZB3IUzipV6pLG4x2Mq5IB4o9ICAABE7fYlx+rkcYPU2Nyi044ZnOjhAN0yMCdNHo9ktRUPxfQCz+Px6OJZI3XxrJEx2ycQSahKi1g2TB5ZkKV/fHeO/r1xr6YU52s6vYfQAwgtAABA1DwejxZMIKxA75CekqyTSgbq1U8rJJmeE+OH5iZ4VEDXFWSnKjnJI19LWxIX64q4owZm6yhbk1kgnggtAAAA0Gf95oKp+sOKLapr9OnKOUcpOYmmgnAvb3KS5o4taA3iBmSlapKteSbgNoQWAAAA6LPyMlN046LxiR4GEDO/Om+yfv3KJlXVN+ub80YrzZvc8ZMAByO0AAAAAIBeoiA7Tb9gJSb0IqweAgAAAAAAHInQAgAAAAAAOBKhBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsAAAAAAOBIhBYAAAAAAMCRCC0AAAAAAIAjEVoAAAAAAABHIrQAAAAAAACORGgBAAAAAAAcidACAAAAAAA4EqEFAAAAAABwJEILAAAAAADgSIQWAAAAAADAkQgtAAAAAACAIxFaAAAAAAAARyK0AAAAAAAAjkRoAQAAAAAAHInQAgAAAAAAOBKhBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsAAAAAAOBIhBYAAAAAAMCRCC0AAAAAAIAjEVoAAAAAAABHIrQAAAAAAACORGgBAAAAAAAcidACAAAAAAA4EqEFAAAAAABwJEILAAAAAADgSIQWAAAAAADAkQgtAAAAAACAIxFaAAAAAAAARyK0AAAAAAAAjkRoAQAAAAAAHInQAgAAAAAAOBKhBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsAAAAAAOBIhBYAAAAAAMCRCC0AAAAAAIAjEVoAAAAAAABHIrQAAAAAAACORGgBAAAAAAAcidACAAAAAAA4EqEFAAAAAABwJEILAAAAAADgSIQWAAAAAADAkeIaWqxZs0a33XabFi5cqKKiIqWlpSk7O1slJSVaunSp3nzzzXi+PAAAAAAAcDFvvHY8d+5cvfHGG0HbGxsbtXnzZm3evFkPP/ywLr74Yt13331KTU2N11AAAAAAAIALxS202L17tySpsLBQ5513nubMmaPhw4fL5/Np1apVuvvuu7Vr1y49+uijampq0mOPPRavoQAAAAAAABfyWJZlxWPHp59+ui6++GKdc845Sk5ODrq/srJSs2fP1qZNmyRJK1as0Ny5c2M6hrKyMhUXF0uSSktLVVRUFNP9AwAAAACA+F1/x62nxfLly3X++eeHDCwkqaCgQHfffXfr7aeeeipeQwEAAAAAAC6U0NVD5s+f3/r11q1bEzgSAAAAAADgNAkNLRoaGlq/DleRAQAAAAAA+qa4NeKMxooVK1q/Hj9+fKefX1ZWFvH+8vLyTu8TAAAAAAA4Q8JCi5aWFt15552tt88///xO78Pf5AMAAAAAAPQ+CZse8utf/1qrV6+WJJ199tmaNm1aooYCAAAAAAAcKCGVFitWrNCPfvQjSdKgQYN0zz33dGk/paWlEe8vLy/Xcccd16V9AwAAAACAxOrx0OKTTz7RkiVL1NzcrPT0dD355JMaNGhQl/YVq3VfAQAAAACA8/To9JDt27dr4cKFOnDggJKTk7Vs2TLNnTu3J4cAAAAAAABcosdCi927d+vUU0/V7t275fF49OCDD+rMM8/sqZcHAAAAAAAu0yOhRWVlpRYsWKBt27ZJkn73u9/p4osv7omXBgAAAAAALhX30OLQoUM67bTTtH79eknSnXfeqWuvvTbeLwsAAAAAAFwurqFFbW2tFi9erPfee0+SdPPNN+uHP/xhPF8SAAAAAAD0EnELLRobG7VkyRKtXLlSknTdddfp5z//ebxeDgAAAAAA9DJxW/L0K1/5il5++WVJ0sknn6zLL79cH3/8cdjHp6amqqSkJF7DAQAAAAAALhO30OLpp59u/frf//63Jk2aFPHxI0aM0I4dO+I1HAAAAAAA4DI9tuQpAAAAAABAZ8St0sKyrHjtGgAAAAAA9AFUWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsAAAAAAOBIhBYA0Nv5mqXXfindO1966WapuSHRIwIAAACi4k30AAAAcfbp36XXbjdf735PGjBGmr40sWMCAAAAokClBQD0dqXvRL4NAAAAOBShBQD0dk21gbd9jYkZBwAAANBJhBYA0NvZe1j4mhIzDgAAAKCTCC0AoLdrqgu8TWgBAAAAlyC0AIDezl5p0UJoAQAAAHcgtACA3q6ZSgsAAAC4E6GF09VUSh88Lu35KNEjAeBWTfWBtwktAAAA4BLeRA8AEdTsk/5nhlS3X/IkS19ZJpUsTPSoALhNsy20YHoIAAAAXIJKCyf76EkTWEiS5ZPWPZrY8QBwJ3toQaUFAAAAXILQwskqPw28Xf15YsYBwN1YPQQAAAAuRWjhZPu3Bd5uqE7MOAC4G6uHAAAAwKUILZwsKLSoSsw4ALgbq4cAAADApQgtnKq5QTpUFriNSgsAXWGvtCC0AAAAgEsQWjjVwc8kqyVwW0O11NIS+vEAEIplBfe0YHoIAAAAXILQwqnsU0MkSZbUeLjHhwLAxXyNkizbNkILAAAAuAOhhVOFDC3EFBEAnWNf7lQitAAAAIBrEFo4VdjQgmacADqhKURowfQQAAAAuAShhVNRaQEgFqi0AAAAgIsRWjhVuNCinkoLAJ0QKrSwfDT1BQAAgCsQWjiRr8msHhIK00MAdIZ95RA/pogAAADABQgtnOhQqdTSHPo+QgsAndHcEHo7U0QAAADgAoQWThRuaohETwsAndMcptLC19iz4wAAAAC6gNDCifZvD38foQWAzgi1eogUvpoLAAAAcBBCCyeKVGlBI04AnRGqEafE9BAAAAC4AqGFEzE9BECshA0tmB4CAAAA5yO0cKJ9W8Pf13Co58YBwP3ChRZMDwEAAIALEFo4TYtPOrAj/P1UWgDojHA9LZgeAgAAABcgtHCaQ2VSS4SLCXpaAOgMVg8BAACAixFaOE2kfhYSlRYAOqe5IfR2pocAAADABQgtnIbQAkAsNYWrtGB6CAAAAJyP0MJp7KFFXnHg7QamhwDohLCVFoQWAAAAcD5CC6fZvz3w9tDJgbebaiUfZd0AokRPCwAAALgYoYXT2Csthk4JfgzVFgCiFXb1EMJPAAAAOB+hhZO0tEgHOqi0kOhrASB6zWFCC6aHAAAAwAW8iR4A2rOk8x811Rb+P0MnS/KY+/yotAAQrXChBdNDAAAA4AKEFk6SlCyVnBa8PS1XajjUdptKCwDRCrt6CNNDAAAA4HxMD3GD9NzA24QWAKLF6iEAAABwMUILN0jLCbxdz/QQAFFi9RAAAAC4GKGFG6TZKy0ILQBEKVylBdNDAAAA4AKEFm5gr7QgtAAQrXA9LZgeAgAAABcgtHCDoNCCnhYAosTqIQAAAHAxQgs3sDfipKcFgGiFDS2YHgIAAADnI7RwAyotAHRVU5jQgukhAAAAcAFCCzdIywu8TWgBIBqWJfnCNeJkeggAAACcj9DCDWjECaArwk0NkSQflRYAAABwPkILNyC0ANAV4VYOkaQWeloAAADA+Qgt3IBGnAC6ojnM1BCJ6SEAAABwBUILN6ARJ4CuaI5QacH0EAAAALgAoYUbpNkqLZgeAiAakSotmB4CAAAAFyC0cAN7aOFrjHwxAgBS5J4WTA8BAACACxBauIG9p4XEFBEAHWP1EAAAALgcoYUb2HtaSFL9oZ4fBwB3iRRaMD0EAAAALkBo4QbedCnJG7iNSgsAHWmKVGnB9BAAAAA4H6GFG3g8NOME0HmsHgIAAACXI7RwC5Y9BdBZrB4CAAAAlyO0cAt7M05CCwAdYfUQAAAAuByhhVvYp4fUMz0EQAciVVowPQQAAAAuQGjhFvS0ANBZkXpaMD0EAAAALkBo4RZBPS0ILQB0gNVDAAAA4HKEFm5hDy2qdkuPf1367RTpjbsTMiQADtccKbRgeggAAACcz5voASBK9kacHz3Z9vW/bpOGnyCNmNWzYwLgbJFCC6aHAAAAwAWotHALe6WF3dZ/9cw4ALhHxEoLpocAAADA+Qgt3MLeiNOu4tOeGQcA94jY04LpIQAAAHA+Qgu3ILQA0FmsHgIAAACXI7RwC3tPC7v9W/nkFECg5obw9zE9BAAAAC5AaOEWHfW0aGmW9m/rmbEAcIemCJUWhJwAAABwAUILt+gotJCYIgIgUKRGnJZPamnpubEAAAAAXUBo4RYd9bSQCC0ABIoUWkhSC9UWAAAAcDZCC7fIGRIYXHiSpbELAx9TSWgBoJ1Iq4dITBEBAACA4xFauEVKhnTyT6TkVEke6fT/Cg4tKjYmZGgAHCpSI06JSgsAAAA4njfRA0AnHH+VNPFcE2CkZEjbXw+8v3Kz1OKTkpITMz4AzhJpyVOJSgsAAAA4HpUWbpPZ3wQWkjRwXOB9zfXSwc96fkwAnInpIQAAAHA5Qgs3yxooZfQL3Fa5KTFjAeA8NOIEAACAyxFauJnHIxUcHbiNvhYAJDNVrKNQgkoLAAAAOByhhdsNtIcWVFoAUMdVFhKhBQAAAByP0MLtgkILKi0AqON+FhLTQwAAAOB4hBZuZw8tKjdJlpWYsQBwjo5WDpGotAAAAIDjEVq4nb2nRUOVVF2emLEAcI7mhuBt3ozA24QWAAAAcDhvogeAbsorklKzpcbDbdv+9TNpz4dmTvvCn0tHL0rc+AAkRpO90sIjpWYGVmAwPQQAAAAOR6WF23k8UsHYwG0fPCZ9/rG0b4v0zDelhsOhnwug97I34kzJkJJTA7f5GntuPAAAAEAXEFr0BgPHhb+v7oC0a23PjQWAM9hDC2+alJQSuM3X3HPjAQAAALqA0KI3KCiJfP+uNT0zDgDOYV89xJshJdtmBDI9BAAAAA5HaNEbjF0Q+f5d7/XMOAA4R6hKC6aHAAAAwGUILXqDIROl038jDZkkTThTmnlt4P1MDwH6nlA9LZgeAgAAAJdh9ZDeYvpS80eSKjdLb/++7b7qcqlqt5RbmJixAeh59tVDvOmSrMBtTA8BAACAw1Fp0Rv1Hy2l5QVuo9oC6FuCpoekh5geQmgBAAAAZyO06I2SkqRhXwjcVkYzTqBPCZoekh5iegihBQAAAJyN0KK3GjYt8DaVFkDfElRpweohAAAAcB9Ci97KHlrsfl9q8SVkKAASIGjJ01CrhxBaAAAAwNniGlrs3btXy5cv1y233KJFixapoKBAHo9HHo9Hl156aTxfGvbpIY3VpkEngL4hqtVDCC0AAADgbHFdPWTw4MHx3D0iyRki5RZJVWVt23atkQaNS9yYAPScoOkhaUwPAQAAgOv02PSQ4cOHa+HChT31cpCkIvpaAH1W0PQQKi0AAADgPnGttLjllls0Y8YMzZgxQ4MHD9aOHTs0atSoeL4k2hs2TVr/bNttQgug72iuC7ydEmrJ08aeGw8AAADQBXENLX7605/Gc/foiL0Z5+efSE11Zm47gN6tuSHwtjc9xPSQ5p4bDwAAANAFrB7Smw2dInna/RO3NEvlHyZsOAB6UJOt0sKbzvQQAAAAuA6hRW+Wli0NtDXe3L4iMWMB0LPslRYpGUwPAQAAgOvEdXpIvJWVlUW8v7y8vIdG4mDDZ0p717fdXvuIdOL3g8vEAfQu9p4WIVcPYXoIAAAAnM3VV67FxcWJHoLzTblIWvNg2+2qMmnTP6TxpyduTADij9VDAAAA0AswPaS3K5omFU4N3PbufYkZC4Ce02wPLdKYHgIAAADXcXWlRWlpacT7y8vLddxxx/XQaBxsxpXSs9e03d72mlS5WSoYm7AhAYgze2iRksH0EAAAALiOq0OLoqKiRA/BHY49W3r5ZqnuQNu2d++XFv0ycWMCEF+hKi2YHgIAAACXYXpIX5CSIU39euC29x+TGmsSMx4A8ddsm/rhTWd6CAAAAFyH0KKvmHG5JE/b7YYq6cMnEjYcAHHmsy15mpwqJdsqLZgeAgAAAIcjtOgr+o2Uxi4M3PbBXxIyFABx1tISXEXhTZOSbDMCmR4CAAAAhyO06EumXRp4e/f7XLQAvVGoaR+sHgIAAAAXIrToS4bPDLzta5AqNiZmLADix96EU5KS05geAgAAANchtOhLMvtL+SMCt+1el5ixAIifkJUW6UwPAQAAgOvEdcnTN998U1u2bGm9XVlZ2fr1li1b9PDDDwc8/tJLL43ncCBJhVOkgzvbbu9+X/rCxcGPO7BT+ugJKadQmniuKS0H4A6hKi28qUwPAQAAgOvENbS4//779cgjj4S8b+XKlVq5cmXANkKLHjB0irT+2bbb5e8HP6bugPTI6dLBz8ztyk+lBbf1xOgAxIJ9uVPpyJKnTA8BAACAuzA9pK8pnBJ4e8/HwSXi7z3aFlhI0ur7pWbb8okAnMu+3Kk8ZmoI00MAAADgMnENLR5++GFZlhX1H/SAoVMCb9ubcbb4pHcfCHxMU4302aq4Dw1AjNinh3jTJI+H6SEAAABwHSot+prM/lL+8MBtu99v+3rzK4E9L9pvB+AO9ukh/p40TA8BAACAyxBa9EX2aov2fS1W3xv6OVv+Fa/RAIg1e6VF8pHQIskWWjA9BAAAAA5HaNEXFU4NvO1f9rRyi7Q1TDhRsUE6VBbfcQGIDfu0D2+6+dteacH0EAAAADgcoUVfFK4Z57v3R34eU0QAdwjqaXGkl4U9tLB8Ev2EAAAA4GCEFn1RqGacu9ZK7/9f4Hb7SgNb/hnXYQGIkaCeFkcqLezTQySmiAAAAMDRCC36olDNOJ+/TmqoarfBI829IfAx214LvhgC4Dz2JU+Tw1RaSEwRAQAAgKMRWvRV9mqL9sueSlLJadL0pYHbGg9Lpe/EdVgAYiDUkqdS6NCihUoLAAAAOBehRV9l72vRnidJmvMDKXtQcLixhb4WgOOFW/I05PQQlj0FAACAcxFa9FX2FUT8PEnSWX+UimeY22MXBN6/4XmpdLXU4ovv+AB0XbglT5keAgAAAJchtOir7BUUkuRJls6+T5p8Qdu2MbbQYv826YEF0l1HSf/+BU38ACcKWvKU6SEAAABwJ0KLviqzvzR8VtvtJK907oPSxHMDH1c0XcroF/z8+oPS63dJHz0V12EC6IJwPS2YHgIAAACXIbToy866R5pwljT6FOniZ6Vjzgp+TFKyNO9H4fex7dV4jQ5AV4XracH0EAAAALiMN9EDQAL1HyWd/0jHj5v5DalohrT+b6anxYEdbfdV7Y7b8AB0UdCSp0dCC4/HVFW1tKuuYHoIAAAAHIzQAtEpmmb+DJ0i/fXytu3V5QkbEoAwwk0PkcwUkfahBdNDAAAA4GBMD0Hn5BYG3q4qlywrMWMBEFq46SGSlJwaeB/TQwAAAOBghBbonJyhgbebaqSGqsSMBUBo4ZY8laRkW4Ed00MAAADgYIQW6Bx7aCGZagsAzhFuyVMpeAURli0GAACAgxFaoHNS0qWM/oHbqmnGCThKpJ4WQdNDCC0AAADgXIQW6LxQfS0AOEfEnhZMDwEAAIB7EFqg8+xTRKi0AJwl3JKnEtNDAAAA4CqEFui8XFtoQaUF4CzNttAioNKC0AIAAADuQWiBzsuxTQ+pJrQAHKUzoQXTQwAAAOBghBbovKBKC6aHAI5iDy2YHgIAAACXIrRA51FpATibvacF00MAAADgUoQW6Dx7pcXhvVz4AE4ScclTpocAAADAPQgt0Hn2SgtZ0uHPEzIUACFEWvKU6SEAAABwEUILdF5m/8A58hIriABOYVmRlzxleggAAABchNACnefxSDlDArdV04wTcISWZslqCdzG9BAAAAC4FKEFuibXNkWESgvAGewrh0hMDwEAAIBrEVqga3JszTiptACcIVRoEWl6CJUWAAAAcDBCC3QNlRaAM9n7WUgseQoAAADXIrRA1wRVWhBaAI5gX+5UYnoIAAAAXIvQAl2TawstqjqYHrL7fWnZ16S/XiEd/CxuwwL6PPtypxLTQwAAAOBa3kQPAC6VY5seUl1ullr0eIIf29wg/eXCtmqMQ2XSZf+I/xiBvsg+PSQpRUpql08zPQQAAAAuQqUFusZeadFUK9UfDP3Y8g8Dp498tkpqOBy3oQF9mr0Rpzc98DbTQwAAAOAihBboGntPCyl8M87KTSEey2ojQFwEhRapgbeZHgIAAAAXIbRA13jTpMwBgdvCLXsaMrQoi/2YAASHFu37WUhMDwEAAICrEFqg6+x9LcJVWuzbErzt0K7YjwdAcE8Lry20YHoIAAAAXITQAl1n72sRbtlTpocAPce+5Kk9tGB6CAAAAFyE0AJdZ+9rESqI8DVJ+7cFb2d6CBAf9iVPg0ILW48LKi0AAADgYIQW6LrcEMue2h3YKbU0B2+n0gKID/v0EHtPiyTbSteEFgAAAHAwQgt0XTSVFqGmhkj0tADiJWj1EKaHAAAAwL0ILdB19kqLzoQWVYQWQFx0GFowPQQAAADuQWiBrssfEXi7tjJ4BZHKzaGf21Al1VfFZ1xAX9bRkqdMDwEAAICLEFqg6waMllJzArftfi/w9r4woYVEXwsgHjpa8pTpIQAAAHARQgt0XVKyVDglcNuutW1fW5ZU8Wn457OCCBB7HS55yvQQAAAAuAehBbpn2LTA2+1Di9p9Uv3B8M+l0gKIvY6WPGV6CAAAAFyE0ALdUzQ98PaudVJLi/k6XBNOP1YQAWKvoyVPmR4CAAAAFyG0QPfYKy0aDkn7tpivwzXh9GMFESD2WD0EAAAAvQihBbont1DKGRq4zT9FpKNKC0ILIPY6Ci2SbJUWhBYAAABwMEILdF+4vhb2Sou84YG3e2p6SFOd5GvumdcCEq2jJU/tIUZzPcEFAAAAHIvQAt037AuBt/2hhX2506PmBd7uiUacG1+Q/mu89IvB0lu/i//rAYnW0ZKnOUNsT7Ckw5/HdUgAAABAVxFaoPvslRZ7PpLqq6QDOwK3j54feLuxWqo/FNeh6eUfS3UHpJZm8/Wu9+L7ekCidTQ9JHNAcF8LVvIBAACAQxFaoPsKp0rytN1uaZI2PCdZLYGPG2WrtJDiO0WksUbavy1w2/Lvxu/1ACfoKLTweIL70BBaAAAAwKEILdB96XlSQUngtnX/F3g7Z6iUVSBlFgRuj+fFUk1F8LbyD6Q9H8fvNYFE62jJU0nKHRZ4m9ACAAAADkVogdiwTxH57K3A2wPGmL/z7BdLZfEbU01l6O1v3B2/1wQSraNKC8ms+tNeNaEFAAAAnInQArFhb8ZpVzTd/G3/hDee00NCVVpI0id/C17ZBOgtogotmB4CAAAAdyC0QGzYKy3aGzJJmv1d83VPlqUf3hvmDkt647/i97pAInW05KkU4vewPH7jAQAAALqB0AKxMfjY0BdHR82Xlv5dysg3t+1l6XGdHhKm0kKSPnxc2r89fq8NJEpHS55KIRpxxrHiCQAAAOgGQgvEhjdVGjU3cNukC6SvPiGl5bRtyysKfExcG3GG6WkhSZZP+uTp+L02kChRTQ+xVVpUl0uWFb8xAQAAAF3kTfQA0It88U7puRrpUJk043Jp9nVmecX2QvW0sKzgx8VCpEoLKXg5VKA36EpPC1+jVLvPrPADAAAAOAihBWKnYIx02YuRH2OfHtJUI9UflDL6xX489tAio79Ut7/tdvWe2L8mkEgtLVJLU+C2UNO2sgdLniTJamnbVrWb0AIAAACOw/QQ9Cx7aCHFb4qIfXrIkIm216X5IHoZez8LKXSlRXKKlDUocBsriAAAAMCBCC3Qs7xpUtbAwG3xWvbUXmkxdFLg7WpCC/Qy9qkhUujQQgoOEKsJLQAAAOA8hBboefZmnAd2xP41WnxSrb3SwhZa1O2Xmupj/9pAooQKLUJND5FCrORDaAEAAADnIbRAzxswNvB25aexf426A4Hz9aXg0EKSDtPXAr1ItNNDJEILAAAAuAKhBXpeQUng7cpNsX+NUCuHDBgteTMCt9HXAr1Jc2PwNkILAAAAuBihBXpegb3SYnPsXyNo5ZB+pvmgfalH+lqgN2m2TXfyJElJYRaJyiG0AAAAgPMRWqDnDTw68HZ1uVRfFdvXsIcW/uafOYQW6MXs00OS0ySPJ/Rjgxpx8rsAAAAA5yG0QM/rf5T5BLi9fTGutrAvd0pogb7A3ogz3NQQKTi0aKiSGqpjPyYAAACgGwgt0PO8aVK/kYHbYj1FJKjSosD8nTMkcDs9LdCbdCa0sAd4Er8PAAAAcBxCCySGvRlnRYxXEDm8N/C2v9IiqCSe1UPQi9hDi3DLnUpSaqbp9dJe1a7YjwkAAADoBkILJEZQM84YryAS9fQQmg+iF7H3tIhUaSEFN+OsLpdK35X++Z/SxhdiOjQAAACgK8K0lQfirMDWjLPHpofYQ4s9kmWFb1YIuIl9ydOOQovcQmnvJ223N74gbX6lLfw4+z5p0vmxHSMAAADQCVRaIDHs00P2b5N8TbHbf1BoMcj8bV/ytKlWqj8Uu9cFEsm+5GmHoYXt92Hj8sBqjRV3mVAPAAAASBBCCySGfXpIS5N0YGfs9h9uekj2kODH0tcCvUWoJU8jyR0W+f59m6Wdb3VvTAAAAEA3EFogMTL7S5kFgdsqY9SMs6lOarQt3egPLVLSpYz+gffR1wK9RWdWD5FCryBit/bhLg8HAAAA6C5CCySOfYpIrJpx2qsspLaeFlLovhZAb9DZ0KKjSgtJWv+sVLu/62MCAAAAuoHQAokz0B5axKgZZ41tudOkFCk9r+22fR5/FZUW6CWCljxNjfx4++9CKL4G6cPHuz4mAAAAoBsILZA4HVVa7N8m3Xey9PMh0vLvB6+MEE6ofhbtVwfJsfW1oNICvUXQkqfpkR+fWxh6e0pm4O21j9CQEwAAAAlBaIHECRVa+C+MGqqlxy6Qdq2VmuukNQ9IL94Q+sKppUV663+kJy4xpezhljv1y7FdqFWXd+/7AJwiaMnTDiot0vODA4qUTOmsPwRuq9ggla7u9vAAAACAziK0QOLYVxCpPyQd3muCieevC668WPuQ9O79wftZfa/08s3S+mekJy+V1j8XeL+/CadfUKUFoQV6iaAlTzuotPB4pLziwG3Tlkrjz5T6jQrcTkNOAAAAJAChBRInrzj4oqpykwkmPv5r6Oe8+ENp22uB29Y82Pa11SJtfinw/uxBgbftJfFVhBboJTq75KkkTb6w7eucQmn2d6SkJGnaJYGP+/BxqWxt98cIAAAAdAKhBRInKVkaYKu2eO1O6R83hn+O5TPTQPZvM7f3b+t4qdSg6SG2SovDn0stvujGDDhZZ1cPkaTZ35XOfUhacJu09O9tvx9TvhbYyNPySU9fITUcjtlwAQAAgI4QWiCx7FNEdr4ptTQFbht7WuDt+oPSyz8xX3/6j45fI2h6iK3SwvIF98EA3KgroUVSknTs2dLs66T+7aaEZA+S5lwf+Nj926SXbur+OAEAAIAoEVogsezNOO1mXyd9ZZk07vTA7Z/+XTq0S9r0YsevYQ8tsgokT3LgNvpaoDfoSmgRyZwfSMOmB2577xFp4wvd2y8AAAAQJUILJNa4L4W/b9Q86eRbzCfBZ90jpWa33We1SKt+L+18q+PXsIcWScnBU0Toa4HeoCs9LSJJ9kpn3yulZAVuf/ZbbVO0AAAAgDjyJnoA6OOGTpa+9pT00ZMmiEjLldLzpAGjpYnnmYsmSUrPlSaeG7iCwdt/kBRiCVQ7e08LyYQWVbvabse70uLwXmnHm1LhFKn/UfF9LfRdnV3yNBoDRkuLfik99622bXX7pf9dIl32UnAACAAAAMQQoQUSb+wC86cj05ball2MIrCQgistJClnaODteIYWB3ZKD5525DU80iXPS6PmxO/10Hd1dsnTaE29SNr0D2nj8rZtB3ZIfz5HuvQFKSM/Nq8DAAAA2DA9BO5ROEUaOiX8/RPODL090aHFSze1278lrfhl/F4LfZvPVmnR3ekhfh6PdNYfpCETA7d//rH0++Ol30ySfjlKun+B9OETkhVloAgAAAB0gNAC7jJ9afj7Fv48OIxIywvdjDDX9riOelp0dUnU7W8EfjotmT4cdQe7tj8gkqBKixiFFpKZtnXR08HTmw7vkQ7uNFNGylZLT18pPbVUqt0fu9cGAABAn0VoAXc59pzAhpx+QyZK+cNN34v2QvWzkILDjc8/luoOBD+upUV67U7pjiLpvyZIn70d/VhbfKGXh7R80pZ/Rr8fIFpBPS1iGFpIZhnUr/9Nyu6gj8Unf5PumS09/nXpkTOke+dLf71S2vSS5GuO7ZgAAADQqxFawF3SckyDTrujj6xCMvViKbld88FRc0Pvx77U6uHPpScvlXxNbdtaWqTl10mv3SE11ZrGncu+GjrcCOWDv0h7Pgx936Z/RLcPoDPiWWnh12+kCS4ywwSCftW7pQ3PSdtfl3a/J330hPTY+dKvJ0gv3SztWhs8jaSh2vxhegkAAACOoBEn3GfapdLahwK3lXzR/D2wRDr/f6V37zdl7Kf8JPQ+hk2Thk2Xdq1p27btNekfN0qL/5/5NPjZa6UPlwU+r3af9O+fS4vvjjzGhsPSv24Lf//mV8xrJPMriBiK9ZKn4QyeIF3ztrT1X22r/tTtl165peNQ7/Dn0qr/MX/yh0ujTzE9X/Z81LaijydJSs2R8oqkyRdIM66QUtstu9rcICV5zfLFAAAA6NU8ltV7P9IqKytTcXGxJKm0tFRFRUUJHhFi5omLpfXPmq+POkn6+jOmWWBnHCozZes1ewO3DxhrLooOfRbmiR7pqtdMY9BQLEt64fvSmgcjv/6lf5dGzu7cmIFIfj5Eaq5ru33lv01A11OqdkvPXCNtezW2+80skI6/2vTJ2PaqVLFR8mZIY06Rxi02oWVm/67t+2CptP4ZE/BMPLdr+ylbK21fYSq4jv6SlEQRIwAA6Hvidf1NaAF3aqw1FxrN9abPRXpe1/ZTtkZ66EvBn1B3pGiGdNnLoS9O/v1z6fVfBW4bc6pU/bn0+Udt2074jrTwZ50fMxCKZUk/7aeApYC/+ZY0+JieHUdLi/TJ09LudaY6Ij3f9HH55G9mSkg8eJJNADjuDLOc8O73pc0vmR40liUNGicNOkYaeLSUXyzlFkmNh6W3/yB9/LQZn2SqO46/Wpp1bejwoqnO/ElOMdPQtr8uvflraefKtseMP0Nacq+Umhmf77Wvsywzjc+b2vFju7Lv/duknCGBlT0AACAqhBZdQGiBqHz4pPT0FeHvT06TSk4z8/PbGzZdOrzXVGr0Hy2NP11qaZbesE0dSU6Vrn5D+vipwDCjoET61rux+R58TeYNd6Q38i0+6YNlpox/ytcCL8osy3x6nZYr5Q2LzZjQs3xN0s9sfSa+tVYqGJOY8YRS8an04eMmwNi/LdGjCS81Wxp8rFllKHOAqcrau0E6+JkCQqFwhk2TvrLMNC6VpP3bzZSwzS8fCXMypaLjpOLjzTS2mr1S9R5T4VU0XRp9ctvUl11rpXcfNFPTjj3H/OmrlRyfvSM9dZmZYjTxPOm0X3S9wsZu++vS32+QKjZI3nRp2lLpxO+aAAMAAETF9aHFzp079d///d964YUXVFpaqrS0NI0ePVrnn3++rr32WmVmxv5TKUILRO31X5kKCbu0XOmC/5WGz5L+MEvav7WTO/ZI5z5gLjTK1kr3nxx497ffkwaM7vKwJZlpKK/daapPpnxVOvU/gz/l9TVLj3+trQFo9hDpshfNBVNjjfTkUvPJdJJXOu1282kz3KWh2qxy0953PzaVBU5jWaaHxYbnTHiRWygNmWRWAUrNkuqrzFKq7/zJXOi7UdYgKaOfCSMaDnXuuXnDpalfk0pXm74h7Q06xvTqGXi0dGiX2X9GvlR8XHQVZ0110r4tkjxSwdj4NGu1rM5P1+vI3o3SAwsDf5Y5Q6Uz/8dUsnVV9R7TGPbjp4Lv86abHkon3Wh+xgAAICJXhxbPP/+8LrroIlVVVYW8v6SkRC+88ILGjIntJ4KEFuiUvRvMJ8GyzJvu5BTzSaj/09It/5T+fE7n9vnl30lfuNh83dIi3X10YA+NE74jjTtd8jVK9YfMn6Zas0LDyBOllIzw+/Y1myVVV/8pcHvB0SYoGTLR3LYs6YXrpTUPBD4uf7h00dPSc9+RPnsr8L5zHghePhbOVrNP+tVRgdt+sLnt+HWrsrWmeqlstalOOmq++d04sF3asNxc1NtXTeksb4aZIma1xGbMieBJlgqnmqlrsszPpKne9DhpbjDnlYOfSQd2qrVaJHuwdPw3pOmXtV2U+1dwaW4w5yVPkpQ/IrrpGPu2mn+rjcvN+WXOD6QJZ0YXYDTVSUkpoZsTH94r3X/KkUqXEKYtNWFtqGDBsqRPXzTn76GTTJVZcoq5b9PL0l8vlxpCvzdpNXCcdNk/TAgFxNq+raYJcfHM+Ex7OlxhVl5LSY/9vgHAxrWhxbp16zR79mzV1dUpOztbN954o+bPn6+6ujotW7ZM9913nyQTXKxZs0Y5OTkxe21CC8TcU5eH/kQulNPukGZdE7jt2WuldX+O7vneDNNkdNA484nqoTITavQbKQ0ab8rM7Z/C+iWnmkDkmCXS1n9Lr4RZRcWT3Daf3/78rz8T3CjUOnIx5E2P/Sep6J5Du8xyou396LOu93txi8Yaacu/pI0vSJteNL8jKVnS6PnS2IXm+9+7Qdr7ibR/h1RV1rbCSdYg6birpBmXSzWVpuLq46c6F14UHC3N/o6pFHn8a+EvrJ0sNcdUu1SXh76AT0ox56Ehk6X0XBNm+BrN9qyBUlaBtOs9M/XHfj4ZMVuaf5M5d1TtlmoqZEITj3lsxafmuZWfmseMmmP+TUoWmQCj4bD06Jc77oeSNVBacJs06cK26TPlH0gv/igwlC38ggl1t60wDZOj/bceNVf62l+7f1HpllWjKj6V1j9nqnHGLW4LehBb7z0qLf++1NJkziEXPxu7KU/1VdLfrpY+/bs51531B2nsgtjsGwDCcG1oMXfuXL3xxhvyer16/fXXNWvWrID7f/WrX+mGG26QJN166636z//8z5i9NqEFYs7XZJZqrNxilmMsGGs+fdv6b2nD89LBneaC/5RbpBO+Hfz8DcvNhY0bpOdLE75sVleo2mUu9OoOmjdXmQOkyV+RZl4T3AOjpUUqe1fa+aYJRUbPN2/GCDnia/826b+nBm67+fO+9emar8mEFmk5kac9NNaYaoLswcHH5aEyc4FcvcdcxNdUmDf8A8eZC/fsweZ1fI3mNXKHte3j8F7pLxeGvsDOLDAXDKNPMU1AS1eb6pH6Q2b/OYPNJ6Ltm/X6ZQ00v3MVG7v+s3GbnEJzcX+wVEF9RFJzpMbq0M8rKDE/rxafVPpO8HMlE7qGqs5Jz5Pm/VCq3GTC5ZbmwPunXmSmz217zZzjkrzm3z9nqKloSs2W0rLN8Zee33Zc1O6XVt8rvfe/JjRLy5OyB5pjqXCqCdeGzzJ9TCo+lXa/ZwKhEbNMxYpf9R7p479K8kjDviANnRy6Gq/6c/N/UnO92f+QieGXB/Y1m+/Tf55o8UkrfyO9envb9z9qnnT+I22VJod2mSAwo5/5naBpadfsXCU9cnrgcVbyRenCv3S/b03NPun/zjEfbvgleaWz7zXTVQEgTlwZWqxevVrHH3+8JOnqq6/WH//4x6DHtLS06Nhjj9WGDRuUn5+vvXv3KiUlNok+oQV6lGWZi3tvuvnkMZTGGuk3k6Tayti+dnKq1G+U+bSyIzlDzcWYXZI3+E16R5K8pvw7r9i8Qa/db/pmHP488HH+RqWZA9ptPPKGPuCisd3XHo+5nT3I9N4YMNq8wS5fZ96IHT7SAHXELNM0sf2bcssyn3jv+dBc9KRlm9fOLJD6jUhMcz3LMhc6Hz1pxpY/wvQhKD7eBGCdCXUsSzpUai4cPv/EXFinZAQ3gb3lQN9t2pgozY1m9ZTDe81xljPEXNj2G9Xxv4VlHWm8ef+RT0cHmqqDL1xsfsc/ekpa8cu23jqpOSbsOPiZCVH6gvzh0hX/MiHxyz82015iYcJZ0pf+nwkTJBNMP7QoeElsT1J01RkZ/Uz/kZzBZnpKR+NMzZFkmUDLL8lr+hRNu9T82695MDBsSfKac9/gY014kDnALAW++ZXAipfUHHOuGTHLhCODjzGhxgfLTJWS5TPBxuhTzIpaO98MHl9BiQls3n/MrNzV+jPwmOq/ohkm5B5zamCQYlkmCCxbLZV/aLblFbX9yR1mflbtz3/+8LHuoPm7pckEPDlDYxfCHthpvv/tK8z/X9MulYbPDBxHi0/a8aYJina8aaqR5t1gqm66q3qP9Ke5wf9XStIpt0pzvt/1fVeVS/97VpiQ0yOd8Vtp2iXBd9XsOxL4xmGKCvoWX5N5P5gzONEjQQK4MrS46aabdMcdd0iS3n777dYAw+7OO+/UjTfeKEl66aWXtHDhwpi8PqEFHGnba9Irt7Y1w/N4zAV3Wq75pM/jMU0Koy1bzhwgXfiY+UTtX7eZSpBwTrlVmnS+9OAXzUWvX9ZAMx3k1V+YiyW3Sctt+8TbskwoU38w/OOzB0tDp0gDS8zjW3zmjXtSinnD5k03F4neNPPHk2QanTbVmk/oq3abN+JVu8yFQ8FY86Y+Z4i5WK0uN397082/aWqWWZ0gXCPXnMK2ACN3qAlaDn5mwq3UbFMunJZr/s0+X2/CinCfNPslp0o/qejqTxSJFqmZ5eEKc5z6p/401kqlb0vb3zgSnKaZY8+bbi4gvWlmulnmAHPMDxgr7dssrfytuciNZS+PlCxp8oXSlldiP1UmLU+6/GVzgS5JB3ZI/7hJ+vSFjp8bbiqcZHpvnPzj4J/3rrXSQ4tNXxBEJzXbXPz7mtrOlYf3RH5OSqY5dzY3mJCifXBjlznAhAe5w4783e7r9HxTEVi7z/ydnGIuwtOOTGk6vNcEvNtfl3a8EbzvoVNMCF+7z1RNfvZOcGglSeO/bKYiZQ4w32Njjfk/PDlFksecn0vfMSG1f0pnQYn5fyJ7sAlpXrxB+mxV6O/RkySd8d/m/7DyD8z/McXHm7Akf7gJf3a+KVVskjL7mfuKjzff34bnTEVP9e7IP/OJ55vf02HTzGpO7z1iPgxIyTJNgGd+03xYEAuHK8x46w6agK1watsUqaZ6U9mU5DU/IzdMnXIKyzLh49v3SJ9/bP4tJ55rqnUi9UOLp6py6e3fS2seNu9RimaYsLP4uO7v27JMqLr6XtM4PneY6Wc05avmgyk4hitDC//UkKysLB08eFBeb+iT0apVq3TCCSdIkm655Rb99Kc/jcnrE1rAtWr3m8Zx214zb4r8n0ql5UiVm6W9680bdv+nX/1HtT139zrp/b+Y8KF9MDH9cmnx3eaNeeUWadlXTWVGQYl0wZ/NagSNNdLDiwNLSuFOqTnSTWWJHgWc7sAO6bO3zYVSzlBz8ZfZ3yz17E0z5589H5k/FRtNwJGcai7QmutNL5DDeyVZ0sg50qxrTaVZU70JUN+934R4ablm39mDzAWKZZnn5Aw1FzFDp5jmqu/8yUyPsCs42qwUEurN76aXzEXggR3B96VkSXOvN29uX7lFWve/bfd5ks05cfrS8D+fDc9Lj39dUS11C8RbUoqpPOmMlCypqaZzz/EkmWlB3nTz3OZGE5JmDjDnh6Zac4FaXW6+9odD6blHQv8jAc7u94OnvKVmmwvsmgozHcofJnrTzVTSwce0VX76wyb/dD1fU1som5JhxpJVYKook1PV+nvaemkT4nbY+2y8aeb7Ss02+26uM+c1X6N57dQs86f9+cyy2kLg1q+tdve3tPUGO7zXVNrU7jdNhPOHmz+ZA4IfH7APyzQufv/P5rxsl5YrDRhjwr/GGvNvmT24bQnv2n1mClnNXvM+YWCJmf6YV2yOreaGIwHiQRM01R0wr+drPNKw2jJBY2qmObaSksy/dUOV6S0Vqurv2HPM/w8HtpvztNViPqzJG2aOq6py80FQdbn5mecXm1W0UrOOTE3eb6rCyt8P3nd6nmmwPGCM+QAuI98cJ80N5t8syWvGm5Jh/l8L9XO1WszPqu6A+b6bG8yxlTnAhIyWdeT7b2rr6eRrMsdoUrL5v8TjMc/3h6PJqaZCeMBo873oyD6aG80qWP4p180NbePwJJnvPyP/yIcSHvNv4mtqe+2xC8xjHMyVocXAgQNVWVmpyZMn6/333w/7uAMHDqh/f9N46LzzztMTTzwR1f7LyiK/IS8vL9dxx5k3OIQW6HMsy6Tvu9ZKuUXSmFMCP0lsbjSfgOUMDWyyVn9Ieu1ICXrOEPOfaN5w88YgI1+Sx8z3XvfnyJ9ADhzXtmIBel7+COm7HyZ6FIB5w9uZT/52rzOfMKflmClg/Y8yF0ORNDeYT9APlR6pnGoxb/pGn9I25UMyIcTq+yRZpsLiqHkdj2fV781KTX7JqdKIE8wn+1W7zZ+6A5EvDD3J5o379KVmbIc/N8u4bnklOCTOKzZvekP13Ujymp/Jvs2RK2RSssz3HSrIiUbecGnhz6QVd5kmtnbedPMG2s0r7jhF3nAztSZSlWRXDRhjqijXPiy98f9iv38APevad03Q5GCuCy3q6+uVkWHepCxevFjLly+P+Pjs7GzV1NRo5syZWrUqTMmcjacTc8AJLYAYq9knffCY6algtZg/niSzssm4000prGWZlQHWP2NKQFvTbSn0pxy2bS3N5iLk4Gdtb469GWbpwrxi82Y/3JQLyVxUDBxn3vzX7gt/IdCT+o+WSk4zy9yVvhN5GktH0vLMzzt7kNlX+/nRp9wizbm+28MFINPTYPc6cxE4ck7ocuQWnwlqayqO9JtZLx3cYS5KJ19o+umEUv15W2PPwqlmHnj1Humt30nvPmDCYU+yKYOe+wMz3aCh2oxnz0dHVsfZYJ4z4Chp0gWmR0dattn22dtmKsLOt0yQbbVI8pjAZtKFpopv67/Nn6pdZrWQU39qQur6KumvV5hybMl86jjzGum4K01wUbHRlGyvf1bauTJ8iJHRTxo23Xw6e2iXeZ3qPeqwgsWbbr73zlYKRCM9Xzp6kfm5HNwZ/nFDJpmpGe8/Zj7xjaXkNDPlacgk6f/ODb8iWFcMnyWd/2jbster/iD989a+0/8G6I2+sVIacmyiRxGR60KLiooKDRpkTpQXXHCBli1bFvHxgwcP1t69e3Xsscfqo49ClDuFQGgB9BHNDaZxmmQ+dW0/77V6j3nz3tjuTW1qtjR4QluDUL+WFhNy7H7flBke3msuFJK9JnDxNZvyR395pK/BVKRYPvNJcUqWedOdPdjsO2+Y+RS5cpMpc607YMoTcwvNY1raNZTL6GcuBopmtI2ppcX0Nil9x/zZtdZ8H3nFpsIlZ8iRksX9Zh+Z/aVBE0z57KDxgStXWJa5cCn/wJSBHnVS3P45APSQ2v1musygY8zvdXfVV5kL9OwhgRUokViWCW2aas0StuHmjx/ea6bqVO0+sorKkekCgyaYsMf+nq250ZSDHyozpeopmSZISM87Uh6d39Z4s77qSFXLrrbqFvvX9YfMeTZroDlX+vtqNFSZ83zWkRVbcoaYwOboL5kpAC0+00D6wyfM95Bb2FaqP/JEE8BL5vz+2i+lNQ9EvvD3ppuldYuPM2Xu+7eb/x/2b2srP7dazPd31j3SuC+Z59Xsk5Z95ciqNx7zcyueYZb93f56YH+NgqPN/yWHSk1o5A918opNz40JZ5r77c1//SvQfPhEW6m9J9kE6ZMuMOHZ2kc67pnUWSlZ5vjdtyW2+4X5IGTS+eZ3dMebSvhUtuQ00xel30jTHLz+UGz3nzfcrOS0e535vU3099vTrnrNhNsO5rrQorS0VMOHm6W6vv71r+vRRx+N+Pjhw4ertLRUo0eP1pYt0Z3UmB4CAACAiM1rY6mpzoToqVlHeh5kmRDCP889LTdyQ8mWFhMKpOUGj9fXbKZtpuW0NdqVzPdWucn0kCkoCQycfM0mDEhOMaF+tD+Dyi2mr1XhFwIDsfpDpnLmwM623g3JKSY4r91n/njTTE+C3KFmrA3VJlhqqG4397/BhEQj55geFt5UEwrteMM0Ec0qMKvWDD7GPH73e9KudWZZYE+yCZqSkk0AlTPUhE0pWW0fLDRUH6mgrDR/tzbata9M1v52pPva8fedaKg2vSF8jabKMyXjSH+LevOBQuNhE3r59+3xmA9A/F/79x2wzWP24V/2OHOA+R4OfmYCxcaawP2F+zt7sAmaxi1uWz3t0C7z8/U1mg9vUrPNhyfV5aZvRN1+KaO/+VlmDzrSV2STqZiq3W/+jfxNyNPzTLCWkW96XySnHFlK3GMCzMYa83f7/hD5w02TV/+qIbX7TXCxfYW53W+UOUaTvEfCxjJzvOUMNRVfOUPN7UOlpiG5r9EEkf4wcvR802jU//3u2yq996gJBWsrzc+x/uCR7+FIM2rLZ35nm2rNcdb672T7maZmtX2/ySntmvoePNJoN7Wtn5P/b3+D55Yj1cYZ+W09Vppqze9l5RbTw0Iyx0Fyatuy2Bn5Jqz1j8Hymd+j+oNHwh5P4OslpUjnPmB+bxzMdaFFT1RadIRGnAAAAACAHmdZJmRLTmkLW3q5eF1/x21toZycts6mhw9HWL7qiJoaU9qWnc2yNQAAAAAAF/N42qa5oVuSOn5I16Snp2vAgAGSOp7GceDAgdbQwp/MAAAAAACAvi1uoYUkTZgwQZK0ZcsWNTc3h33cxo0bW78eP358PIcEAAAAAABcIq6hxYknnijJTP1Yu3Zt2MetWLGi9evZs2fHc0gAAAAAAMAl4hpanHXWWa1fP/TQQyEf09LS0rqySH5+vubPnx/PIQEAAAAAAJeIa2hx3HHHac6cOZKkBx54QKtWrQp6zN13360NGzZIkq677jqlpKTEc0gAAAAAAMAl4rZ6iN9vf/tbzZ49W3V1dVq4cKFuuukmzZ8/X3V1dVq2bJnuvfdeSVJJSYmuv/76eA8HAAAAAAC4RNxDi6lTp+rxxx/XRRddpKqqKt10001BjykpKdELL7wQsEwqAAAAAADo2+I6PcTvjDPO0Icffqjvfe97KikpUWZmpvLz8zV9+nT98pe/1Lp16zRmzJieGAoAAAAAAHAJj2VZVqIHES9lZWUqLi6WJJWWlqqoqCjBIwIAAAAAoPeJ1/V3j1RaAAAAAAAAdBahBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsAAAAAAOBIhBYAAAAAAMCRCC0AAAAAAIAjEVoAAAAAAABHIrQAAAAAAACORGgBAAAAAAAcidACAAAAAAA4EqEFAAAAAABwJEILAAAAAADgSN5EDyCempubW78uLy9P4EgAAAAAAOi92l9zt78W765eHVpUVFS0fn3cccclcCQAAAAAAPQNFRUVGjlyZEz2xfQQAAAAAADgSB7LsqxEDyJe6uvr9dFHH0mSBg4cKK/X+YUl5eXlrVUhq1ev1tChQxM8IqANxyecjOMTTsWxCSfj+IRTcWy6T3Nzc+tsh4kTJyo9PT0m+3X+VXw3pKena8aMGYkeRpcNHTpURUVFiR4GEBLHJ5yM4xNOxbEJJ+P4hFNxbLpHrKaEtMf0EAAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADiSx7IsK9GDAAAAAAAAsKPSAgAAAAAAOBKhBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHAkQgsH2blzp66//nqNGzdOWVlZ6t+/v2bMmKFf/epXqq2tTfTw0Mt4PJ6o/px00kkd7uvFF1/UkiVLVFRUpLS0NBUVFWnJkiV68cUX4/+NwHX27t2r5cuX65ZbbtGiRYtUUFDQerxdeumlnd5fLI6/5uZm/fGPf9ScOXM0cOBAZWRkaPTo0br66qv1ySefdHpMcK9YHJ8PP/xw1OfYhx9+uMP91dbW6q677tKMGTPUv39/ZWVlady4cbr++uu1c+fO7n3DcI01a9botttu08KFC1vPd9nZ2SopKdHSpUv15ptvdmp/nDsRS7E4Pjl3IiwLjvDcc89Zubm5lqSQf0pKSqzNmzcnepjoRcIda/Y/8+bNC7sPn89nXX755RGff8UVV1g+n6/nvjE4XqTj5ZJLLol6P7E6/ioqKqwZM2aE3UdaWpp13333dfO7hlvE4vh86KGHoj7HPvTQQxH3tXnzZmvs2LFhn5+bm2s9//zz3f/G4Whz5syJ6ni6+OKLrYaGhoj74tyJWIvV8cm5E+F4g1IM9Lh169bpggsuUF1dnbKzs3XjjTdq/vz5qqur07Jly3Tfffdp06ZNWrx4sdasWaOcnJxEDxm9yDe/+U1dc801Ye/PysoKe9/NN9+sBx54QJI0depU3XDDDRo9erS2bt2qu+66S+vWrdP999+vgQMH6vbbb4/52OF+w4cP17hx4/Tyyy93+rmxOP58Pp+WLFmid999V5J09tln68orr1T//v31zjvv6Oc//7n27t2rq6++WsOGDdOiRYu6/s3CdbpzfPq99NJLKiwsDHt/UVFR2Puqq6u1ePFibd68WZJ05ZVX6sILL1RGRoZeffVV3XHHHaqqqtIFF1yglStXasqUKV0eJ5xt9+7dkqTCwkKdd955mjNnjoYPHy6fz6dVq1bp7rvv1q5du/Too4+qqalJjz32WNh9ce5ErMXy+PTj3IkAiU5N0JZOer1e66233gq6/6677mpNBW+99daeHyB6pe4eU59++qnl9XotSdb06dOt2tragPtramqs6dOntx7bVArB75ZbbrGef/55a8+ePZZlWdb27ds7/Ul2rI6/Bx54oPW1r7nmmqD7N2/e3FoFN2bMGKupqalz3yxcJxbHZ/tPC7dv397lsfzkJz9p3c9dd90VdP/KlStbfw8iVcXB/RYvXmw9/vjjVnNzc8j7KyoqrJKSktbjZcWKFSEfx7kT8RCr45NzJ8IhtEiwd955p/WX6uqrrw75GJ/PZ40fP96SZOXn51uNjY09PEr0Rt0NLb75zW+27mPVqlUhH7Nq1aqIb2oAy+raRWGsjj//ubV///5WTU1NyMfccccdrft54oknohofeo9EhRaNjY1WXl6eJckaP3582FL9q6++uvW1Vq9e3aXXQu/w/PPPtx4L3/72t0M+hnMnEiWa45NzJ8KhEWeCPfPMM61fL126NORjkpKSdPHFF0uSDh48qFdffbUnhgaEZVmWnn32WUnSuHHjNHPmzJCPmzlzpo4++mhJ0rPPPivLsnpsjOi9YnX8bdq0SRs2bJAknX/++crMzAy5n/bNF//2t791d/hAVF599VUdOnRIknTJJZcoKSn0WzaOT/jNnz+/9eutW7cG3c+5E4nU0fEZK5w7eydCiwTzd9LNysrStGnTwj5u3rx5rV+vXLky7uMCItm+fXvr/MX2x2Yo/vt37dqlHTt2xHto6ANidfy172QeaT9DhgxRSUmJJM6/6DnRHp/Tp09vvWjk+OzbGhoaWr9OTk4Oup9zJxKpo+MzVjh39k6EFgnmT6rHjBkjrzd8X9Rx48YFPQeIhSeffFITJkxQZmamcnJyNHbsWF1yySURK3rWr1/f+nX7YzMUjl3EWqyOv67sp7S0VDU1NVGPFVi6dKkKCwuVmpqqgoICzZw5Uz/+8Y+1a9euiM+L9vj0er0aM2aMJM6xfd2KFStavx4/fnzQ/Zw7kUgdHZ92nDvRHqFFAtXX16uyslJS5A64ktSvX7/WVRxKS0vjPjb0HevXr9eGDRtUV1enw4cPa8uWLXr00Ud18skna8mSJa0ldu2VlZW1ft3RsVtcXNz6NccuYiFWx19X9mNZVsDzgI689tprKi8vV1NTk/bt26d33nlHv/jFLzRmzBj96U9/Cvs8/3GWlZWl/Pz8iK/hPz4rKioCPs1E39HS0qI777yz9fb5558f9BjOnUiUaI5PO86daI8lTxOourq69evs7OwOH5+VlaWamhodPnw4nsNCH5GZmakvf/nLOuWUUzRu3DhlZ2eroqJCK1as0B//+Eft27dPzzzzjM4880y98sorSklJaX1uZ47d9kumcuwiFmJ1/HEcI56OOuoonX322Zo1a1brG+Nt27bpr3/9q5566inV19frG9/4hjwej6666qqg5/uPz2jfH/gdPnxYaWlpMfou4Ba//vWvtXr1aklm+dFQU445dyJRojk+/Th3IhRCiwSqr69v/To1NbXDx/t/kerq6uI2JvQdu3btCplAL1iwQN/+9re1aNEirVu3TitWrNA999yj73znO62P6cyx2/4/AI5dxEKsjj+OY8TLkiVLdMkll8jj8QRsnzFjhi644AItX75cZ599tpqamvS9731PX/7ylzVkyJCAx/qPz868P5A4PvuiFStW6Ec/+pEkadCgQbrnnntCPo5zJxIh2uNT4tyJ8JgekkDp6emtXzc2Nnb4eH/ZUkZGRtzGhL4jUsnc4MGD9dRTT7VWV/zud78LuL8zx277cjuOXcRCrI4/jmPES15eXtCb7vZOP/103XLLLZKk2tpaPfDAA0GP8R+fnXl/IHF89jWffPKJlixZoubmZqWnp+vJJ5/UoEGDQj6Wcyd6WmeOT4lzJ8IjtEignJyc1q+jKZnzNzCKptwJ6K6jjjpKCxYskCRt2bKlteO41Lljt33jLY5dxEKsjj+OYyTSVVdd1frmvH2DOj//8dmZ9wcSx2dfsn37di1cuFAHDhxQcnKyli1bprlz54Z9POdO9KTOHp/R4tzZNxFaJFB6eroGDBggSR02Jzpw4EDrL1b75khAPE2YMKH16/bdmts33uro2G3fwItjF7EQq+OvK/vxeDwdNp4DojFo0KDW9wChuuH7j7OamhodPHgw4r78x+fAgQOZk91H7N69W6eeeqp2794tj8ejBx98UGeeeWbE53DuRE/pyvEZLc6dfROhRYL5Lwq3bNmi5ubmsI/buHFj69fRLBMExEK4Er32YUb7YzMUjl3EWqyOv67sp7i4OKBxF9Adkcqgoz0+m5ubtXXrVkmcY/uKyspKLViwQNu2bZNkpnBefPHFHT6Pcyd6QlePz87g3Nn3EFok2IknnijJpIFr164N+7j25U+zZ8+O+7gAKXCt68LCwtavR40a1Xo7VGlee6+//rokadiwYRo5cmTsB4k+J1bHn//829F+9uzZo02bNkni/IvYqaioaF32vP351S/a43PNmjWtlZgcn73foUOHdNppp7X+/3znnXfq2muvjeq5nDsRb905PqPFubNvIrRIsLPOOqv164ceeijkY1paWvToo49KMs0T58+f3xNDQx+3fft2vfLKK5Kk0aNHa9iwYa33eTye1jK/jRs36u233w65j7fffrs15T7zzDMjJuNAtGJ1/JWUlLR+uvLEE0+otrY25H4efvjh1q+XLFnS3eEDkqR7771XlmVJkubNmxd0/0knnaS8vDxJ0iOPPNL6WDuOz76jtrZWixcv1nvvvSdJuvnmm/XDH/4w6udz7kQ8dff4jBbnzj7KQsLNmTPHkmR5vV7rrbfeCrr/rrvusiRZkqxbb7215weIXue5556zmpqawt6/Z88ea+rUqa3H3d133x30mE8//dRKTk62JFnTp0+3amtrA+6vra21pk+f3npsb9q0KebfB3qH7du3tx5rl1xySVTPidXx98ADD7S+9rXXXht0/5YtW6zc3FxLkjVmzJiIvzfonTp7fG7fvt167733Ij7m+eeft1JTUy1JVkZGhlVWVhbycT/5yU9aX/uuu+4Kuv+tt96yvF6vJcmaN29eNN8OXKqhocFauHBh6/Fw3XXXdWk/nDsRD7E4Pjl3IhKPZYWJn9Bj1q1bp9mzZ6uurk7Z2dm66aabNH/+fNXV1WnZsmW69957JZlke82aNQFdm4GuGDlypJqamnTOOedo1qxZGjlypDIyMlRZWanXXntNf/rTn1pL70488UT985//DNmg6MYbb9Sdd94pSZo6dap++MMfavTo0dq6dat++ctfat26da2Pu/3223vuG4Sjvfnmm9qyZUvr7crKSv3Hf/yHJFOiecUVVwQ8/tJLLw25n1gcfz6fT/PmzdPKlSslSeecc46uvPJK9evXT6tXr9bPfvYz7d27V0lJSVq+fLkWLVrUre8dztfd4/O1117T/PnzNWvWLJ1xxhmaPHly6xJ/27Zt01NPPaWnnnqq9dO/3//+97rmmmtCjqW6ulrTp09vLbG/6qqrdOGFFyojI0Ovvvqqbr/9dh0+fFgZGRl66623NGXKlFj8COBA55xzjp5++mlJ0sknn6zf/OY3EasXU1NTVVJSEvI+zp2ItVgcn5w7EVFiMxP4Pffcc62JdKg/JSUl1ubNmxM9TPQSI0aMCHustf9zzjnnWAcOHAi7H5/PZ1122WUR93H55ZdbPp+v5745ON4ll1wS1fHn/xNOrI6/iooKa8aMGWH3kZaWZt13332x/jHAobp7fL766qtRPS8zM9P605/+1OF4Nm/ebI0dOzbsfnJzc63nn38+Hj8KOEhnjklJ1ogRI8Lui3MnYi0WxyfnTkRCpYWD7Ny5U7/97W/1wgsvqKysTKmpqRozZozOO+88fetb31JmZmaih4heYsWKFVqxYoVWrVqlbdu2qbKyUlVVVcrOzlZxcbFOOOEEXXLJJZo1a1ZU+/v73/+ue++9V++++64qKytVUFCgGTNm6Oqrr+bTFQS59NJL9cgjj0T9+I7+m4rF8dfc3Kz77rtPjz32mDZs2KCamhoVFhbqlFNO0XXXXadjjjkm6vHC3bp7fFZXV+u5557TqlWrtGbNGpWXl6uyslLNzc3q16+fjjnmGJ1yyim64oorWj9F7EhNTY1+//vf68knn9SWLVvU2Nio4uJifelLX9J1112nESNGdOp7hPt0tifUiBEjtGPHjoiP4dyJWInF8cm5E5EQWgAAAAAAAEdi9RAAAAAAAOBIhBYAAAAAAMCRCC0AAAAAAIAjEVoAAAAAAABHIrQAAAAAAACORGgBAAAAAAAcidACAAAAAAA4EqEFAAAAAABwJEILAAAAAADgSIQWAAAAAADAkQgtAAAAAACAIxFaAAAAAAAARyK0AAAAAAAAjkRoAQAAAAAAHInQAgAAAAAAOBKhBQAAAAAAcCRCCwAAAAAA4EiEFgAAAAAAwJEILQAAAAAAgCMRWgAAAAAAAEcitAAAAAAAAI5EaAEAAAAAAByJ0AIAAAAAADgSoQUAAAAAAHCk/w+xEqRLUY/FEwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 413,
       "width": 534
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot losses\n",
    "import matplotlib.pyplot as plt\n",
    "%config InlineBackend.figure_format = 'retina' # Makes the plots clear on high-res screens\n",
    "plt.plot(G_losses, label='G_losses')\n",
    "plt.plot(D_losses, label='D_losses')\n",
    "plt.legend(frameon=False)\n",
    "plt.tight_layout\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(netG.state_dict(), \"checkpoints/generator.pt\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "model.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = []\n",
    "original_musics = []\n",
    "\n",
    "with torch.no_grad():\n",
    "    for batch in test_dataloader:\n",
    "        input = batch['input'].to(device)\n",
    "        target = batch['target'].to(device)\n",
    "        for i in range(len(input)):\n",
    "            generated_tokens = []\n",
    "            output = netG(input[i])\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "                # current_token = torch.argmax(probabilities[j]).item()\n",
    "                generated_tokens.append(current_token)\n",
    "\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "            generated_musics.append(generated_tokens)\n",
    "            # original_musics.append(target[i].tolist())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(generated_musics)):\n",
    "    midi_encoder_remi.words_to_midi(generated_musics[i],f'generated_musics/gan/gan{i}.mid')\n",
    "    # midi_encoder_remi.words_to_midi(original_musics[i],f'generated_musics/original/original{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Q1\n",
      "Q2\n",
      "Q3\n",
      "Q4\n"
     ]
    }
   ],
   "source": [
    "max_length = 256\n",
    "temperature = 1.0\n",
    "end_token = 1\n",
    "\n",
    "model.eval()\n",
    "# current_token = start_token\n",
    "generated_musics = {'Q1':[], 'Q2':[], 'Q3':[], 'Q4':[]}\n",
    "\n",
    "with torch.no_grad():\n",
    "    for c in classifier_valid.keys():\n",
    "        print(c)\n",
    "        for batch in classifier_valid[c]:\n",
    "            input = batch['input'].to(device)\n",
    "            target = batch['target'].to(device)\n",
    "            generated_tokens = [input[0]]\n",
    "            output = netG(input)\n",
    "            # Apply temperature to the output probabilities for diversity\n",
    "            probabilities = softmax(output.squeeze() / temperature, dim=-1)\n",
    "            for j in range(max_length):\n",
    "                current_token = torch.multinomial(probabilities[j], 1).item()\n",
    "                # current_token = torch.argmax(probabilities[j]).item()\n",
    "                if current_token == end_token:\n",
    "                    break\n",
    "                else:\n",
    "                    generated_tokens.append(current_token)\n",
    "\n",
    "            generated_musics[c].append(generated_tokens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in generated_musics.keys():\n",
    "    for i in range(len(generated_musics[c])):\n",
    "        midi_encoder_remi.words_to_midi(generated_musics[c][i],f'generated_musics/gan_by_emotion/{c}/transformer{i}.mid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}