Preprocessing_Training.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import torch\n",
    "import mne\n",
    "import matplotlib as plt \n",
    "import os \n",
    "import mne\n",
    "from torch.utils.data import DataLoader\n",
    "from torch.autograd import Variable\n",
    "import torch.autograd as autograd\n",
    "from torchvision.models import vgg19\n",
    "\n",
    "from torch import nn\n",
    "from torch import Tensor\n",
    "from PIL import Image\n",
    "from torchvision.transforms import Compose, Resize, ToTensor\n",
    "from einops import rearrange, reduce, repeat\n",
    "from einops.layers.torch import Rearrange, Reduce"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importation & Pre-processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "# Chemin vers le dossier parent\n",
    "dossier_parent = \"C:/Users/MAREZ10/OneDrive - Université Laval/Bureau/Projet Transformers/eeg_data\"\n",
    "\n",
    "# Dictionnaire pour stocker les données MNE par sous-dossier\n",
    "donnees_par_sous_dossier = {}\n",
    "\n",
    "# Parcours des sous-dossiers\n",
    "for nom_sous_dossier in os.listdir(dossier_parent):\n",
    "    \n",
    "    chemin_sous_dossier = os.path.join(dossier_parent, nom_sous_dossier)\n",
    "    \n",
    "    if os.path.isdir(chemin_sous_dossier):\n",
    "        \n",
    "        donnees_par_sous_dossier[nom_sous_dossier] = []\n",
    "        \n",
    "        for nom_fichier in os.listdir(chemin_sous_dossier):\n",
    "            \n",
    "            if nom_fichier.endswith('.edf'):\n",
    "                \n",
    "                chemin_fichier = os.path.join(chemin_sous_dossier, nom_fichier)\n",
    "                donnees_mne = mne.io.read_raw_edf(chemin_fichier)\n",
    "                donnees_par_sous_dossier[nom_sous_dossier].append(donnees_mne)\n",
    "\n",
    "# Affichage des informations sur les données MNE\n",
    "for nom_sous_dossier, donnees_mne in donnees_par_sous_dossier.items():\n",
    "    print(f\"Sujet {nom_sous_dossier} : {len(donnees_mne)} fichiers .edf\")\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = list(donnees_par_sous_dossier.keys())\n",
    "Y = list(len(elem) for elem in donnees_par_sous_dossier.values())\n",
    "\n",
    "# Création de l'histogramme\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.bar(X,Y, color='skyblue')\n",
    "plt.xlabel('Sujets')\n",
    "plt.ylabel('Nombre d\\'enregistrements')\n",
    "plt.title('Nombre d\\'enregistrements par sujet')\n",
    "plt.xticks(rotation=45, ha='right')  # Rotation des étiquettes sur l'axe des x pour une meilleure lisibilité\n",
    "plt.tight_layout()  # Ajustement automatique du tracé pour éviter les chevauchements\n",
    "plt.show()\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Temps d'échantillonnage - Nombre d'enregistrements n'ayant pas une durée de 1 heure par sujet "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Dictionnaire pour stocker le nombre de fichiers par sous-dossier\n",
    "nombre_par_sous_dossier = {}\n",
    "\n",
    "# Parcours des sous-dossiers\n",
    "for nom_sous_dossier, donnees_mne_liste in donnees_par_sous_dossier.items():\n",
    "    # Initialisation du compteur pour ce sous-dossier\n",
    "    cpt = 0\n",
    "    # Parcours des données MNE dans ce sous-dossier\n",
    "    for donnees_mne in donnees_mne_liste:\n",
    "        # Vérification si la durée de la donnée MNE est inférieure à la valeur donnée\n",
    "        if donnees_mne.times[-1] < 3599.99609375:\n",
    "            # Incrémentation du compteur\n",
    "            cpt += 1\n",
    "    # Stockage du nombre dans le dictionnaire\n",
    "    nombre_par_sous_dossier[nom_sous_dossier] = cpt\n",
    "\n",
    "# Affichage du nombre de fichiers qui satisfont la condition pour chaque sous-dossier\n",
    "for nom_sous_dossier, nombre in nombre_par_sous_dossier.items():\n",
    "    print(f\"Sous-dossier {nom_sous_dossier} : {nombre} fichiers\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# Chemin d'accès au fichier\n",
    "file_path = \"C:/Users/MAREZ10/OneDrive - Université Laval/Bureau/Projet Transformers/eeg_data\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_summary(file_path):\n",
    "    # Initialisation des listes pour stocker les informations\n",
    "    seizure_presence = {}\n",
    "    \n",
    "    # Chemin d'accès au fichier\n",
    "    #file_path = \"chb04-summary.txt\"\n",
    "    \n",
    "    # Ouvrir et lire le fichier\n",
    "    is_seizure = False\n",
    "    all_files_str = []\n",
    "    seizure_start = 0\n",
    "    with open(file_path, \"r\") as file:\n",
    "        # Lire chaque ligne du fichier\n",
    "        for line in file:\n",
    "            # Traiter la ligne actuelle\n",
    "            if line.strip():\n",
    "                if \"File Name:\" in line:\n",
    "                  start = len(\"File Name: \")\n",
    "                  all_files_str.append(str(line[start:len(line)-1]))\n",
    "                elif (\"Seizure\" in line) and (\"Start Time: \" in line):\n",
    "                  if line[len(\"Seizure S\")-1] == \"S\":\n",
    "                    start = len(\"Seizure Start Time: \")\n",
    "                  else:\n",
    "                    start = len(\"Seizure 1 Start Time: \")\n",
    "                  end = len(\" seconds\")+1\n",
    "                  seizure_start = int(line[start:len(line)-end])\n",
    "                elif (\"Seizure\"in line) and (\"End Time: \" in line):\n",
    "                  if line[len(\"Seizure E\")-1] == \"E\":\n",
    "                    start = len(\"Seizure End Time: \")\n",
    "                  else:\n",
    "                    start = len(\"Seizure 1 End Time: \")\n",
    "                  end = len(\" seconds\")+1\n",
    "                  seizure_end = int(line[start:len(line)-end])\n",
    "                  if not all_files_str[len(all_files_str)-1] in seizure_presence.keys():\n",
    "                    seizure_presence[all_files_str[len(all_files_str)-1]] = []\n",
    "                  seizure_presence[all_files_str[len(all_files_str)-1]].append((seizure_start, seizure_end))\n",
    "    print(\"Seizure presence init\" ,seizure_presence)\n",
    "    return seizure_presence, all_files_str\n",
    "\n",
    "\n",
    "def separate_data_intervals(file_str, seizure_presence,path):\n",
    "  # Durée de chaque intervalle en secondes (2 minutes)\n",
    "  interval_duration = 60\n",
    "\n",
    "  raw = mne.io.read_raw_edf(path + file_str)\n",
    "  #data, times = raw[:, :]\n",
    "  total_duration = raw.times[-1] # en secondes\n",
    "\n",
    "  # Nombre total d'intervalle de 10 minutes\n",
    "  num_intervals = int(total_duration / interval_duration)\n",
    "  labels = []\n",
    "  data = []\n",
    "  \n",
    "  # print(\"Seizure presence: \", seizure_presence) \n",
    "  \n",
    "  # Diviser les données en intervalles de 10 minutes\n",
    "  for i in range(num_intervals):\n",
    "      # Calculer le temps de début et de fin de chaque intervalle\n",
    "      start_time = i * interval_duration\n",
    "      end_time = (i + 1) * interval_duration\n",
    "            \n",
    "\n",
    "      # Convertir le temps en indice\n",
    "      start_idx = raw.time_as_index(start_time)\n",
    "      end_idx = raw.time_as_index(end_time)\n",
    "\n",
    "      # Extraire les données de l'intervalle\n",
    "      interval_data, interval_times = raw[:, start_idx:end_idx]\n",
    "      data.append(interval_data)\n",
    "\n",
    "      \n",
    "      if file_str in seizure_presence.keys():\n",
    "        is_seizure = False\n",
    "        for start_seizure, end_seizure in seizure_presence[file_str]:\n",
    "\n",
    "\n",
    "          if (start_seizure >= start_time and start_seizure <= end_time) or (end_seizure >= start_time and end_seizure <= end_time) or (start_seizure <= start_time and end_seizure >= end_time):\n",
    "                \n",
    "            is_seizure = True\n",
    "  \n",
    "            break\n",
    "        if is_seizure:\n",
    "          labels.append(1)\n",
    "        else:\n",
    "          labels.append(0)\n",
    "      else:\n",
    "        labels.append(0)\n",
    "  return data, labels\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data_for_patient(file_path, patient):\n",
    "      \n",
    "    path = file_path +\"/\"+ patient + \"/\"\n",
    "    seizure_presence, all_files_str = read_summary(path+patient+\"-summary.txt\")\n",
    "    data = []\n",
    "    labels = []\n",
    "    for file in all_files_str:\n",
    "      print(\"Reading file : \", file)\n",
    "      interval_data, label = separate_data_intervals(file, seizure_presence,path)\n",
    "      print(\"End of file\")\n",
    "      for i in range(len(interval_data)):\n",
    "        data.append(interval_data[i])\n",
    "        labels.append(label[i])\n",
    "    return data, labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data, labels = load_data_for_patient(file_path, \"chb01\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0: 2377, 1: 15}\n"
     ]
    }
   ],
   "source": [
    "oc={}\n",
    "for elem in labels:\n",
    "    oc[elem] = labels.count(elem)\n",
    "\n",
    "print(oc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "data = np.array(data)\n",
    "labels = np.array(labels)\n",
    "\n",
    "train_data, test_data, train_labels, test_labels = train_test_split(data, labels, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1913, 23, 15360)\n",
      "(1913,)\n",
      "(479, 23, 15360)\n",
      "(479,)\n"
     ]
    }
   ],
   "source": [
    "print(train_data.shape)\n",
    "print(train_labels.shape)\n",
    "print(test_data.shape)\n",
    "print(test_labels.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import transformers\n",
    "import convolution\n",
    "from importlib import reload\n",
    "reload(transformers)\n",
    "reload(convolution)\n",
    "\n",
    "class Conformer(nn.Sequential):\n",
    "    def __init__(self, emb_size=40, nb_channels =23, depth=6, num_heads=2, n_classes=2, **kwargs):\n",
    "        super().__init__(\n",
    "\n",
    "            convolution.PatchEmbedding(emb_size, nb_channels),\n",
    "            transformers.TransformerEncoder(depth, emb_size, num_heads, out_size = 10, dropout=0.1),\n",
    "            transformers.TransformerClassifier(10, n_classes)\n",
    "        )\n",
    "\n",
    "batch_size = 4\n",
    "n_epochs = 1\n",
    "\n",
    "torch.cuda.empty_cache()\n",
    "model = Conformer().cuda()\n",
    "\n",
    "Tensor = torch.cuda.FloatTensor\n",
    "LongTensor = torch.cuda.LongTensor\n",
    "\n",
    "criterion = torch.nn.BCELoss()\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=0.001, betas=(0.5, 0.999))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn.functional as F\n",
    "import tqdm\n",
    "def train(train_data, train_label, test_data, test_label):\n",
    "\n",
    "        \n",
    "        train_data = torch.from_numpy(train_data)\n",
    "        train_label = torch.from_numpy(train_label)\n",
    "        dataset = torch.utils.data.TensorDataset(train_data, train_label)\n",
    "        dataloader = torch.utils.data.DataLoader(dataset=dataset, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "        test_data = torch.from_numpy(test_data)\n",
    "        test_label = torch.from_numpy(test_label)\n",
    "        #test_dataset = torch.utils.data.TensorDataset(test_data, test_label)\n",
    "        #test_dataloader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=True)\n",
    "\n",
    "\n",
    "        test_data = Variable(test_data.type(Tensor))\n",
    "        test_label = Variable(test_label.type(Tensor))\n",
    "\n",
    "        bestAcc = 0\n",
    "        averAcc = 0\n",
    "        num = 0\n",
    "        Y_true = 0\n",
    "        Y_pred = 0\n",
    "\n",
    "\n",
    "\n",
    "        for e in tqdm(range(n_epochs)):\n",
    "            model.train()\n",
    "            for i, (train_data, train_label) in enumerate(dataloader):\n",
    "\n",
    "                train_data = Variable(train_data.cuda().type(Tensor))\n",
    "                train_label = Variable(train_label.cuda().type(LongTensor))\n",
    "\n",
    "                outputs = model(train_data)\n",
    "                \n",
    "                train_label_one_hot = F.one_hot(train_label.view(-1), num_classes=2).float()\n",
    "                loss = criterion(outputs, train_label_one_hot) \n",
    "\n",
    "                optimizer.zero_grad()\n",
    "                loss.backward()\n",
    "                optimizer.step()\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "            # test process\n",
    "            if (e + 1) % 1 == 0:\n",
    "                model.eval()\n",
    "                Cls = model(test_data)\n",
    "\n",
    "                test_label_one_hot = F.one_hot(test_label.view(-1), num_classes=2).float()\n",
    "\n",
    "                loss_test = criterion(Cls, test_label_one_hot)\n",
    "                y_pred = torch.max(Cls, 1)[1]\n",
    "                acc = float((y_pred == test_label).cpu().numpy().astype(int).sum()) / float(test_label.size(0))\n",
    "                train_pred = torch.max(outputs, 1)[1]\n",
    "                train_acc = float((train_pred == train_label).cpu().numpy().astype(int).sum()) / float(train_label.size(0))\n",
    "\n",
    "                print('Epoch:', e,\n",
    "                      '  Train loss: %.6f' % loss.detach().cpu().numpy(),\n",
    "                      '  Test loss: %.6f' % loss_test.detach().cpu().numpy(),\n",
    "                      '  Train accuracy %.6f' % train_acc,\n",
    "                      '  Test accuracy is %.6f' % acc)\n",
    "\n",
    "                \n",
    "                num = num + 1\n",
    "                averAcc = averAcc + acc\n",
    "                if acc > bestAcc:\n",
    "                    bestAcc = acc\n",
    "                    Y_true = test_label\n",
    "                    Y_pred = y_pred\n",
    "\n",
    "\n",
    "        torch.save(model.module.state_dict(), 'model.pth')\n",
    "        averAcc = averAcc / num\n",
    "        print('The average accuracy is:', averAcc)\n",
    "        print('The best accuracy is:', bestAcc)\n",
    "        \n",
    "\n",
    "        return bestAcc, averAcc, Y_true, Y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.4887, 0.5263],\n",
      "        [0.4984, 0.5335],\n",
      "        [0.5027, 0.5346],\n",
      "        [0.4929, 0.5251]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.5873, 0.3992],\n",
      "        [0.5754, 0.3916],\n",
      "        [0.5658, 0.3875],\n",
      "        [0.5736, 0.4126]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.6858, 0.2869],\n",
      "        [0.6932, 0.2666],\n",
      "        [0.6871, 0.3088],\n",
      "        [0.6874, 0.2917]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.7819, 0.1814],\n",
      "        [0.7890, 0.1838],\n",
      "        [0.7907, 0.1855],\n",
      "        [0.7856, 0.1899]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.8687, 0.1104],\n",
      "        [0.8649, 0.1300],\n",
      "        [0.8530, 0.1135],\n",
      "        [0.8586, 0.1206]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9241, 0.0607],\n",
      "        [0.9251, 0.0716],\n",
      "        [0.9329, 0.0599],\n",
      "        [0.9205, 0.0754]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9414, 0.0369],\n",
      "        [0.9371, 0.0405],\n",
      "        [0.9520, 0.0428],\n",
      "        [0.9495, 0.0442]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9605, 0.0256],\n",
      "        [0.9686, 0.0272],\n",
      "        [0.9696, 0.0253],\n",
      "        [0.9680, 0.0247]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9799, 0.0166],\n",
      "        [0.9794, 0.0178],\n",
      "        [0.9802, 0.0154],\n",
      "        [0.9740, 0.0179]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9871, 0.0127],\n",
      "        [0.9837, 0.0136],\n",
      "        [0.9866, 0.0118],\n",
      "        [0.9886, 0.0110]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9868, 0.0102],\n",
      "        [0.9893, 0.0109],\n",
      "        [0.9889, 0.0130],\n",
      "        [0.9875, 0.0118]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9879, 0.0050],\n",
      "        [0.9913, 0.0060],\n",
      "        [0.9882, 0.0068],\n",
      "        [0.9917, 0.0067]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9896, 0.0059],\n",
      "        [0.9916, 0.0047],\n",
      "        [0.9921, 0.0066],\n",
      "        [0.9902, 0.0058]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9945, 0.0048],\n",
      "        [0.9958, 0.0032],\n",
      "        [0.9937, 0.0074],\n",
      "        [0.9945, 0.0053]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9951, 0.0041],\n",
      "        [0.9941, 0.0059],\n",
      "        [0.9946, 0.0040],\n",
      "        [0.9942, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9957, 0.0032],\n",
      "        [0.9965, 0.0026],\n",
      "        [0.9961, 0.0030],\n",
      "        [0.9965, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9958, 0.0035],\n",
      "        [0.9942, 0.0049],\n",
      "        [0.9945, 0.0032],\n",
      "        [0.9963, 0.0032]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9958, 0.0027],\n",
      "        [0.9960, 0.0024],\n",
      "        [0.9972, 0.0022],\n",
      "        [0.9970, 0.0027]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9960, 0.0037],\n",
      "        [0.9963, 0.0031],\n",
      "        [0.9973, 0.0022],\n",
      "        [0.9969, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0023],\n",
      "        [0.9977, 0.0022],\n",
      "        [0.9981, 0.0022],\n",
      "        [0.9968, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0036],\n",
      "        [0.9965, 0.0020],\n",
      "        [0.9963, 0.0025],\n",
      "        [0.9982, 0.0019]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9968, 0.0039],\n",
      "        [0.9973, 0.0022],\n",
      "        [0.9974, 0.0013],\n",
      "        [0.9981, 0.0024]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0018],\n",
      "        [0.9983, 0.0019],\n",
      "        [0.9974, 0.0016],\n",
      "        [0.9970, 0.0014]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0022],\n",
      "        [0.9985, 0.0011],\n",
      "        [0.9979, 0.0018],\n",
      "        [0.9977, 0.0018]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0013],\n",
      "        [0.9973, 0.0024],\n",
      "        [0.9985, 0.0014],\n",
      "        [0.9982, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9808e-01, 1.2670e-03],\n",
      "        [9.9869e-01, 8.3144e-04],\n",
      "        [9.9846e-01, 7.8021e-04],\n",
      "        [9.9851e-01, 1.1734e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9988, 0.0012],\n",
      "        [0.9983, 0.0013],\n",
      "        [0.9985, 0.0014],\n",
      "        [0.9984, 0.0020]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9895e-01, 1.3195e-03],\n",
      "        [9.9877e-01, 1.3147e-03],\n",
      "        [9.9892e-01, 6.5439e-04],\n",
      "        [9.9875e-01, 1.5626e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9842e-01, 7.9335e-04],\n",
      "        [9.9868e-01, 1.2708e-03],\n",
      "        [9.9874e-01, 1.1433e-03],\n",
      "        [9.9849e-01, 7.7499e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9916e-01, 7.3640e-04],\n",
      "        [9.9890e-01, 8.7915e-04],\n",
      "        [9.9885e-01, 6.8777e-04],\n",
      "        [9.9858e-01, 7.8636e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9934e-01, 8.1063e-04],\n",
      "        [9.9894e-01, 7.3750e-04],\n",
      "        [9.9816e-01, 1.2736e-03],\n",
      "        [9.9843e-01, 9.6418e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9901e-01, 9.0426e-04],\n",
      "        [9.9869e-01, 1.0964e-03],\n",
      "        [9.9899e-01, 8.7469e-04],\n",
      "        [9.9871e-01, 1.1463e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9857e-01, 9.9252e-04],\n",
      "        [9.9894e-01, 8.0462e-04],\n",
      "        [9.9880e-01, 8.4119e-04],\n",
      "        [9.9910e-01, 6.0218e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9899e-01, 1.0100e-03],\n",
      "        [9.9899e-01, 6.1635e-04],\n",
      "        [9.9925e-01, 5.8756e-04],\n",
      "        [9.9919e-01, 1.0016e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9897e-01, 6.6051e-04],\n",
      "        [9.9923e-01, 8.6062e-04],\n",
      "        [9.9918e-01, 6.2178e-04],\n",
      "        [9.9896e-01, 6.9486e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9900e-01, 7.4229e-04],\n",
      "        [9.9909e-01, 6.1348e-04],\n",
      "        [9.9843e-01, 6.1449e-04],\n",
      "        [9.9905e-01, 6.9577e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9890e-01, 9.7611e-04],\n",
      "        [9.9871e-01, 5.5951e-04],\n",
      "        [9.9941e-01, 4.2668e-04],\n",
      "        [9.9920e-01, 6.1916e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9873e-01, 5.1508e-04],\n",
      "        [9.9923e-01, 7.8536e-04],\n",
      "        [9.9934e-01, 6.7892e-04],\n",
      "        [9.9916e-01, 3.3725e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9918e-01, 5.5370e-04],\n",
      "        [9.9914e-01, 8.8429e-04],\n",
      "        [9.9916e-01, 1.1421e-03],\n",
      "        [9.9932e-01, 6.4195e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9938e-01, 6.4464e-04],\n",
      "        [9.9932e-01, 6.3480e-04],\n",
      "        [9.9921e-01, 7.8556e-04],\n",
      "        [9.9904e-01, 7.0268e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9942e-01, 5.3864e-04],\n",
      "        [9.9853e-01, 6.9824e-04],\n",
      "        [9.9889e-01, 6.2483e-04],\n",
      "        [9.9900e-01, 8.9457e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9936e-01, 3.9531e-04],\n",
      "        [9.9959e-01, 5.7827e-04],\n",
      "        [9.9932e-01, 5.8178e-04],\n",
      "        [9.9933e-01, 7.4076e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9955e-01, 4.1909e-04],\n",
      "        [9.9957e-01, 4.1704e-04],\n",
      "        [9.9953e-01, 4.3147e-04],\n",
      "        [9.9955e-01, 2.5846e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9932e-01, 3.7189e-04],\n",
      "        [9.9950e-01, 4.7091e-04],\n",
      "        [9.9950e-01, 5.8869e-04],\n",
      "        [9.9957e-01, 3.5798e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9944e-01, 4.5393e-04],\n",
      "        [9.9950e-01, 3.7689e-04],\n",
      "        [9.9897e-01, 4.8637e-04],\n",
      "        [9.9945e-01, 6.0817e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9944e-01, 2.7329e-04],\n",
      "        [9.9939e-01, 2.7226e-04],\n",
      "        [9.9934e-01, 3.2201e-04],\n",
      "        [9.9937e-01, 5.4122e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9938e-01, 4.4573e-04],\n",
      "        [9.9974e-01, 2.8678e-04],\n",
      "        [9.9942e-01, 6.3714e-04],\n",
      "        [9.9958e-01, 3.6230e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9867, 0.0163],\n",
      "        [0.9905, 0.0144],\n",
      "        [0.9882, 0.0104],\n",
      "        [0.9902, 0.0120]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9775, 0.0438],\n",
      "        [0.9718, 0.0489],\n",
      "        [0.9739, 0.0339],\n",
      "        [0.9748, 0.0340]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9710, 0.0334],\n",
      "        [0.9693, 0.0371],\n",
      "        [0.9694, 0.0326],\n",
      "        [0.9708, 0.0341]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9704, 0.0264],\n",
      "        [0.9685, 0.0277],\n",
      "        [0.9638, 0.0432],\n",
      "        [0.9613, 0.0338]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9745, 0.0174],\n",
      "        [0.9763, 0.0190],\n",
      "        [0.9693, 0.0224],\n",
      "        [0.9714, 0.0228]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9873, 0.0126],\n",
      "        [0.9762, 0.0186],\n",
      "        [0.9822, 0.0158],\n",
      "        [0.9763, 0.0176]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9877, 0.0084],\n",
      "        [0.9858, 0.0095],\n",
      "        [0.9877, 0.0105],\n",
      "        [0.9879, 0.0111]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9902, 0.0066],\n",
      "        [0.9874, 0.0070],\n",
      "        [0.9909, 0.0072],\n",
      "        [0.9911, 0.0054]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9898, 0.0062],\n",
      "        [0.9925, 0.0075],\n",
      "        [0.9909, 0.0070],\n",
      "        [0.9927, 0.0066]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9936, 0.0051],\n",
      "        [0.9914, 0.0058],\n",
      "        [0.9943, 0.0074],\n",
      "        [0.9921, 0.0077]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9943, 0.0041],\n",
      "        [0.9956, 0.0042],\n",
      "        [0.9949, 0.0040],\n",
      "        [0.9943, 0.0043]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9958, 0.0031],\n",
      "        [0.9936, 0.0041],\n",
      "        [0.9957, 0.0038],\n",
      "        [0.9963, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9949, 0.0032],\n",
      "        [0.9959, 0.0028],\n",
      "        [0.9955, 0.0040],\n",
      "        [0.9960, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9968, 0.0020],\n",
      "        [0.9966, 0.0027],\n",
      "        [0.9956, 0.0023],\n",
      "        [0.9956, 0.0036]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0033],\n",
      "        [0.9972, 0.0023],\n",
      "        [0.9965, 0.0022],\n",
      "        [0.9957, 0.0037]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0025],\n",
      "        [0.9965, 0.0042],\n",
      "        [0.9969, 0.0023],\n",
      "        [0.9977, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0032],\n",
      "        [0.9979, 0.0021],\n",
      "        [0.9937, 0.0031],\n",
      "        [0.9983, 0.0014]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0033],\n",
      "        [0.9960, 0.0028],\n",
      "        [0.9977, 0.0029],\n",
      "        [0.9974, 0.0015]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0020],\n",
      "        [0.9985, 0.0020],\n",
      "        [0.9976, 0.0029],\n",
      "        [0.9963, 0.0029]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0014],\n",
      "        [0.9966, 0.0026],\n",
      "        [0.9986, 0.0014],\n",
      "        [0.9981, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9968, 0.0018],\n",
      "        [0.9975, 0.0012],\n",
      "        [0.9978, 0.0016],\n",
      "        [0.9976, 0.0015]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0025],\n",
      "        [0.9982, 0.0015],\n",
      "        [0.9974, 0.0026],\n",
      "        [0.9980, 0.0023]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([1, 0, 0, 0], device='cuda:0')\n",
      "tensor([[0., 1.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9850, 0.0191],\n",
      "        [0.9829, 0.0157],\n",
      "        [0.9865, 0.0145],\n",
      "        [0.9870, 0.0205]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9792, 0.0224],\n",
      "        [0.9733, 0.0284],\n",
      "        [0.9706, 0.0316],\n",
      "        [0.9709, 0.0344]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9648, 0.0379],\n",
      "        [0.9741, 0.0324],\n",
      "        [0.9729, 0.0259],\n",
      "        [0.9690, 0.0307]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9552, 0.0426],\n",
      "        [0.9735, 0.0281],\n",
      "        [0.9552, 0.0418],\n",
      "        [0.9690, 0.0235]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9712, 0.0173],\n",
      "        [0.9710, 0.0259],\n",
      "        [0.9743, 0.0207],\n",
      "        [0.9720, 0.0324]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9842, 0.0143],\n",
      "        [0.9759, 0.0120],\n",
      "        [0.9814, 0.0168],\n",
      "        [0.9795, 0.0133]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 1, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [0., 1.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9498, 0.0483],\n",
      "        [0.9472, 0.0566],\n",
      "        [0.9413, 0.0609],\n",
      "        [0.9441, 0.0606]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9449, 0.0669],\n",
      "        [0.9486, 0.0702],\n",
      "        [0.9317, 0.0830],\n",
      "        [0.9305, 0.0809]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9456, 0.0606],\n",
      "        [0.9426, 0.0696],\n",
      "        [0.9439, 0.0707],\n",
      "        [0.9360, 0.0553]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9540, 0.0375],\n",
      "        [0.9554, 0.0540],\n",
      "        [0.9610, 0.0359],\n",
      "        [0.9495, 0.0451]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9691, 0.0262],\n",
      "        [0.9618, 0.0254],\n",
      "        [0.9741, 0.0223],\n",
      "        [0.9688, 0.0321]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9759, 0.0147],\n",
      "        [0.9695, 0.0172],\n",
      "        [0.9730, 0.0159],\n",
      "        [0.9788, 0.0245]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9845, 0.0139],\n",
      "        [0.9767, 0.0126],\n",
      "        [0.9854, 0.0121],\n",
      "        [0.9836, 0.0103]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9867, 0.0293],\n",
      "        [0.9885, 0.0097],\n",
      "        [0.9915, 0.0083],\n",
      "        [0.9856, 0.0118]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9885, 0.0089],\n",
      "        [0.9853, 0.0116],\n",
      "        [0.9905, 0.0069],\n",
      "        [0.9925, 0.0065]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9888, 0.0094],\n",
      "        [0.9923, 0.0068],\n",
      "        [0.9895, 0.0157],\n",
      "        [0.9898, 0.0121]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9905, 0.0064],\n",
      "        [0.9891, 0.0057],\n",
      "        [0.9937, 0.0128],\n",
      "        [0.9906, 0.0054]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9940, 0.0040],\n",
      "        [0.9933, 0.0113],\n",
      "        [0.9925, 0.0056],\n",
      "        [0.9907, 0.0058]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0050],\n",
      "        [0.9945, 0.0056],\n",
      "        [0.9932, 0.0048],\n",
      "        [0.9941, 0.0046]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9962, 0.0038],\n",
      "        [0.9942, 0.0029],\n",
      "        [0.9953, 0.0072],\n",
      "        [0.9948, 0.0049]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9960, 0.0029],\n",
      "        [0.9922, 0.0064],\n",
      "        [0.9948, 0.0041],\n",
      "        [0.9952, 0.0042]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9969, 0.0037],\n",
      "        [0.9956, 0.0076],\n",
      "        [0.9955, 0.0036],\n",
      "        [0.9962, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9944, 0.0034],\n",
      "        [0.9940, 0.0039],\n",
      "        [0.9970, 0.0033],\n",
      "        [0.9954, 0.0038]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0034],\n",
      "        [0.9947, 0.0030],\n",
      "        [0.9966, 0.0025],\n",
      "        [0.9961, 0.0030]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0028],\n",
      "        [0.9969, 0.0020],\n",
      "        [0.9945, 0.0036],\n",
      "        [0.9955, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9972, 0.0024],\n",
      "        [0.9961, 0.0027],\n",
      "        [0.9974, 0.0027],\n",
      "        [0.9972, 0.0023]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0025],\n",
      "        [0.9971, 0.0023],\n",
      "        [0.9958, 0.0028],\n",
      "        [0.9975, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0023],\n",
      "        [0.9975, 0.0026],\n",
      "        [0.9965, 0.0030],\n",
      "        [0.9976, 0.0039]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0020],\n",
      "        [0.9966, 0.0024],\n",
      "        [0.9959, 0.0020],\n",
      "        [0.9959, 0.0034]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0016],\n",
      "        [0.9971, 0.0023],\n",
      "        [0.9974, 0.0022],\n",
      "        [0.9975, 0.0024]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9981, 0.0015],\n",
      "        [0.9976, 0.0024],\n",
      "        [0.9982, 0.0015],\n",
      "        [0.9981, 0.0019]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9963, 0.0054],\n",
      "        [0.9983, 0.0022],\n",
      "        [0.9981, 0.0024],\n",
      "        [0.9975, 0.0027]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0016],\n",
      "        [0.9975, 0.0018],\n",
      "        [0.9981, 0.0014],\n",
      "        [0.9979, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0017],\n",
      "        [0.9983, 0.0015],\n",
      "        [0.9973, 0.0016],\n",
      "        [0.9982, 0.0011]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9978, 0.0022],\n",
      "        [0.9982, 0.0014],\n",
      "        [0.9984, 0.0016],\n",
      "        [0.9985, 0.0015]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9985, 0.0011],\n",
      "        [0.9971, 0.0023],\n",
      "        [0.9985, 0.0014],\n",
      "        [0.9986, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0025],\n",
      "        [0.9979, 0.0017],\n",
      "        [0.9975, 0.0024],\n",
      "        [0.9985, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9962, 0.0025],\n",
      "        [0.9978, 0.0018],\n",
      "        [0.9982, 0.0016],\n",
      "        [0.9972, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9881e-01, 9.6759e-04],\n",
      "        [9.9818e-01, 1.9592e-03],\n",
      "        [9.9888e-01, 1.5948e-03],\n",
      "        [9.9867e-01, 1.0776e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9985, 0.0015],\n",
      "        [0.9983, 0.0015],\n",
      "        [0.9985, 0.0016],\n",
      "        [0.9982, 0.0019]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9901e-01, 9.3755e-04],\n",
      "        [9.9782e-01, 1.7179e-03],\n",
      "        [9.9729e-01, 2.0630e-03],\n",
      "        [9.9889e-01, 1.0796e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9989, 0.0014],\n",
      "        [0.9984, 0.0011],\n",
      "        [0.9990, 0.0013],\n",
      "        [0.9985, 0.0011]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9874e-01, 9.8866e-04],\n",
      "        [9.9849e-01, 1.5676e-03],\n",
      "        [9.9787e-01, 1.3946e-03],\n",
      "        [9.9854e-01, 1.3198e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9869e-01, 1.5611e-03],\n",
      "        [9.9881e-01, 1.1546e-03],\n",
      "        [9.9727e-01, 1.7860e-03],\n",
      "        [9.9897e-01, 9.0457e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9918e-01, 1.0251e-03],\n",
      "        [9.9889e-01, 1.1416e-03],\n",
      "        [9.9867e-01, 8.8739e-04],\n",
      "        [9.9908e-01, 1.1429e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9989, 0.0021],\n",
      "        [0.9982, 0.0014],\n",
      "        [0.9986, 0.0011],\n",
      "        [0.9987, 0.0011]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9986, 0.0013],\n",
      "        [0.9986, 0.0014],\n",
      "        [0.9989, 0.0011],\n",
      "        [0.9984, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9892e-01, 1.0601e-03],\n",
      "        [9.9867e-01, 2.1188e-03],\n",
      "        [9.9875e-01, 8.3471e-04],\n",
      "        [9.9830e-01, 1.3675e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 1, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [0., 1.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9863, 0.0098],\n",
      "        [0.9896, 0.0092],\n",
      "        [0.9886, 0.0112],\n",
      "        [0.9896, 0.0115]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9859, 0.0183],\n",
      "        [0.9819, 0.0370],\n",
      "        [0.9826, 0.0201],\n",
      "        [0.9755, 0.0221]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9713, 0.0312],\n",
      "        [0.9756, 0.0237],\n",
      "        [0.9746, 0.0308],\n",
      "        [0.9676, 0.0254]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9743, 0.0248],\n",
      "        [0.9706, 0.0244],\n",
      "        [0.9779, 0.0248],\n",
      "        [0.9773, 0.0240]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9747, 0.0252],\n",
      "        [0.9817, 0.0164],\n",
      "        [0.9780, 0.0244],\n",
      "        [0.9777, 0.0227]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9774, 0.0164],\n",
      "        [0.9841, 0.0163],\n",
      "        [0.9786, 0.0203],\n",
      "        [0.9866, 0.0182]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9819, 0.0147],\n",
      "        [0.9775, 0.0154],\n",
      "        [0.9863, 0.0118],\n",
      "        [0.9843, 0.0137]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9857, 0.0122],\n",
      "        [0.9901, 0.0096],\n",
      "        [0.9863, 0.0224],\n",
      "        [0.9823, 0.0192]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9906, 0.0082],\n",
      "        [0.9902, 0.0088],\n",
      "        [0.9821, 0.0130],\n",
      "        [0.9900, 0.0068]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9919, 0.0171],\n",
      "        [0.9908, 0.0077],\n",
      "        [0.9921, 0.0086],\n",
      "        [0.9934, 0.0060]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9866, 0.0089],\n",
      "        [0.9893, 0.0073],\n",
      "        [0.9924, 0.0058],\n",
      "        [0.9929, 0.0053]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9942, 0.0062],\n",
      "        [0.9839, 0.0110],\n",
      "        [0.9940, 0.0072],\n",
      "        [0.9931, 0.0078]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9914, 0.0049],\n",
      "        [0.9852, 0.0115],\n",
      "        [0.9916, 0.0064],\n",
      "        [0.9917, 0.0068]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0043],\n",
      "        [0.9951, 0.0048],\n",
      "        [0.9943, 0.0044],\n",
      "        [0.9929, 0.0056]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9954, 0.0036],\n",
      "        [0.9955, 0.0042],\n",
      "        [0.9956, 0.0049],\n",
      "        [0.9896, 0.0054]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0032],\n",
      "        [0.9962, 0.0040],\n",
      "        [0.9963, 0.0038],\n",
      "        [0.9908, 0.0054]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9945, 0.0041],\n",
      "        [0.9959, 0.0032],\n",
      "        [0.9943, 0.0042],\n",
      "        [0.9967, 0.0034]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9936, 0.0057],\n",
      "        [0.9940, 0.0070],\n",
      "        [0.9957, 0.0037],\n",
      "        [0.9953, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9966, 0.0029],\n",
      "        [0.9976, 0.0028],\n",
      "        [0.9967, 0.0023],\n",
      "        [0.9956, 0.0032]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0030],\n",
      "        [0.9970, 0.0027],\n",
      "        [0.9970, 0.0029],\n",
      "        [0.9908, 0.0047]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9975, 0.0025],\n",
      "        [0.9967, 0.0035],\n",
      "        [0.9971, 0.0020],\n",
      "        [0.9944, 0.0038]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0034],\n",
      "        [0.9965, 0.0034],\n",
      "        [0.9982, 0.0021],\n",
      "        [0.9972, 0.0025]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0034],\n",
      "        [0.9948, 0.0032],\n",
      "        [0.9981, 0.0018],\n",
      "        [0.9977, 0.0030]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9966, 0.0020],\n",
      "        [0.9974, 0.0027],\n",
      "        [0.9973, 0.0022],\n",
      "        [0.9966, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0024],\n",
      "        [0.9979, 0.0019],\n",
      "        [0.9963, 0.0024],\n",
      "        [0.9975, 0.0019]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0025],\n",
      "        [0.9954, 0.0032],\n",
      "        [0.9953, 0.0031],\n",
      "        [0.9977, 0.0052]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9984, 0.0015],\n",
      "        [0.9974, 0.0018],\n",
      "        [0.9980, 0.0017],\n",
      "        [0.9963, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0021],\n",
      "        [0.9984, 0.0018],\n",
      "        [0.9976, 0.0059],\n",
      "        [0.9976, 0.0025]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9984, 0.0015],\n",
      "        [0.9983, 0.0021],\n",
      "        [0.9980, 0.0018],\n",
      "        [0.9970, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0015],\n",
      "        [0.9979, 0.0017],\n",
      "        [0.9984, 0.0016],\n",
      "        [0.9972, 0.0023]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0015],\n",
      "        [0.9984, 0.0022],\n",
      "        [0.9975, 0.0013],\n",
      "        [0.9987, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0015],\n",
      "        [0.9983, 0.0014],\n",
      "        [0.9983, 0.0013],\n",
      "        [0.9981, 0.0018]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9972, 0.0023],\n",
      "        [0.9963, 0.0023],\n",
      "        [0.9986, 0.0010],\n",
      "        [0.9984, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0018],\n",
      "        [0.9987, 0.0013],\n",
      "        [0.9987, 0.0014],\n",
      "        [0.9979, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0020],\n",
      "        [0.9988, 0.0011],\n",
      "        [0.9982, 0.0013],\n",
      "        [0.9984, 0.0038]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0012],\n",
      "        [0.9991, 0.0011],\n",
      "        [0.9985, 0.0012],\n",
      "        [0.9983, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9981, 0.0014],\n",
      "        [0.9988, 0.0012],\n",
      "        [0.9984, 0.0019],\n",
      "        [0.9990, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9864, 0.0138],\n",
      "        [0.9897, 0.0135],\n",
      "        [0.9907, 0.0137],\n",
      "        [0.9908, 0.0111]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9703, 0.0291],\n",
      "        [0.9787, 0.0243],\n",
      "        [0.9787, 0.0319],\n",
      "        [0.9749, 0.0440]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9696, 0.0305],\n",
      "        [0.9669, 0.0312],\n",
      "        [0.9749, 0.0290],\n",
      "        [0.9731, 0.0337]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9736, 0.0281],\n",
      "        [0.9737, 0.0247],\n",
      "        [0.9696, 0.0270],\n",
      "        [0.9658, 0.0256]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9681, 0.0229],\n",
      "        [0.9785, 0.0255],\n",
      "        [0.9814, 0.0224],\n",
      "        [0.9778, 0.0208]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9801, 0.0207],\n",
      "        [0.9769, 0.0185],\n",
      "        [0.9788, 0.0177],\n",
      "        [0.9796, 0.0182]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9804, 0.0293],\n",
      "        [0.9819, 0.0156],\n",
      "        [0.9730, 0.0178],\n",
      "        [0.9857, 0.0168]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9842, 0.0129],\n",
      "        [0.9813, 0.0163],\n",
      "        [0.9817, 0.0150],\n",
      "        [0.9769, 0.0128]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9838, 0.0200],\n",
      "        [0.9850, 0.0104],\n",
      "        [0.9862, 0.0117],\n",
      "        [0.9902, 0.0104]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9852, 0.0092],\n",
      "        [0.9910, 0.0064],\n",
      "        [0.9855, 0.0095],\n",
      "        [0.9919, 0.0064]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9923, 0.0062],\n",
      "        [0.9913, 0.0076],\n",
      "        [0.9894, 0.0129],\n",
      "        [0.9819, 0.0101]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9937, 0.0053],\n",
      "        [0.9927, 0.0059],\n",
      "        [0.9865, 0.0096],\n",
      "        [0.9873, 0.0092]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9932, 0.0061],\n",
      "        [0.9923, 0.0053],\n",
      "        [0.9935, 0.0072],\n",
      "        [0.9907, 0.0065]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9926, 0.0064],\n",
      "        [0.9959, 0.0034],\n",
      "        [0.9950, 0.0038],\n",
      "        [0.9937, 0.0052]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9941, 0.0057],\n",
      "        [0.9950, 0.0056],\n",
      "        [0.9954, 0.0053],\n",
      "        [0.9943, 0.0059]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9948, 0.0052],\n",
      "        [0.9963, 0.0034],\n",
      "        [0.9966, 0.0036],\n",
      "        [0.9909, 0.0058]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9929, 0.0047],\n",
      "        [0.9960, 0.0037],\n",
      "        [0.9953, 0.0043],\n",
      "        [0.9958, 0.0059]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9948, 0.0045],\n",
      "        [0.9946, 0.0097],\n",
      "        [0.9966, 0.0038],\n",
      "        [0.9956, 0.0043]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9967, 0.0028],\n",
      "        [0.9971, 0.0072],\n",
      "        [0.9970, 0.0026],\n",
      "        [0.9965, 0.0036]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9947, 0.0052],\n",
      "        [0.9974, 0.0028],\n",
      "        [0.9959, 0.0032],\n",
      "        [0.9894, 0.0077]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0033],\n",
      "        [0.9955, 0.0032],\n",
      "        [0.9970, 0.0034],\n",
      "        [0.9960, 0.0067]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0028],\n",
      "        [0.9953, 0.0036],\n",
      "        [0.9972, 0.0020],\n",
      "        [0.9974, 0.0030]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0021],\n",
      "        [0.9972, 0.0030],\n",
      "        [0.9969, 0.0020],\n",
      "        [0.9971, 0.0029]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9818, 0.0191],\n",
      "        [0.9891, 0.0140],\n",
      "        [0.9857, 0.0153],\n",
      "        [0.9823, 0.0209]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9584, 0.0306],\n",
      "        [0.9803, 0.0255],\n",
      "        [0.9681, 0.0338],\n",
      "        [0.9786, 0.0288]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9783, 0.0502],\n",
      "        [0.9727, 0.0357],\n",
      "        [0.9745, 0.0305],\n",
      "        [0.9638, 0.0329]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9656, 0.0518],\n",
      "        [0.9712, 0.0246],\n",
      "        [0.9724, 0.0254],\n",
      "        [0.9719, 0.0368]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9661, 0.0383],\n",
      "        [0.9742, 0.0245],\n",
      "        [0.9772, 0.0229],\n",
      "        [0.9801, 0.0218]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9800, 0.0187],\n",
      "        [0.9761, 0.0213],\n",
      "        [0.9785, 0.0168],\n",
      "        [0.9799, 0.0264]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9811, 0.0125],\n",
      "        [0.9807, 0.0149],\n",
      "        [0.9857, 0.0126],\n",
      "        [0.9815, 0.0192]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9897, 0.0105],\n",
      "        [0.9868, 0.0107],\n",
      "        [0.9887, 0.0099],\n",
      "        [0.9862, 0.0124]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9879, 0.0097],\n",
      "        [0.9826, 0.0134],\n",
      "        [0.9849, 0.0147],\n",
      "        [0.9882, 0.0096]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9850, 0.0081],\n",
      "        [0.9904, 0.0082],\n",
      "        [0.9772, 0.0117],\n",
      "        [0.9876, 0.0080]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9909, 0.0130],\n",
      "        [0.9915, 0.0083],\n",
      "        [0.9908, 0.0117],\n",
      "        [0.9915, 0.0071]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9926, 0.0068],\n",
      "        [0.9924, 0.0086],\n",
      "        [0.9921, 0.0070],\n",
      "        [0.9924, 0.0092]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9926, 0.0057],\n",
      "        [0.9934, 0.0061],\n",
      "        [0.9918, 0.0059],\n",
      "        [0.9899, 0.0101]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9942, 0.0047],\n",
      "        [0.9917, 0.0070],\n",
      "        [0.9892, 0.0068],\n",
      "        [0.9938, 0.0063]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9932, 0.0062],\n",
      "        [0.9944, 0.0058],\n",
      "        [0.9947, 0.0037],\n",
      "        [0.9943, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9938, 0.0073],\n",
      "        [0.9964, 0.0034],\n",
      "        [0.9951, 0.0055],\n",
      "        [0.9950, 0.0047]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9963, 0.0044],\n",
      "        [0.9951, 0.0047],\n",
      "        [0.9951, 0.0083],\n",
      "        [0.9959, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9935, 0.0045],\n",
      "        [0.9957, 0.0051],\n",
      "        [0.9953, 0.0039],\n",
      "        [0.9966, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0043],\n",
      "        [0.9953, 0.0041],\n",
      "        [0.9956, 0.0037],\n",
      "        [0.9954, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0032],\n",
      "        [0.9964, 0.0030],\n",
      "        [0.9952, 0.0042],\n",
      "        [0.9940, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9948, 0.0047],\n",
      "        [0.9955, 0.0039],\n",
      "        [0.9934, 0.0043],\n",
      "        [0.9969, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9942, 0.0042],\n",
      "        [0.9964, 0.0034],\n",
      "        [0.9963, 0.0036],\n",
      "        [0.9968, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0024],\n",
      "        [0.9969, 0.0062],\n",
      "        [0.9974, 0.0028],\n",
      "        [0.9959, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9975, 0.0027],\n",
      "        [0.9967, 0.0075],\n",
      "        [0.9971, 0.0024],\n",
      "        [0.9970, 0.0025]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0023],\n",
      "        [0.9973, 0.0023],\n",
      "        [0.9946, 0.0049],\n",
      "        [0.9973, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0021],\n",
      "        [0.9950, 0.0031],\n",
      "        [0.9964, 0.0025],\n",
      "        [0.9971, 0.0020]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0017],\n",
      "        [0.9979, 0.0028],\n",
      "        [0.9972, 0.0025],\n",
      "        [0.9972, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9972, 0.0024],\n",
      "        [0.9977, 0.0019],\n",
      "        [0.9978, 0.0025],\n",
      "        [0.9979, 0.0027]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0016],\n",
      "        [0.9980, 0.0023],\n",
      "        [0.9970, 0.0031],\n",
      "        [0.9981, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9968, 0.0021],\n",
      "        [0.9969, 0.0019],\n",
      "        [0.9983, 0.0021],\n",
      "        [0.9973, 0.0027]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0023],\n",
      "        [0.9950, 0.0033],\n",
      "        [0.9979, 0.0035],\n",
      "        [0.9971, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9871, 0.0133],\n",
      "        [0.9826, 0.0119],\n",
      "        [0.9904, 0.0096],\n",
      "        [0.9881, 0.0117]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9799, 0.0203],\n",
      "        [0.9792, 0.0189],\n",
      "        [0.9724, 0.0274],\n",
      "        [0.9790, 0.0205]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9768, 0.0285],\n",
      "        [0.9751, 0.0258],\n",
      "        [0.9741, 0.0269],\n",
      "        [0.9756, 0.0257]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9758, 0.0257],\n",
      "        [0.9787, 0.0226],\n",
      "        [0.9788, 0.0247],\n",
      "        [0.9796, 0.0207]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9799, 0.0191],\n",
      "        [0.9767, 0.0223],\n",
      "        [0.9804, 0.0189],\n",
      "        [0.9731, 0.0316]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9790, 0.0235],\n",
      "        [0.9780, 0.0200],\n",
      "        [0.9743, 0.0258],\n",
      "        [0.9793, 0.0193]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9827, 0.0130],\n",
      "        [0.9856, 0.0182],\n",
      "        [0.9868, 0.0158],\n",
      "        [0.9862, 0.0138]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9877, 0.0112],\n",
      "        [0.9846, 0.0143],\n",
      "        [0.9859, 0.0282],\n",
      "        [0.9872, 0.0194]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9921, 0.0083],\n",
      "        [0.9903, 0.0103],\n",
      "        [0.9780, 0.0138],\n",
      "        [0.9885, 0.0100]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9886, 0.0089],\n",
      "        [0.9903, 0.0096],\n",
      "        [0.9899, 0.0087],\n",
      "        [0.9903, 0.0100]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9914, 0.0074],\n",
      "        [0.9921, 0.0093],\n",
      "        [0.9897, 0.0179],\n",
      "        [0.9915, 0.0081]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9660, 0.0312],\n",
      "        [0.9709, 0.0265],\n",
      "        [0.9731, 0.0258],\n",
      "        [0.9758, 0.0221]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9618, 0.0369],\n",
      "        [0.9564, 0.0632],\n",
      "        [0.9613, 0.0371],\n",
      "        [0.9503, 0.0409]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9677, 0.0333],\n",
      "        [0.9396, 0.0538],\n",
      "        [0.9575, 0.0377],\n",
      "        [0.9530, 0.0390]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9647, 0.0320],\n",
      "        [0.9684, 0.0313],\n",
      "        [0.9676, 0.0263],\n",
      "        [0.9574, 0.0358]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9710, 0.0293],\n",
      "        [0.9706, 0.0348],\n",
      "        [0.9549, 0.0477],\n",
      "        [0.9738, 0.0261]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9792, 0.0236],\n",
      "        [0.9764, 0.0300],\n",
      "        [0.9798, 0.0202],\n",
      "        [0.9831, 0.0196]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9779, 0.0202],\n",
      "        [0.9803, 0.0158],\n",
      "        [0.9795, 0.0359],\n",
      "        [0.9632, 0.0260]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9800, 0.0178],\n",
      "        [0.9831, 0.0171],\n",
      "        [0.9836, 0.0130],\n",
      "        [0.9839, 0.0176]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9875, 0.0114],\n",
      "        [0.9822, 0.0176],\n",
      "        [0.9837, 0.0143],\n",
      "        [0.9778, 0.0185]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9849, 0.0127],\n",
      "        [0.9860, 0.0099],\n",
      "        [0.9866, 0.0115],\n",
      "        [0.9891, 0.0105]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9925, 0.0089],\n",
      "        [0.9898, 0.0092],\n",
      "        [0.9891, 0.0106],\n",
      "        [0.9896, 0.0107]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9936, 0.0061],\n",
      "        [0.9859, 0.0138],\n",
      "        [0.9927, 0.0085],\n",
      "        [0.9868, 0.0066]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9942, 0.0069],\n",
      "        [0.9901, 0.0179],\n",
      "        [0.9910, 0.0077],\n",
      "        [0.9922, 0.0075]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9927, 0.0072],\n",
      "        [0.9867, 0.0112],\n",
      "        [0.9948, 0.0050],\n",
      "        [0.9931, 0.0058]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9934, 0.0065],\n",
      "        [0.9947, 0.0055],\n",
      "        [0.9940, 0.0117],\n",
      "        [0.9953, 0.0044]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0033],\n",
      "        [0.9932, 0.0062],\n",
      "        [0.9915, 0.0066],\n",
      "        [0.9913, 0.0063]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9895, 0.0113],\n",
      "        [0.9951, 0.0041],\n",
      "        [0.9948, 0.0035],\n",
      "        [0.9939, 0.0139]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9942, 0.0054],\n",
      "        [0.9945, 0.0041],\n",
      "        [0.9925, 0.0054],\n",
      "        [0.9936, 0.0044]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9953, 0.0086],\n",
      "        [0.9950, 0.0045],\n",
      "        [0.9960, 0.0034],\n",
      "        [0.9959, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9966, 0.0036],\n",
      "        [0.9963, 0.0031],\n",
      "        [0.9952, 0.0073],\n",
      "        [0.9912, 0.0053]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0036],\n",
      "        [0.9957, 0.0036],\n",
      "        [0.9966, 0.0025],\n",
      "        [0.9957, 0.0043]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9954, 0.0081],\n",
      "        [0.9924, 0.0054],\n",
      "        [0.9962, 0.0028],\n",
      "        [0.9958, 0.0045]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9960, 0.0036],\n",
      "        [0.9973, 0.0024],\n",
      "        [0.9957, 0.0033],\n",
      "        [0.9950, 0.0038]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0021],\n",
      "        [0.9962, 0.0064],\n",
      "        [0.9953, 0.0029],\n",
      "        [0.9962, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9954, 0.0029],\n",
      "        [0.9974, 0.0026],\n",
      "        [0.9975, 0.0027],\n",
      "        [0.9953, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9969, 0.0028],\n",
      "        [0.9974, 0.0022],\n",
      "        [0.9975, 0.0019],\n",
      "        [0.9969, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0018],\n",
      "        [0.9967, 0.0023],\n",
      "        [0.9974, 0.0027],\n",
      "        [0.9982, 0.0018]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0027],\n",
      "        [0.9982, 0.0021],\n",
      "        [0.9967, 0.0029],\n",
      "        [0.9970, 0.0023]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9981, 0.0020],\n",
      "        [0.9963, 0.0051],\n",
      "        [0.9976, 0.0027],\n",
      "        [0.9981, 0.0015]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0018],\n",
      "        [0.9986, 0.0018],\n",
      "        [0.9975, 0.0024],\n",
      "        [0.9981, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0020],\n",
      "        [0.9982, 0.0022],\n",
      "        [0.9971, 0.0034],\n",
      "        [0.9971, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0072],\n",
      "        [0.9974, 0.0019],\n",
      "        [0.9964, 0.0045],\n",
      "        [0.9973, 0.0049]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0020],\n",
      "        [0.9976, 0.0015],\n",
      "        [0.9980, 0.0018],\n",
      "        [0.9978, 0.0032]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0021],\n",
      "        [0.9958, 0.0031],\n",
      "        [0.9980, 0.0026],\n",
      "        [0.9972, 0.0037]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0015],\n",
      "        [0.9980, 0.0017],\n",
      "        [0.9977, 0.0017],\n",
      "        [0.9969, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9975, 0.0018],\n",
      "        [0.9981, 0.0014],\n",
      "        [0.9982, 0.0013],\n",
      "        [0.9975, 0.0022]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9760e-01, 1.3528e-03],\n",
      "        [9.9643e-01, 1.5286e-03],\n",
      "        [9.9903e-01, 9.8046e-04],\n",
      "        [9.9756e-01, 3.2835e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9988, 0.0013],\n",
      "        [0.9975, 0.0022],\n",
      "        [0.9979, 0.0016],\n",
      "        [0.9975, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0010],\n",
      "        [0.9987, 0.0016],\n",
      "        [0.9985, 0.0025],\n",
      "        [0.9984, 0.0022]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0014],\n",
      "        [0.9983, 0.0012],\n",
      "        [0.9981, 0.0017],\n",
      "        [0.9983, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0012],\n",
      "        [0.9986, 0.0013],\n",
      "        [0.9984, 0.0025],\n",
      "        [0.9979, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9782e-01, 1.4995e-03],\n",
      "        [9.9856e-01, 1.1919e-03],\n",
      "        [9.9841e-01, 1.2793e-03],\n",
      "        [9.9864e-01, 9.4230e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0016],\n",
      "        [0.9983, 0.0012],\n",
      "        [0.9974, 0.0031],\n",
      "        [0.9985, 0.0011]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0017],\n",
      "        [0.9987, 0.0012],\n",
      "        [0.9987, 0.0013],\n",
      "        [0.9988, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9985, 0.0014],\n",
      "        [0.9988, 0.0011],\n",
      "        [0.9977, 0.0015],\n",
      "        [0.9973, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0032],\n",
      "        [0.9983, 0.0011],\n",
      "        [0.9985, 0.0026],\n",
      "        [0.9978, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0026],\n",
      "        [0.9971, 0.0026],\n",
      "        [0.9972, 0.0014],\n",
      "        [0.9986, 0.0010]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9860e-01, 1.1959e-03],\n",
      "        [9.9881e-01, 1.1757e-03],\n",
      "        [9.9880e-01, 2.1170e-03],\n",
      "        [9.9893e-01, 6.9958e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9903e-01, 7.6640e-04],\n",
      "        [9.9863e-01, 1.8397e-03],\n",
      "        [9.9886e-01, 1.0068e-03],\n",
      "        [9.9851e-01, 1.0835e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9875e-01, 1.3358e-03],\n",
      "        [9.9894e-01, 8.9113e-04],\n",
      "        [9.9823e-01, 1.2603e-03],\n",
      "        [9.9805e-01, 2.8005e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9747e-01, 8.1852e-04],\n",
      "        [9.9912e-01, 5.5675e-04],\n",
      "        [9.9907e-01, 1.1224e-03],\n",
      "        [9.9809e-01, 3.2616e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9918e-01, 7.5583e-04],\n",
      "        [9.9855e-01, 1.3580e-03],\n",
      "        [9.9891e-01, 7.0051e-04],\n",
      "        [9.9792e-01, 1.5156e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9912e-01, 9.8230e-04],\n",
      "        [9.9879e-01, 8.2757e-04],\n",
      "        [9.9833e-01, 1.2286e-03],\n",
      "        [9.9839e-01, 1.1782e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9906e-01, 6.9047e-04],\n",
      "        [9.9856e-01, 8.3886e-04],\n",
      "        [9.9898e-01, 9.8224e-04],\n",
      "        [9.9857e-01, 8.2587e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9917e-01, 7.9880e-04],\n",
      "        [9.9908e-01, 1.0054e-03],\n",
      "        [9.9901e-01, 1.2045e-03],\n",
      "        [9.9907e-01, 8.0479e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9875e-01, 8.7545e-04],\n",
      "        [9.9883e-01, 7.8356e-04],\n",
      "        [9.9926e-01, 6.1455e-04],\n",
      "        [9.9866e-01, 1.2880e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9914e-01, 8.8954e-04],\n",
      "        [9.9881e-01, 1.1254e-03],\n",
      "        [9.9872e-01, 9.7168e-04],\n",
      "        [9.9915e-01, 8.9992e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9899e-01, 9.4771e-04],\n",
      "        [9.9815e-01, 9.5032e-04],\n",
      "        [9.9843e-01, 1.3565e-03],\n",
      "        [9.9897e-01, 1.0351e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9898e-01, 9.5097e-04],\n",
      "        [9.9850e-01, 1.6878e-03],\n",
      "        [9.9869e-01, 8.4691e-04],\n",
      "        [9.9904e-01, 7.2009e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9921e-01, 5.0604e-04],\n",
      "        [9.9928e-01, 8.5518e-04],\n",
      "        [9.9863e-01, 8.3269e-04],\n",
      "        [9.9876e-01, 8.2408e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9909e-01, 8.4318e-04],\n",
      "        [9.9920e-01, 6.3439e-04],\n",
      "        [9.9894e-01, 7.7562e-04],\n",
      "        [9.9929e-01, 6.8494e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9932e-01, 5.6539e-04],\n",
      "        [9.9889e-01, 1.0900e-03],\n",
      "        [9.9838e-01, 1.0292e-03],\n",
      "        [9.9896e-01, 1.0583e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9887e-01, 8.3595e-04],\n",
      "        [9.9944e-01, 5.4782e-04],\n",
      "        [9.9904e-01, 1.0032e-03],\n",
      "        [9.9905e-01, 7.1064e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9940e-01, 6.1608e-04],\n",
      "        [9.9896e-01, 1.0104e-03],\n",
      "        [9.9948e-01, 7.5769e-04],\n",
      "        [9.9839e-01, 1.1033e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9917e-01, 1.2536e-03],\n",
      "        [9.9949e-01, 7.2859e-04],\n",
      "        [9.9933e-01, 5.9717e-04],\n",
      "        [9.9936e-01, 6.0430e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9918e-01, 8.9665e-04],\n",
      "        [9.9837e-01, 7.9731e-04],\n",
      "        [9.9908e-01, 8.1491e-04],\n",
      "        [9.9927e-01, 8.2207e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9934e-01, 6.6578e-04],\n",
      "        [9.9906e-01, 6.8052e-04],\n",
      "        [9.9911e-01, 5.9040e-04],\n",
      "        [9.9892e-01, 2.5069e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9922e-01, 4.8799e-04],\n",
      "        [9.9893e-01, 1.6883e-03],\n",
      "        [9.9880e-01, 6.2546e-04],\n",
      "        [9.9923e-01, 7.4540e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9951e-01, 4.1941e-04],\n",
      "        [9.9894e-01, 8.9538e-04],\n",
      "        [9.9939e-01, 6.4913e-04],\n",
      "        [9.9915e-01, 7.9150e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9875e-01, 1.1427e-03],\n",
      "        [9.9950e-01, 5.0259e-04],\n",
      "        [9.9919e-01, 7.4417e-04],\n",
      "        [9.9920e-01, 5.9547e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9840e-01, 1.7557e-03],\n",
      "        [9.9938e-01, 6.4486e-04],\n",
      "        [9.9929e-01, 6.9777e-04],\n",
      "        [9.9920e-01, 8.7049e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9949e-01, 1.1173e-03],\n",
      "        [9.9928e-01, 7.0283e-04],\n",
      "        [9.9922e-01, 7.4083e-04],\n",
      "        [9.9941e-01, 5.4346e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9933e-01, 6.6934e-04],\n",
      "        [9.9919e-01, 1.6907e-03],\n",
      "        [9.9944e-01, 4.3911e-04],\n",
      "        [9.9941e-01, 5.2781e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9858e-01, 1.4524e-03],\n",
      "        [9.9942e-01, 5.2135e-04],\n",
      "        [9.9893e-01, 6.6735e-04],\n",
      "        [9.9924e-01, 1.1217e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9932e-01, 7.8564e-04],\n",
      "        [9.9948e-01, 2.5846e-04],\n",
      "        [9.9951e-01, 5.4500e-04],\n",
      "        [9.9943e-01, 4.5135e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9900e-01, 6.2399e-04],\n",
      "        [9.9840e-01, 1.3389e-03],\n",
      "        [9.9903e-01, 1.4733e-03],\n",
      "        [9.9942e-01, 4.0908e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9895e-01, 4.6141e-04],\n",
      "        [9.9929e-01, 5.8366e-04],\n",
      "        [9.9911e-01, 6.5435e-04],\n",
      "        [9.9883e-01, 5.0628e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9923e-01, 7.5850e-04],\n",
      "        [9.9850e-01, 1.7050e-03],\n",
      "        [9.9941e-01, 4.8159e-04],\n",
      "        [9.9896e-01, 7.9224e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9923e-01, 6.2882e-04],\n",
      "        [9.9889e-01, 8.2967e-04],\n",
      "        [9.9931e-01, 5.2736e-04],\n",
      "        [9.9937e-01, 4.9045e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9902e-01, 4.9313e-04],\n",
      "        [9.9937e-01, 4.5076e-04],\n",
      "        [9.9913e-01, 4.6665e-04],\n",
      "        [9.9935e-01, 1.0226e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9935e-01, 6.4567e-04],\n",
      "        [9.9922e-01, 6.4483e-04],\n",
      "        [9.9937e-01, 5.4866e-04],\n",
      "        [9.9926e-01, 4.0738e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9904e-01, 7.1229e-04],\n",
      "        [9.9949e-01, 3.7067e-04],\n",
      "        [9.9950e-01, 4.6248e-04],\n",
      "        [9.9905e-01, 1.1065e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9955e-01, 4.8426e-04],\n",
      "        [9.9928e-01, 5.8796e-04],\n",
      "        [9.9888e-01, 5.4614e-04],\n",
      "        [9.9875e-01, 7.0851e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9930e-01, 8.4250e-04],\n",
      "        [9.9931e-01, 4.5582e-04],\n",
      "        [9.9930e-01, 4.2290e-04],\n",
      "        [9.9939e-01, 1.5060e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9953e-01, 4.6776e-04],\n",
      "        [9.9900e-01, 7.9816e-04],\n",
      "        [9.9938e-01, 6.0203e-04],\n",
      "        [9.9928e-01, 5.8311e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9883e-01, 1.1485e-03],\n",
      "        [9.9955e-01, 4.5695e-04],\n",
      "        [9.9943e-01, 4.1404e-04],\n",
      "        [9.9937e-01, 6.3588e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9942e-01, 1.4393e-03],\n",
      "        [9.9953e-01, 5.3062e-04],\n",
      "        [9.9947e-01, 4.0625e-04],\n",
      "        [9.9946e-01, 6.0246e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9944, 0.0044],\n",
      "        [0.9941, 0.0057],\n",
      "        [0.9924, 0.0076],\n",
      "        [0.9944, 0.0056]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9862, 0.0145],\n",
      "        [0.9854, 0.0194],\n",
      "        [0.9853, 0.0112],\n",
      "        [0.9774, 0.0157]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9735, 0.0221],\n",
      "        [0.9778, 0.0145],\n",
      "        [0.9802, 0.0162],\n",
      "        [0.9844, 0.0168]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9816, 0.0173],\n",
      "        [0.9783, 0.0212],\n",
      "        [0.9842, 0.0159],\n",
      "        [0.9817, 0.0188]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9869, 0.0160],\n",
      "        [0.9854, 0.0180],\n",
      "        [0.9807, 0.0174],\n",
      "        [0.9807, 0.0175]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9852, 0.0149],\n",
      "        [0.9841, 0.0134],\n",
      "        [0.9882, 0.0120],\n",
      "        [0.9818, 0.0185]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9743, 0.0193],\n",
      "        [0.9887, 0.0206],\n",
      "        [0.9871, 0.0117],\n",
      "        [0.9888, 0.0122]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9892, 0.0127],\n",
      "        [0.9894, 0.0093],\n",
      "        [0.9807, 0.0149],\n",
      "        [0.9892, 0.0096]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9860, 0.0180],\n",
      "        [0.9911, 0.0095],\n",
      "        [0.9866, 0.0114],\n",
      "        [0.9869, 0.0119]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9925, 0.0117],\n",
      "        [0.9921, 0.0125],\n",
      "        [0.9896, 0.0091],\n",
      "        [0.9908, 0.0079]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9943, 0.0059],\n",
      "        [0.9932, 0.0097],\n",
      "        [0.9934, 0.0053],\n",
      "        [0.9868, 0.0199]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9925, 0.0082],\n",
      "        [0.9907, 0.0153],\n",
      "        [0.9919, 0.0073],\n",
      "        [0.9930, 0.0075]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9933, 0.0053],\n",
      "        [0.9920, 0.0110],\n",
      "        [0.9940, 0.0093],\n",
      "        [0.9936, 0.0060]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9859, 0.0149],\n",
      "        [0.9897, 0.0067],\n",
      "        [0.9917, 0.0120],\n",
      "        [0.9921, 0.0063]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9891, 0.0064],\n",
      "        [0.9941, 0.0048],\n",
      "        [0.9911, 0.0066],\n",
      "        [0.9957, 0.0043]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9953, 0.0037],\n",
      "        [0.9893, 0.0070],\n",
      "        [0.9946, 0.0050],\n",
      "        [0.9946, 0.0050]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9956, 0.0041],\n",
      "        [0.9949, 0.0039],\n",
      "        [0.9955, 0.0031],\n",
      "        [0.9953, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9928, 0.0048],\n",
      "        [0.9957, 0.0039],\n",
      "        [0.9954, 0.0047],\n",
      "        [0.9956, 0.0036]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9933, 0.0037],\n",
      "        [0.9945, 0.0083],\n",
      "        [0.9965, 0.0036],\n",
      "        [0.9939, 0.0052]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0063],\n",
      "        [0.9941, 0.0035],\n",
      "        [0.9950, 0.0035],\n",
      "        [0.9959, 0.0032]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 1], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9841, 0.0178],\n",
      "        [0.9894, 0.0128],\n",
      "        [0.9832, 0.0178],\n",
      "        [0.9840, 0.0143]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9808, 0.0240],\n",
      "        [0.9735, 0.0386],\n",
      "        [0.9769, 0.0255],\n",
      "        [0.9800, 0.0253]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9750, 0.0301],\n",
      "        [0.9577, 0.0274],\n",
      "        [0.9747, 0.0402],\n",
      "        [0.9770, 0.0198]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9760, 0.0274],\n",
      "        [0.9717, 0.0359],\n",
      "        [0.9640, 0.0392],\n",
      "        [0.9725, 0.0269]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9779, 0.0333],\n",
      "        [0.9749, 0.0225],\n",
      "        [0.9746, 0.0231],\n",
      "        [0.9742, 0.0195]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9777, 0.0174],\n",
      "        [0.9801, 0.0201],\n",
      "        [0.9803, 0.0186],\n",
      "        [0.9822, 0.0183]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9818, 0.0144],\n",
      "        [0.9871, 0.0162],\n",
      "        [0.9802, 0.0158],\n",
      "        [0.9827, 0.0125]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9875, 0.0102],\n",
      "        [0.9889, 0.0112],\n",
      "        [0.9889, 0.0123],\n",
      "        [0.9805, 0.0166]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9873, 0.0094],\n",
      "        [0.9876, 0.0100],\n",
      "        [0.9894, 0.0107],\n",
      "        [0.9855, 0.0141]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9907, 0.0092],\n",
      "        [0.9896, 0.0081],\n",
      "        [0.9901, 0.0094],\n",
      "        [0.9883, 0.0098]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9922, 0.0138],\n",
      "        [0.9901, 0.0099],\n",
      "        [0.9866, 0.0159],\n",
      "        [0.9861, 0.0095]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9901, 0.0101],\n",
      "        [0.9913, 0.0076],\n",
      "        [0.9899, 0.0085],\n",
      "        [0.9912, 0.0082]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9934, 0.0075],\n",
      "        [0.9926, 0.0066],\n",
      "        [0.9922, 0.0081],\n",
      "        [0.9898, 0.0085]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9929, 0.0099],\n",
      "        [0.9885, 0.0081],\n",
      "        [0.9905, 0.0081],\n",
      "        [0.9854, 0.0084]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0047],\n",
      "        [0.9944, 0.0056],\n",
      "        [0.9934, 0.0055],\n",
      "        [0.9931, 0.0068]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9902, 0.0118],\n",
      "        [0.9917, 0.0081],\n",
      "        [0.9945, 0.0056],\n",
      "        [0.9951, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0044],\n",
      "        [0.9949, 0.0054],\n",
      "        [0.9932, 0.0057],\n",
      "        [0.9951, 0.0049]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9950, 0.0113],\n",
      "        [0.9929, 0.0049],\n",
      "        [0.9954, 0.0048],\n",
      "        [0.9929, 0.0050]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0062],\n",
      "        [0.9929, 0.0080],\n",
      "        [0.9937, 0.0051],\n",
      "        [0.9938, 0.0049]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9940, 0.0058],\n",
      "        [0.9966, 0.0046],\n",
      "        [0.9938, 0.0039],\n",
      "        [0.9923, 0.0063]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9960, 0.0045],\n",
      "        [0.9958, 0.0041],\n",
      "        [0.9905, 0.0096],\n",
      "        [0.9956, 0.0045]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0036],\n",
      "        [0.9969, 0.0034],\n",
      "        [0.9939, 0.0039],\n",
      "        [0.9951, 0.0037]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9953, 0.0038],\n",
      "        [0.9967, 0.0025],\n",
      "        [0.9949, 0.0081],\n",
      "        [0.9963, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9977, 0.0024],\n",
      "        [0.9972, 0.0043],\n",
      "        [0.9974, 0.0030],\n",
      "        [0.9976, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9956, 0.0036],\n",
      "        [0.9962, 0.0037],\n",
      "        [0.9970, 0.0024],\n",
      "        [0.9975, 0.0025]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9945, 0.0045],\n",
      "        [0.9963, 0.0037],\n",
      "        [0.9959, 0.0043],\n",
      "        [0.9974, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0018],\n",
      "        [0.9976, 0.0028],\n",
      "        [0.9975, 0.0033],\n",
      "        [0.9976, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0023],\n",
      "        [0.9977, 0.0021],\n",
      "        [0.9947, 0.0035],\n",
      "        [0.9974, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0028],\n",
      "        [0.9970, 0.0034],\n",
      "        [0.9973, 0.0022],\n",
      "        [0.9970, 0.0026]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0030],\n",
      "        [0.9978, 0.0022],\n",
      "        [0.9968, 0.0028],\n",
      "        [0.9960, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0033],\n",
      "        [0.9958, 0.0038],\n",
      "        [0.9962, 0.0070],\n",
      "        [0.9975, 0.0036]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0024],\n",
      "        [0.9971, 0.0030],\n",
      "        [0.9982, 0.0016],\n",
      "        [0.9978, 0.0020]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0052],\n",
      "        [0.9977, 0.0028],\n",
      "        [0.9967, 0.0037],\n",
      "        [0.9983, 0.0017]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0023],\n",
      "        [0.9971, 0.0023],\n",
      "        [0.9972, 0.0046],\n",
      "        [0.9968, 0.0062]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0022],\n",
      "        [0.9939, 0.0047],\n",
      "        [0.9984, 0.0017],\n",
      "        [0.9980, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9979, 0.0016],\n",
      "        [0.9969, 0.0025],\n",
      "        [0.9982, 0.0033],\n",
      "        [0.9973, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9974, 0.0024],\n",
      "        [0.9970, 0.0025],\n",
      "        [0.9982, 0.0018],\n",
      "        [0.9975, 0.0023]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0013],\n",
      "        [0.9969, 0.0021],\n",
      "        [0.9966, 0.0033],\n",
      "        [0.9971, 0.0020]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0030],\n",
      "        [0.9983, 0.0023],\n",
      "        [0.9976, 0.0039],\n",
      "        [0.9986, 0.0015]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9990, 0.0022],\n",
      "        [0.9986, 0.0010],\n",
      "        [0.9987, 0.0011],\n",
      "        [0.9970, 0.0019]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0029],\n",
      "        [0.9988, 0.0014],\n",
      "        [0.9983, 0.0014],\n",
      "        [0.9989, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0027],\n",
      "        [0.9974, 0.0019],\n",
      "        [0.9984, 0.0019],\n",
      "        [0.9978, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9978, 0.0013],\n",
      "        [0.9985, 0.0012],\n",
      "        [0.9964, 0.0023],\n",
      "        [0.9975, 0.0049]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0017],\n",
      "        [0.9985, 0.0026],\n",
      "        [0.9984, 0.0014],\n",
      "        [0.9979, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9984, 0.0019],\n",
      "        [0.9988, 0.0016],\n",
      "        [0.9979, 0.0016],\n",
      "        [0.9988, 0.0010]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9819e-01, 1.7878e-03],\n",
      "        [9.9865e-01, 1.4645e-03],\n",
      "        [9.9898e-01, 1.1348e-03],\n",
      "        [9.9869e-01, 9.2045e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9889e-01, 1.0908e-03],\n",
      "        [9.9684e-01, 2.3287e-03],\n",
      "        [9.9851e-01, 1.0986e-03],\n",
      "        [9.9906e-01, 8.4134e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9756e-01, 2.3827e-03],\n",
      "        [9.9876e-01, 7.6788e-04],\n",
      "        [9.9854e-01, 1.1561e-03],\n",
      "        [9.9815e-01, 1.3620e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9891e-01, 1.0288e-03],\n",
      "        [9.9884e-01, 9.4034e-04],\n",
      "        [9.9801e-01, 1.3583e-03],\n",
      "        [9.9780e-01, 1.7529e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9849e-01, 1.5741e-03],\n",
      "        [9.9887e-01, 8.5852e-04],\n",
      "        [9.9811e-01, 1.9659e-03],\n",
      "        [9.9889e-01, 9.0628e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9990, 0.0010],\n",
      "        [0.9983, 0.0013],\n",
      "        [0.9981, 0.0025],\n",
      "        [0.9992, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9827e-01, 2.1309e-03],\n",
      "        [9.9896e-01, 8.3975e-04],\n",
      "        [9.9868e-01, 9.2729e-04],\n",
      "        [9.9713e-01, 1.5000e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9832e-01, 1.3434e-03],\n",
      "        [9.9907e-01, 9.2554e-04],\n",
      "        [9.9909e-01, 7.3755e-04],\n",
      "        [9.9880e-01, 1.0878e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9877e-01, 9.6449e-04],\n",
      "        [9.9865e-01, 1.1020e-03],\n",
      "        [9.9835e-01, 1.4933e-03],\n",
      "        [9.9829e-01, 1.1091e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9986, 0.0013],\n",
      "        [0.9981, 0.0025],\n",
      "        [0.9986, 0.0013],\n",
      "        [0.9990, 0.0013]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9905e-01, 8.3508e-04],\n",
      "        [9.9802e-01, 1.4949e-03],\n",
      "        [9.9881e-01, 9.9930e-04],\n",
      "        [9.9885e-01, 1.1296e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9859e-01, 2.6126e-03],\n",
      "        [9.9872e-01, 1.6053e-03],\n",
      "        [9.9907e-01, 8.3021e-04],\n",
      "        [9.9783e-01, 1.4392e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9818e-01, 1.2887e-03],\n",
      "        [9.9830e-01, 1.3720e-03],\n",
      "        [9.9829e-01, 1.2939e-03],\n",
      "        [9.9929e-01, 8.6014e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9870e-01, 8.2877e-04],\n",
      "        [9.9916e-01, 8.3736e-04],\n",
      "        [9.9891e-01, 1.9990e-03],\n",
      "        [9.9850e-01, 9.2295e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9869e-01, 1.2603e-03],\n",
      "        [9.9904e-01, 1.1929e-03],\n",
      "        [9.9866e-01, 1.2272e-03],\n",
      "        [9.9875e-01, 9.3266e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9929e-01, 6.9986e-04],\n",
      "        [9.9887e-01, 1.2230e-03],\n",
      "        [9.9919e-01, 7.2286e-04],\n",
      "        [9.9824e-01, 1.9553e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9892e-01, 1.1284e-03],\n",
      "        [9.9915e-01, 7.0612e-04],\n",
      "        [9.9936e-01, 1.3748e-03],\n",
      "        [9.9914e-01, 1.1068e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9783e-01, 1.5068e-03],\n",
      "        [9.9934e-01, 1.2001e-03],\n",
      "        [9.9908e-01, 6.4388e-04],\n",
      "        [9.9918e-01, 7.2519e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9929e-01, 7.1858e-04],\n",
      "        [9.9891e-01, 7.7636e-04],\n",
      "        [9.9873e-01, 1.1321e-03],\n",
      "        [9.9898e-01, 1.2066e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9894e-01, 1.0642e-03],\n",
      "        [9.9893e-01, 8.7313e-04],\n",
      "        [9.9902e-01, 8.9651e-04],\n",
      "        [9.9920e-01, 8.7392e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9896e-01, 8.3967e-04],\n",
      "        [9.9923e-01, 8.8181e-04],\n",
      "        [9.9782e-01, 2.5829e-03],\n",
      "        [9.9916e-01, 7.5953e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9907e-01, 1.4534e-03],\n",
      "        [9.9897e-01, 8.3425e-04],\n",
      "        [9.9891e-01, 1.0210e-03],\n",
      "        [9.9938e-01, 1.0156e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9883e-01, 7.9780e-04],\n",
      "        [9.9925e-01, 6.4808e-04],\n",
      "        [9.9890e-01, 1.0800e-03],\n",
      "        [9.9910e-01, 5.8280e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9904e-01, 1.1930e-03],\n",
      "        [9.9914e-01, 6.3075e-04],\n",
      "        [9.9866e-01, 1.9213e-03],\n",
      "        [9.9812e-01, 2.0706e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9845e-01, 9.8951e-04],\n",
      "        [9.9895e-01, 9.5162e-04],\n",
      "        [9.9896e-01, 1.2726e-03],\n",
      "        [9.9937e-01, 6.6929e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9943e-01, 4.4075e-04],\n",
      "        [9.9944e-01, 4.4762e-04],\n",
      "        [9.9852e-01, 1.1969e-03],\n",
      "        [9.9888e-01, 7.3290e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9918e-01, 5.1979e-04],\n",
      "        [9.9912e-01, 8.2431e-04],\n",
      "        [9.9934e-01, 6.8329e-04],\n",
      "        [9.9951e-01, 4.4799e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9924e-01, 8.2549e-04],\n",
      "        [9.9919e-01, 1.2791e-03],\n",
      "        [9.9866e-01, 7.2510e-04],\n",
      "        [9.9935e-01, 6.6508e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9884e-01, 1.3533e-03],\n",
      "        [9.9940e-01, 5.5546e-04],\n",
      "        [9.9944e-01, 4.7257e-04],\n",
      "        [9.9889e-01, 9.2064e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9946e-01, 6.6408e-04],\n",
      "        [9.9916e-01, 7.4573e-04],\n",
      "        [9.9931e-01, 5.5626e-04],\n",
      "        [9.9891e-01, 9.1353e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9928e-01, 5.5275e-04],\n",
      "        [9.9927e-01, 9.0694e-04],\n",
      "        [9.9937e-01, 6.7916e-04],\n",
      "        [9.9924e-01, 5.2906e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9900e-01, 1.6755e-03],\n",
      "        [9.9914e-01, 7.9493e-04],\n",
      "        [9.9937e-01, 6.7563e-04],\n",
      "        [9.9869e-01, 9.1648e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9870e-01, 9.6456e-04],\n",
      "        [9.9941e-01, 4.3218e-04],\n",
      "        [9.9902e-01, 1.0105e-03],\n",
      "        [9.9921e-01, 6.3487e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9920e-01, 7.4620e-04],\n",
      "        [9.9919e-01, 4.9532e-04],\n",
      "        [9.9952e-01, 5.4978e-04],\n",
      "        [9.9950e-01, 5.7542e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9900e-01, 1.1161e-03],\n",
      "        [9.9928e-01, 9.7976e-04],\n",
      "        [9.9938e-01, 5.0281e-04],\n",
      "        [9.9953e-01, 4.0291e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9879e-01, 7.1772e-04],\n",
      "        [9.9952e-01, 5.3959e-04],\n",
      "        [9.9922e-01, 9.1743e-04],\n",
      "        [9.9902e-01, 8.9328e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9949e-01, 3.5751e-04],\n",
      "        [9.9949e-01, 7.2945e-04],\n",
      "        [9.9913e-01, 1.5710e-03],\n",
      "        [9.9913e-01, 5.4949e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9946e-01, 6.4653e-04],\n",
      "        [9.9924e-01, 6.8271e-04],\n",
      "        [9.9870e-01, 1.9246e-03],\n",
      "        [9.9925e-01, 6.6005e-04]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 1, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9954, 0.0036],\n",
      "        [0.9954, 0.0042],\n",
      "        [0.9933, 0.0046],\n",
      "        [0.9918, 0.0057]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9915, 0.0092],\n",
      "        [0.9930, 0.0068],\n",
      "        [0.9902, 0.0088],\n",
      "        [0.9922, 0.0112]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9877, 0.0114],\n",
      "        [0.9886, 0.0119],\n",
      "        [0.9830, 0.0155],\n",
      "        [0.9892, 0.0113]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9871, 0.0127],\n",
      "        [0.9861, 0.0239],\n",
      "        [0.9871, 0.0104],\n",
      "        [0.9857, 0.0185]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9837, 0.0140],\n",
      "        [0.9883, 0.0119],\n",
      "        [0.9855, 0.0123],\n",
      "        [0.9891, 0.0099]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9866, 0.0110],\n",
      "        [0.9889, 0.0121],\n",
      "        [0.9876, 0.0098],\n",
      "        [0.9847, 0.0119]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9897, 0.0075],\n",
      "        [0.9925, 0.0090],\n",
      "        [0.9888, 0.0076],\n",
      "        [0.9877, 0.0134]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9855, 0.0138],\n",
      "        [0.9865, 0.0131],\n",
      "        [0.9907, 0.0112],\n",
      "        [0.9913, 0.0111]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9918, 0.0071],\n",
      "        [0.9930, 0.0073],\n",
      "        [0.9927, 0.0083],\n",
      "        [0.9879, 0.0128]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9921, 0.0077],\n",
      "        [0.9929, 0.0071],\n",
      "        [0.9933, 0.0067],\n",
      "        [0.9925, 0.0088]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9896, 0.0094],\n",
      "        [0.9941, 0.0055],\n",
      "        [0.9938, 0.0063],\n",
      "        [0.9890, 0.0097]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9937, 0.0083],\n",
      "        [0.9939, 0.0089],\n",
      "        [0.9911, 0.0070],\n",
      "        [0.9902, 0.0090]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9922, 0.0066],\n",
      "        [0.9952, 0.0039],\n",
      "        [0.9935, 0.0048],\n",
      "        [0.9956, 0.0054]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9956, 0.0046],\n",
      "        [0.9934, 0.0055],\n",
      "        [0.9951, 0.0037],\n",
      "        [0.9961, 0.0038]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9959, 0.0041],\n",
      "        [0.9964, 0.0032],\n",
      "        [0.9951, 0.0053],\n",
      "        [0.9954, 0.0048]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9953, 0.0057],\n",
      "        [0.9936, 0.0061],\n",
      "        [0.9926, 0.0131],\n",
      "        [0.9961, 0.0061]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9952, 0.0049],\n",
      "        [0.9958, 0.0045],\n",
      "        [0.9951, 0.0062],\n",
      "        [0.9967, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9960, 0.0037],\n",
      "        [0.9949, 0.0056],\n",
      "        [0.9943, 0.0078],\n",
      "        [0.9949, 0.0061]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0036],\n",
      "        [0.9966, 0.0035],\n",
      "        [0.9976, 0.0026],\n",
      "        [0.9975, 0.0029]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0032],\n",
      "        [0.9951, 0.0040],\n",
      "        [0.9963, 0.0035],\n",
      "        [0.9963, 0.0028]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0027],\n",
      "        [0.9979, 0.0016],\n",
      "        [0.9977, 0.0024],\n",
      "        [0.9962, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9955, 0.0035],\n",
      "        [0.9955, 0.0027],\n",
      "        [0.9974, 0.0021],\n",
      "        [0.9942, 0.0059]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([1, 0, 0, 0], device='cuda:0')\n",
      "tensor([[0., 1.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9843, 0.0195],\n",
      "        [0.9761, 0.0202],\n",
      "        [0.9881, 0.0087],\n",
      "        [0.9842, 0.0152]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9784, 0.0202],\n",
      "        [0.9756, 0.0313],\n",
      "        [0.9843, 0.0210],\n",
      "        [0.9800, 0.0287]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9723, 0.0281],\n",
      "        [0.9711, 0.0264],\n",
      "        [0.9760, 0.0236],\n",
      "        [0.9755, 0.0226]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9825, 0.0200],\n",
      "        [0.9755, 0.0214],\n",
      "        [0.9816, 0.0211],\n",
      "        [0.9764, 0.0195]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9755, 0.0252],\n",
      "        [0.9815, 0.0199],\n",
      "        [0.9776, 0.0191],\n",
      "        [0.9721, 0.0369]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9813, 0.0157],\n",
      "        [0.9801, 0.0184],\n",
      "        [0.9770, 0.0209],\n",
      "        [0.9812, 0.0202]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9802, 0.0247],\n",
      "        [0.9828, 0.0164],\n",
      "        [0.9864, 0.0152],\n",
      "        [0.9820, 0.0157]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9883, 0.0184],\n",
      "        [0.9862, 0.0123],\n",
      "        [0.9885, 0.0158],\n",
      "        [0.9894, 0.0141]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9857, 0.0151],\n",
      "        [0.9880, 0.0120],\n",
      "        [0.9883, 0.0113],\n",
      "        [0.9845, 0.0129]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9926, 0.0098],\n",
      "        [0.9857, 0.0101],\n",
      "        [0.9907, 0.0097],\n",
      "        [0.9911, 0.0084]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9895, 0.0114],\n",
      "        [0.9915, 0.0063],\n",
      "        [0.9867, 0.0080],\n",
      "        [0.9922, 0.0104]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9884, 0.0071],\n",
      "        [0.9933, 0.0056],\n",
      "        [0.9904, 0.0068],\n",
      "        [0.9918, 0.0090]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9923, 0.0062],\n",
      "        [0.9936, 0.0057],\n",
      "        [0.9931, 0.0122],\n",
      "        [0.9931, 0.0073]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9926, 0.0067],\n",
      "        [0.9924, 0.0129],\n",
      "        [0.9940, 0.0046],\n",
      "        [0.9940, 0.0046]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9936, 0.0062],\n",
      "        [0.9933, 0.0066],\n",
      "        [0.9929, 0.0042],\n",
      "        [0.9953, 0.0040]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9959, 0.0045],\n",
      "        [0.9945, 0.0064],\n",
      "        [0.9953, 0.0080],\n",
      "        [0.9952, 0.0048]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9956, 0.0040],\n",
      "        [0.9891, 0.0114],\n",
      "        [0.9963, 0.0030],\n",
      "        [0.9955, 0.0052]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9936, 0.0046],\n",
      "        [0.9956, 0.0045],\n",
      "        [0.9944, 0.0041],\n",
      "        [0.9938, 0.0061]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9943, 0.0054],\n",
      "        [0.9950, 0.0050],\n",
      "        [0.9943, 0.0046],\n",
      "        [0.9939, 0.0058]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9953, 0.0033],\n",
      "        [0.9966, 0.0039],\n",
      "        [0.9967, 0.0027],\n",
      "        [0.9945, 0.0045]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9957, 0.0036],\n",
      "        [0.9950, 0.0063],\n",
      "        [0.9973, 0.0026],\n",
      "        [0.9965, 0.0033]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9964, 0.0034],\n",
      "        [0.9959, 0.0031],\n",
      "        [0.9971, 0.0020],\n",
      "        [0.9955, 0.0041]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9969, 0.0032],\n",
      "        [0.9954, 0.0040],\n",
      "        [0.9963, 0.0032],\n",
      "        [0.9961, 0.0034]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9962, 0.0035],\n",
      "        [0.9968, 0.0047],\n",
      "        [0.9973, 0.0027],\n",
      "        [0.9968, 0.0029]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9971, 0.0033],\n",
      "        [0.9968, 0.0028],\n",
      "        [0.9959, 0.0038],\n",
      "        [0.9955, 0.0061]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9970, 0.0030],\n",
      "        [0.9975, 0.0023],\n",
      "        [0.9971, 0.0020],\n",
      "        [0.9967, 0.0030]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0022],\n",
      "        [0.9980, 0.0022],\n",
      "        [0.9978, 0.0017],\n",
      "        [0.9971, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9973, 0.0055],\n",
      "        [0.9963, 0.0028],\n",
      "        [0.9955, 0.0030],\n",
      "        [0.9982, 0.0020]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9968, 0.0030],\n",
      "        [0.9958, 0.0024],\n",
      "        [0.9971, 0.0024],\n",
      "        [0.9976, 0.0022]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9963, 0.0034],\n",
      "        [0.9983, 0.0018],\n",
      "        [0.9970, 0.0034],\n",
      "        [0.9982, 0.0016]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9981, 0.0019],\n",
      "        [0.9976, 0.0029],\n",
      "        [0.9984, 0.0016],\n",
      "        [0.9977, 0.0025]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9980, 0.0020],\n",
      "        [0.9983, 0.0038],\n",
      "        [0.9980, 0.0017],\n",
      "        [0.9971, 0.0018]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9975, 0.0032],\n",
      "        [0.9973, 0.0020],\n",
      "        [0.9981, 0.0018],\n",
      "        [0.9967, 0.0022]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0023],\n",
      "        [0.9965, 0.0026],\n",
      "        [0.9982, 0.0019],\n",
      "        [0.9979, 0.0039]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9978, 0.0018],\n",
      "        [0.9979, 0.0022],\n",
      "        [0.9975, 0.0022],\n",
      "        [0.9979, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9985, 0.0017],\n",
      "        [0.9972, 0.0020],\n",
      "        [0.9976, 0.0030],\n",
      "        [0.9978, 0.0035]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9984, 0.0022],\n",
      "        [0.9982, 0.0022],\n",
      "        [0.9985, 0.0015],\n",
      "        [0.9983, 0.0021]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9986, 0.0015],\n",
      "        [0.9980, 0.0025],\n",
      "        [0.9977, 0.0026],\n",
      "        [0.9985, 0.0014]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9983, 0.0011],\n",
      "        [0.9966, 0.0026],\n",
      "        [0.9980, 0.0018],\n",
      "        [0.9973, 0.0031]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9982, 0.0011],\n",
      "        [0.9985, 0.0019],\n",
      "        [0.9961, 0.0024],\n",
      "        [0.9985, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[9.9889e-01, 9.6877e-04],\n",
      "        [9.9678e-01, 3.2537e-03],\n",
      "        [9.9751e-01, 1.7144e-03],\n",
      "        [9.9839e-01, 1.3920e-03]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9976, 0.0030],\n",
      "        [0.9977, 0.0022],\n",
      "        [0.9985, 0.0013],\n",
      "        [0.9985, 0.0012]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 1, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [0., 1.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9871, 0.0180],\n",
      "        [0.9839, 0.0163],\n",
      "        [0.9832, 0.0135],\n",
      "        [0.9811, 0.0183]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9789, 0.0329],\n",
      "        [0.9811, 0.0274],\n",
      "        [0.9736, 0.0218],\n",
      "        [0.9790, 0.0189]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9749, 0.0274],\n",
      "        [0.9743, 0.0189],\n",
      "        [0.9791, 0.0204],\n",
      "        [0.9752, 0.0232]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9840, 0.0194],\n",
      "        [0.9820, 0.0173],\n",
      "        [0.9722, 0.0215],\n",
      "        [0.9772, 0.0244]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([4, 1, 23, 15360])\n",
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "tensor([[0.9764, 0.0225],\n",
      "        [0.9781, 0.0191],\n",
      "        [0.9743, 0.0232],\n",
      "        [0.9787, 0.0227]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0, 0, 0, 0], device='cuda:0')\n",
      "tensor([[1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.],\n",
      "        [1., 0.]], device='cuda:0')\n",
      "torch.Size([1, 1, 23, 15360])\n",
      "batch_size :  1\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (1, 1018, 10)\n",
      "tensor([[0.9813, 0.0154]], device='cuda:0', grad_fn=<SigmoidBackward0>)\n",
      "tensor([0], device='cuda:0')\n",
      "tensor([[1., 0.]], device='cuda:0')\n",
      "torch.Size([479, 1, 23, 15360])\n"
     ]
    },
    {
     "ename": "OutOfMemoryError",
     "evalue": "CUDA out of memory. Tried to allocate 25.18 GiB. GPU 0 has a total capacity of 4.00 GiB of which 2.49 GiB is free. Of the allocated memory 668.29 MiB is allocated by PyTorch, and 37.71 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mOutOfMemoryError\u001b[0m                          Traceback (most recent call last)",
      "Cell \u001b[1;32mIn[10], line 1\u001b[0m\n\u001b[1;32m----> 1\u001b[0m \u001b[43mtrain\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtrain_data\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtrain_labels\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtest_data\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtest_labels\u001b[49m\u001b[43m)\u001b[49m\n",
      "Cell \u001b[1;32mIn[9], line 51\u001b[0m, in \u001b[0;36mtrain\u001b[1;34m(train_data, train_label, test_data, test_label)\u001b[0m\n\u001b[0;32m     49\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m (e \u001b[38;5;241m+\u001b[39m \u001b[38;5;241m1\u001b[39m) \u001b[38;5;241m%\u001b[39m \u001b[38;5;241m1\u001b[39m \u001b[38;5;241m==\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[0;32m     50\u001b[0m     model\u001b[38;5;241m.\u001b[39meval()\n\u001b[1;32m---> 51\u001b[0m     Cls \u001b[38;5;241m=\u001b[39m \u001b[43mmodel\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtest_data\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m     54\u001b[0m     loss_test \u001b[38;5;241m=\u001b[39m criterion(Cls, test_label)\n\u001b[0;32m     55\u001b[0m     y_pred \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mmax(Cls, \u001b[38;5;241m1\u001b[39m)[\u001b[38;5;241m1\u001b[39m]\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1509\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[0;32m   1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m-> 1511\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[0;32m   1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[0;32m   1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[0;32m   1518\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[0;32m   1519\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[1;32m-> 1520\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m   1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m   1523\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\container.py:217\u001b[0m, in \u001b[0;36mSequential.forward\u001b[1;34m(self, input)\u001b[0m\n\u001b[0;32m    215\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m):\n\u001b[0;32m    216\u001b[0m     \u001b[38;5;28;01mfor\u001b[39;00m module \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m:\n\u001b[1;32m--> 217\u001b[0m         \u001b[38;5;28minput\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[43mmodule\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[0;32m    218\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28minput\u001b[39m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1509\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[0;32m   1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m-> 1511\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[0;32m   1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[0;32m   1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[0;32m   1518\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[0;32m   1519\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[1;32m-> 1520\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m   1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m   1523\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
      "File \u001b[1;32mc:\\Users\\MAREZ10\\OneDrive - Université Laval\\Documents\\GitHub\\EEG_Seizures_Transformers-\\convolution.py:46\u001b[0m, in \u001b[0;36mPatchEmbedding.forward\u001b[1;34m(self, x)\u001b[0m\n\u001b[0;32m     44\u001b[0m x \u001b[38;5;241m=\u001b[39m x\u001b[38;5;241m.\u001b[39munsqueeze(\u001b[38;5;241m1\u001b[39m)\n\u001b[0;32m     45\u001b[0m \u001b[38;5;28mprint\u001b[39m(x\u001b[38;5;241m.\u001b[39mshape)\n\u001b[1;32m---> 46\u001b[0m x \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mshallownet\u001b[49m\u001b[43m(\u001b[49m\u001b[43mx\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m     47\u001b[0m x \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mprojection(x)\n\u001b[0;32m     48\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m x\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1509\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[0;32m   1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m-> 1511\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[0;32m   1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[0;32m   1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[0;32m   1518\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[0;32m   1519\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[1;32m-> 1520\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m   1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m   1523\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\container.py:217\u001b[0m, in \u001b[0;36mSequential.forward\u001b[1;34m(self, input)\u001b[0m\n\u001b[0;32m    215\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m):\n\u001b[0;32m    216\u001b[0m     \u001b[38;5;28;01mfor\u001b[39;00m module \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m:\n\u001b[1;32m--> 217\u001b[0m         \u001b[38;5;28minput\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[43mmodule\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[0;32m    218\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28minput\u001b[39m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1509\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)  \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[0;32m   1510\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m-> 1511\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1515\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[0;32m   1516\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[0;32m   1517\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[0;32m   1518\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[0;32m   1519\u001b[0m         \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[1;32m-> 1520\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[0;32m   1522\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m   1523\u001b[0m     result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\conv.py:460\u001b[0m, in \u001b[0;36mConv2d.forward\u001b[1;34m(self, input)\u001b[0m\n\u001b[0;32m    459\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Tensor) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Tensor:\n\u001b[1;32m--> 460\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_conv_forward\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[1;32m~\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python311\\site-packages\\torch\\nn\\modules\\conv.py:456\u001b[0m, in \u001b[0;36mConv2d._conv_forward\u001b[1;34m(self, input, weight, bias)\u001b[0m\n\u001b[0;32m    452\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode \u001b[38;5;241m!=\u001b[39m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mzeros\u001b[39m\u001b[38;5;124m'\u001b[39m:\n\u001b[0;32m    453\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m F\u001b[38;5;241m.\u001b[39mconv2d(F\u001b[38;5;241m.\u001b[39mpad(\u001b[38;5;28minput\u001b[39m, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_reversed_padding_repeated_twice, mode\u001b[38;5;241m=\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode),\n\u001b[0;32m    454\u001b[0m                     weight, bias, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mstride,\n\u001b[0;32m    455\u001b[0m                     _pair(\u001b[38;5;241m0\u001b[39m), \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdilation, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mgroups)\n\u001b[1;32m--> 456\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mF\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv2d\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mbias\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstride\u001b[49m\u001b[43m,\u001b[49m\n\u001b[0;32m    457\u001b[0m \u001b[43m                \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mpadding\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdilation\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgroups\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[1;31mOutOfMemoryError\u001b[0m: CUDA out of memory. Tried to allocate 25.18 GiB. GPU 0 has a total capacity of 4.00 GiB of which 2.49 GiB is free. Of the allocated memory 668.29 MiB is allocated by PyTorch, and 37.71 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)"
     ]
    }
   ],
   "source": [
    "train(train_data, train_labels, test_data, test_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TEST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([4, 1, 23, 15360])\n"
     ]
    }
   ],
   "source": [
    "from convolution import PatchEmbedding\n",
    "\n",
    "x = torch.randn((4, 23, 15360))\n",
    "\n",
    "conv = PatchEmbedding(40,23)\n",
    "\n",
    "X = conv(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "batch_size :  4\n",
      "Sequence length :  1018\n",
      "dimension embeddings :  40\n",
      "Shape output \t\t= (4, 1018, 10)\n",
      "Shape output \t\t= (4, 1018, 10)\n"
     ]
    }
   ],
   "source": [
    "reload(transformers)\n",
    "from transformers import TransformerEncoder\n",
    "\n",
    "x = torch.randn((2, 64, 64))\n",
    "\n",
    "transformer_enc = TransformerEncoder(6, 40, 2, 10, 0.0)\n",
    "output = transformer_enc(X)\n",
    "\n",
    "print(f'Shape output \\t\\t= {tuple(output.shape)}')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[-0.0124, -0.0361],\n",
       "        [-0.0234,  0.0205],\n",
       "        [-0.0421, -0.0397],\n",
       "        [-0.0418,  0.0534]], grad_fn=<AddmmBackward0>)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reload(transformers)\n",
    "from transformers import TransformerClassifier\n",
    "\n",
    "clc = TransformerClassifier(output, 2)\n",
    "\n",
    "res = clc(output)\n",
    "res"
   ]
  }
 ],
 "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}