10_DSCMR_CVPR2019.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "DSCMR_CVPR2019.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNEx0Phm3lcHXRIrU1H0+wV",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/OUCTheoryGroup/colab_demo/blob/master/DSCMR_CVPR2019.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fbBzhs2W0cIB",
        "colab_type": "text"
      },
      "source": [
        "## DSCMR 深度监督跨模态检索 \n",
        "\n",
        "Liangli Zhen, Peng Hu, Xu Wang, Dezhong Peng, Deep Supervised Cross-modal Retrieval. *CVPR 2029*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cXtXi0pm_Ucn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "! mkdir pascal\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/test_img.mat\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/test_img_lab.mat\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/test_txt.mat\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/train_img.mat\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/train_img_lab.mat\n",
        "! wget -P pascal https://raw.githubusercontent.com/penghu-cs/DSCMR/master/data/pascal/train_txt.mat"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CV8nHBqm0Zi9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "from datetime import datetime\n",
        "import torch.optim as optim\n",
        "import matplotlib.pyplot as plt\n",
        "from torch.utils.data.dataset import Dataset\n",
        "from scipy.io import loadmat, savemat\n",
        "from torch.utils.data import DataLoader\n",
        "import numpy as np\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "from torchvision import models\n",
        "import time\n",
        "import copy\n",
        "import scipy.spatial"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Og8AlAbIFFVv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class ImgNN(nn.Module):\n",
        "    \"\"\"Network to learn image representations\"\"\"\n",
        "    def __init__(self, input_dim=4096, output_dim=1024):\n",
        "        super(ImgNN, self).__init__()\n",
        "        self.denseL1 = nn.Linear(input_dim, output_dim)\n",
        "        \n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.denseL1(x))\n",
        "        return out\n",
        "\n",
        "class TextNN(nn.Module):\n",
        "    \"\"\"Network to learn text representations\"\"\"\n",
        "    def __init__(self, input_dim=1024, output_dim=1024):\n",
        "        super(TextNN, self).__init__()\n",
        "        self.denseL1 = nn.Linear(input_dim, output_dim)\n",
        "\n",
        "    def forward(self, x):\n",
        "        out = F.relu(self.denseL1(x))\n",
        "        return out"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1o-oP0vtCvSL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class IDCM_NN(nn.Module):\n",
        "    \"\"\"Network to learn text representations\"\"\"\n",
        "    def __init__(self, img_input_dim=4096, img_output_dim=2048,\n",
        "                 text_input_dim=1024, text_output_dim=2048, minus_one_dim=1024, output_dim=10):\n",
        "        super(IDCM_NN, self).__init__()\n",
        "        self.img_net = ImgNN(img_input_dim, img_output_dim)\n",
        "        self.text_net = TextNN(text_input_dim, text_output_dim)\n",
        "        self.linearLayer = nn.Linear(img_output_dim, minus_one_dim)\n",
        "        self.linearLayer2 = nn.Linear(minus_one_dim, output_dim)\n",
        "\n",
        "    def forward(self, img, text):\n",
        "        view1_feature = self.img_net(img)\n",
        "        view2_feature = self.text_net(text)\n",
        "        view1_feature = self.linearLayer(view1_feature)\n",
        "        view2_feature = self.linearLayer(view2_feature)\n",
        "\n",
        "        view1_predict = self.linearLayer2(view1_feature)\n",
        "        view2_predict = self.linearLayer2(view2_feature)\n",
        "        return view1_feature, view2_feature, view1_predict, view2_predict"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ogvLSuoX8LIF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class CustomDataSet(Dataset):\n",
        "    def __init__(\n",
        "            self,\n",
        "            images,\n",
        "            texts,\n",
        "            labels):\n",
        "        self.images = images\n",
        "        self.texts = texts\n",
        "        self.labels = labels\n",
        "\n",
        "    def __getitem__(self, index):\n",
        "        img = self.images[index]\n",
        "        text = self.texts[index]\n",
        "        label = self.labels[index]\n",
        "        return img, text, label\n",
        "\n",
        "    def __len__(self):\n",
        "        count = len(self.images)\n",
        "        assert len(\n",
        "            self.images) == len(self.labels)\n",
        "        return count\n",
        "\n",
        "def ind2vec(ind, N=None):\n",
        "    ind = np.asarray(ind)\n",
        "    if N is None:\n",
        "        N = ind.max() + 1\n",
        "    return np.arange(N) == np.repeat(ind, N, axis=1)\n",
        "\n",
        "def get_loader(path, batch_size):\n",
        "    img_train = loadmat(path+\"train_img.mat\")['train_img']\n",
        "    img_test = loadmat(path + \"test_img.mat\")['test_img']\n",
        "    text_train = loadmat(path+\"train_txt.mat\")['train_txt']\n",
        "    text_test = loadmat(path + \"test_txt.mat\")['test_txt']\n",
        "    label_train = loadmat(path+\"train_img_lab.mat\")['train_img_lab']\n",
        "    label_test = loadmat(path + \"test_img_lab.mat\")['test_img_lab']\n",
        "\n",
        "    label_train = ind2vec(label_train).astype(int)\n",
        "    label_test = ind2vec(label_test).astype(int)\n",
        "\n",
        "    imgs = {'train': img_train, 'test': img_test}\n",
        "    texts = {'train': text_train, 'test': text_test}\n",
        "    labels = {'train': label_train, 'test': label_test}\n",
        "    dataset = {x: CustomDataSet(images=imgs[x], texts=texts[x], labels=labels[x])\n",
        "               for x in ['train', 'test']}\n",
        "\n",
        "    shuffle = {'train': False, 'test': False}\n",
        "\n",
        "    dataloader = {x: DataLoader(dataset[x], batch_size=batch_size,\n",
        "                                shuffle=shuffle[x], num_workers=0) for x in ['train', 'test']}\n",
        "\n",
        "    img_dim = img_train.shape[1]\n",
        "    text_dim = text_train.shape[1]\n",
        "    num_class = label_train.shape[1]\n",
        "\n",
        "    input_data_par = {}\n",
        "    input_data_par['img_test'] = img_test\n",
        "    input_data_par['text_test'] = text_test\n",
        "    input_data_par['label_test'] = label_test\n",
        "    input_data_par['img_train'] = img_train\n",
        "    input_data_par['text_train'] = text_train\n",
        "    input_data_par['label_train'] = label_train\n",
        "    input_data_par['img_dim'] = img_dim\n",
        "    input_data_par['text_dim'] = text_dim\n",
        "    input_data_par['num_class'] = num_class\n",
        "    return dataloader, input_data_par"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uHOEGdPPFXdV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def train_model(model, data_loaders, optimizer, alpha, beta, device=\"cpu\", num_epochs=500):\n",
        "    since = time.time()\n",
        "    # device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "    test_img_acc_history = []\n",
        "    test_txt_acc_history = []\n",
        "    epoch_loss_history =[]\n",
        "\n",
        "    best_model_wts = copy.deepcopy(model.state_dict())\n",
        "    best_acc = 0.0\n",
        "\n",
        "    for epoch in range(num_epochs):\n",
        "        print('Epoch {}/{}'.format(epoch+1, num_epochs))\n",
        "        print('-' * 20)\n",
        "\n",
        "        # Each epoch has a training and validation phase\n",
        "        for phase in ['train', 'test']:\n",
        "            if phase == 'train':\n",
        "                # Set model to training mode\n",
        "                model.train()\n",
        "            else:\n",
        "                # Set model to evaluate mode\n",
        "                model.eval()\n",
        "\n",
        "            running_loss = 0.0\n",
        "            running_corrects_img = 0\n",
        "            running_corrects_txt = 0\n",
        "            # Iterate over data.\n",
        "            for imgs, txts, labels in data_loaders[phase]:\n",
        "                # imgs = imgs.to(device)\n",
        "                # txts = txts.to(device)\n",
        "                # labels = labels.to(device)\n",
        "                if torch.sum(imgs != imgs)>1 or torch.sum(txts != txts)>1:\n",
        "                    print(\"Data contains Nan.\")\n",
        "\n",
        "                # zero the parameter gradients\n",
        "                optimizer.zero_grad()\n",
        "\n",
        "                # forward\n",
        "                # track history if only in train\n",
        "                with torch.set_grad_enabled(phase == 'train'):\n",
        "                    # Get model outputs and calculate loss\n",
        "                    # Special case for inception because in training it has an auxiliary output. In train\n",
        "                    #   mode we calculate the loss by summing the final output and the auxiliary output\n",
        "                    #   but in testing we only consider the final output.\n",
        "                    if torch.cuda.is_available():\n",
        "                        imgs = imgs.cuda()\n",
        "                        txts = txts.cuda()\n",
        "                        labels = labels.cuda()\n",
        "\n",
        "\n",
        "                    # zero the parameter gradients\n",
        "                    optimizer.zero_grad()\n",
        "\n",
        "                    # Forward\n",
        "                    view1_feature, view2_feature, view1_predict, view2_predict = model(imgs, txts)\n",
        "\n",
        "                    loss = calc_loss(view1_feature, view2_feature, view1_predict,\n",
        "                                     view2_predict, labels, labels, alpha, beta)\n",
        "\n",
        "                    img_preds = view1_predict\n",
        "                    txt_preds = view2_predict\n",
        "\n",
        "                    # backward + optimize only if in training phase\n",
        "                    if phase == 'train':\n",
        "                        loss.backward()\n",
        "                        optimizer.step()\n",
        "\n",
        "                # statistics\n",
        "                running_loss += loss.item()\n",
        "                running_corrects_img += torch.sum(torch.argmax(img_preds, dim=1) == torch.argmax(labels, dim=1))\n",
        "                running_corrects_txt += torch.sum(torch.argmax(txt_preds, dim=1) == torch.argmax(labels, dim=1))\n",
        "\n",
        "            epoch_loss = running_loss / len(data_loaders[phase].dataset)\n",
        "            # epoch_img_acc = running_corrects_img.double() / len(data_loaders[phase].dataset)\n",
        "            # epoch_txt_acc = running_corrects_txt.double() / len(data_loaders[phase].dataset)\n",
        "            t_imgs, t_txts, t_labels = [], [], []\n",
        "            with torch.no_grad():\n",
        "                for imgs, txts, labels in data_loaders['test']:\n",
        "                    if torch.cuda.is_available():\n",
        "                            imgs = imgs.cuda()\n",
        "                            txts = txts.cuda()\n",
        "                            labels = labels.cuda()\n",
        "                    t_view1_feature, t_view2_feature, _, _ = model(imgs, txts)\n",
        "                    t_imgs.append(t_view1_feature.cpu().numpy())\n",
        "                    t_txts.append(t_view2_feature.cpu().numpy())\n",
        "                    t_labels.append(labels.cpu().numpy())\n",
        "            t_imgs = np.concatenate(t_imgs)\n",
        "            t_txts = np.concatenate(t_txts)\n",
        "            t_labels = np.concatenate(t_labels).argmax(1)\n",
        "            img2text = fx_calc_map_label(t_imgs, t_txts, t_labels)\n",
        "            txt2img = fx_calc_map_label(t_txts, t_imgs, t_labels)\n",
        "\n",
        "            print('{} Loss: {:.4f} Img2Txt: {:.4f}  Txt2Img: {:.4f}'.format(phase, epoch_loss, img2text, txt2img))\n",
        "\n",
        "            # deep copy the model\n",
        "            if phase == 'test' and (img2text + txt2img) / 2. > best_acc:\n",
        "                best_acc = (img2text + txt2img) / 2.\n",
        "                best_model_wts = copy.deepcopy(model.state_dict())\n",
        "            if phase == 'test':\n",
        "                test_img_acc_history.append(img2text)\n",
        "                test_txt_acc_history.append(txt2img)\n",
        "                epoch_loss_history.append(epoch_loss)\n",
        "\n",
        "        print()\n",
        "\n",
        "    time_elapsed = time.time() - since\n",
        "    print('Training complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))\n",
        "    print('Best average ACC: {:4f}'.format(best_acc))\n",
        "\n",
        "    # load best model weights\n",
        "    model.load_state_dict(best_model_wts)\n",
        "    return model, test_img_acc_history, test_txt_acc_history, epoch_loss_history\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r6rxREl4Fo4G",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def calc_loss(view1_feature, view2_feature, view1_predict, view2_predict, labels_1, labels_2, alpha, beta):\n",
        "    term1 = ((view1_predict-labels_1.float())**2).sum(1).sqrt().mean() + ((view2_predict-labels_2.float())**2).sum(1).sqrt().mean()\n",
        "\n",
        "    cos = lambda x, y: x.mm(y.t()) / ((x ** 2).sum(1, keepdim=True).sqrt().mm((y ** 2).sum(1, keepdim=True).sqrt().t())).clamp(min=1e-6) / 2.\n",
        "    theta11 = cos(view1_feature, view1_feature)\n",
        "    theta12 = cos(view1_feature, view2_feature)\n",
        "    theta22 = cos(view2_feature, view2_feature)\n",
        "    Sim11 = calc_label_sim(labels_1, labels_1).float()\n",
        "    Sim12 = calc_label_sim(labels_1, labels_2).float()\n",
        "    Sim22 = calc_label_sim(labels_2, labels_2).float()\n",
        "    term21 = ((1+torch.exp(theta11)).log() - Sim11 * theta11).mean()\n",
        "    term22 = ((1+torch.exp(theta12)).log() - Sim12 * theta12).mean()\n",
        "    term23 = ((1 + torch.exp(theta22)).log() - Sim22 * theta22).mean()\n",
        "    term2 = term21 + term22 + term23\n",
        "\n",
        "    term3 = ((view1_feature - view2_feature)**2).sum(1).sqrt().mean()\n",
        "\n",
        "    im_loss = term1 + alpha * term2 + beta * term3\n",
        "    return im_loss\n",
        "\n",
        "def calc_label_sim(label_1, label_2):\n",
        "    Sim = label_1.float().mm(label_2.float().t())\n",
        "    return Sim"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QEfNLK1DFyfl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def fx_calc_map_label(image, text, label, k = 0, dist_method='COS'):\n",
        "  if dist_method == 'L2':\n",
        "    dist = scipy.spatial.distance.cdist(image, text, 'euclidean')\n",
        "  elif dist_method == 'COS':\n",
        "    dist = scipy.spatial.distance.cdist(image, text, 'cosine')\n",
        "  ord = dist.argsort()\n",
        "  numcases = dist.shape[0]\n",
        "  if k == 0:\n",
        "    k = numcases\n",
        "  res = []\n",
        "  for i in range(numcases):\n",
        "    order = ord[i]\n",
        "    p = 0.0\n",
        "    r = 0.0\n",
        "    for j in range(k):\n",
        "      if label[i] == label[order[j]]:\n",
        "        r += 1\n",
        "        p += (r / (j + 1))\n",
        "    if r > 0:\n",
        "      res += [p / r]\n",
        "    else:\n",
        "      res += [0]\n",
        "  return np.mean(res)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bkfjTvSk73dB",
        "colab_type": "code",
        "outputId": "5920c8ec-e4f2-4f02-9c50-9c68a2a920d5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "# data parameters\n",
        "DATA_DIR = 'pascal/'\n",
        "alpha = 1e-3\n",
        "beta = 1e-1\n",
        "MAX_EPOCH = 200\n",
        "batch_size = 100\n",
        "# batch_size = 512\n",
        "lr = 1e-4\n",
        "betas = (0.5, 0.999)\n",
        "weight_decay = 0\n",
        "\n",
        "print('...Data loading is beginning...')\n",
        "\n",
        "data_loader, input_data_par = get_loader(DATA_DIR, batch_size)\n",
        "\n",
        "print('...Data loading is completed...')\n",
        "\n",
        "model_ft = IDCM_NN(img_input_dim=input_data_par['img_dim'], text_input_dim=input_data_par['text_dim'], output_dim=input_data_par['num_class']).to(device)\n",
        "params_to_update = list(model_ft.parameters())\n",
        "\n",
        "# Observe that all parameters are being optimized\n",
        "optimizer = optim.Adam(params_to_update, lr=lr, betas=betas)\n",
        "\n",
        "print('...Training is beginning...')\n",
        "# Train and evaluate\n",
        "model_ft, img_acc_hist, txt_acc_hist, loss_hist = train_model(model_ft, data_loader, optimizer, alpha, beta, MAX_EPOCH)\n",
        "print('...Training is completed...')\n",
        "\n",
        "print('...Evaluation on testing data...')\n",
        "view1_feature, view2_feature, view1_predict, view2_predict = model_ft(torch.tensor(input_data_par['img_test']).to(device), torch.tensor(input_data_par['text_test']).to(device))\n",
        "label = torch.argmax(torch.tensor(input_data_par['label_test']), dim=1)\n",
        "view1_feature = view1_feature.detach().cpu().numpy()\n",
        "view2_feature = view2_feature.detach().cpu().numpy()\n",
        "view1_predict = view1_predict.detach().cpu().numpy()\n",
        "view2_predict = view2_predict.detach().cpu().numpy()\n",
        "img_to_txt = fx_calc_map_label(view1_feature, view2_feature, label)\n",
        "print('...Image to Text MAP = {}'.format(img_to_txt))\n",
        "\n",
        "txt_to_img = fx_calc_map_label(view2_feature, view1_feature, label)\n",
        "print('...Text to Image MAP = {}'.format(txt_to_img))\n",
        "\n",
        "print('...Average MAP = {}'.format(((img_to_txt + txt_to_img) / 2.)))\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "...Data loading is beginning...\n",
            "...Data loading is completed...\n",
            "...Training is beginning...\n",
            "Epoch 1/500\n",
            "--------------------\n",
            "train Loss: 0.0267 Img2Txt: 0.2081  Txt2Img: 0.2477\n",
            "test Loss: 0.0229 Img2Txt: 0.2081  Txt2Img: 0.2477\n",
            "\n",
            "Epoch 2/500\n",
            "--------------------\n",
            "train Loss: 0.0228 Img2Txt: 0.4067  Txt2Img: 0.4179\n",
            "test Loss: 0.0204 Img2Txt: 0.4067  Txt2Img: 0.4179\n",
            "\n",
            "Epoch 3/500\n",
            "--------------------\n",
            "train Loss: 0.0207 Img2Txt: 0.5217  Txt2Img: 0.5267\n",
            "test Loss: 0.0187 Img2Txt: 0.5217  Txt2Img: 0.5267\n",
            "\n",
            "Epoch 4/500\n",
            "--------------------\n",
            "train Loss: 0.0187 Img2Txt: 0.5594  Txt2Img: 0.5960\n",
            "test Loss: 0.0174 Img2Txt: 0.5594  Txt2Img: 0.5960\n",
            "\n",
            "Epoch 5/500\n",
            "--------------------\n",
            "train Loss: 0.0170 Img2Txt: 0.5938  Txt2Img: 0.6371\n",
            "test Loss: 0.0165 Img2Txt: 0.5938  Txt2Img: 0.6371\n",
            "\n",
            "Epoch 6/500\n",
            "--------------------\n",
            "train Loss: 0.0156 Img2Txt: 0.6224  Txt2Img: 0.6591\n",
            "test Loss: 0.0158 Img2Txt: 0.6224  Txt2Img: 0.6591\n",
            "\n",
            "Epoch 7/500\n",
            "--------------------\n",
            "train Loss: 0.0146 Img2Txt: 0.6271  Txt2Img: 0.6685\n",
            "test Loss: 0.0155 Img2Txt: 0.6271  Txt2Img: 0.6685\n",
            "\n",
            "Epoch 8/500\n",
            "--------------------\n",
            "train Loss: 0.0140 Img2Txt: 0.6461  Txt2Img: 0.6762\n",
            "test Loss: 0.0153 Img2Txt: 0.6461  Txt2Img: 0.6762\n",
            "\n",
            "Epoch 9/500\n",
            "--------------------\n",
            "train Loss: 0.0136 Img2Txt: 0.6570  Txt2Img: 0.6914\n",
            "test Loss: 0.0152 Img2Txt: 0.6570  Txt2Img: 0.6914\n",
            "\n",
            "Epoch 10/500\n",
            "--------------------\n",
            "train Loss: 0.0132 Img2Txt: 0.6597  Txt2Img: 0.6969\n",
            "test Loss: 0.0151 Img2Txt: 0.6597  Txt2Img: 0.6969\n",
            "\n",
            "Epoch 11/500\n",
            "--------------------\n",
            "train Loss: 0.0129 Img2Txt: 0.6583  Txt2Img: 0.7032\n",
            "test Loss: 0.0149 Img2Txt: 0.6583  Txt2Img: 0.7032\n",
            "\n",
            "Epoch 12/500\n",
            "--------------------\n",
            "train Loss: 0.0127 Img2Txt: 0.6686  Txt2Img: 0.7086\n",
            "test Loss: 0.0150 Img2Txt: 0.6686  Txt2Img: 0.7086\n",
            "\n",
            "Epoch 13/500\n",
            "--------------------\n",
            "train Loss: 0.0122 Img2Txt: 0.6693  Txt2Img: 0.7114\n",
            "test Loss: 0.0149 Img2Txt: 0.6693  Txt2Img: 0.7114\n",
            "\n",
            "Epoch 14/500\n",
            "--------------------\n",
            "train Loss: 0.0122 Img2Txt: 0.6628  Txt2Img: 0.7001\n",
            "test Loss: 0.0147 Img2Txt: 0.6628  Txt2Img: 0.7001\n",
            "\n",
            "Epoch 15/500\n",
            "--------------------\n",
            "train Loss: 0.0115 Img2Txt: 0.6705  Txt2Img: 0.7034\n",
            "test Loss: 0.0147 Img2Txt: 0.6705  Txt2Img: 0.7034\n",
            "\n",
            "Epoch 16/500\n",
            "--------------------\n",
            "train Loss: 0.0115 Img2Txt: 0.6731  Txt2Img: 0.7106\n",
            "test Loss: 0.0145 Img2Txt: 0.6731  Txt2Img: 0.7106\n",
            "\n",
            "Epoch 17/500\n",
            "--------------------\n",
            "train Loss: 0.0113 Img2Txt: 0.6743  Txt2Img: 0.7111\n",
            "test Loss: 0.0143 Img2Txt: 0.6743  Txt2Img: 0.7111\n",
            "\n",
            "Epoch 18/500\n",
            "--------------------\n",
            "train Loss: 0.0112 Img2Txt: 0.6703  Txt2Img: 0.7110\n",
            "test Loss: 0.0146 Img2Txt: 0.6703  Txt2Img: 0.7110\n",
            "\n",
            "Epoch 19/500\n",
            "--------------------\n",
            "train Loss: 0.0110 Img2Txt: 0.6691  Txt2Img: 0.7066\n",
            "test Loss: 0.0145 Img2Txt: 0.6691  Txt2Img: 0.7066\n",
            "\n",
            "Epoch 20/500\n",
            "--------------------\n",
            "train Loss: 0.0108 Img2Txt: 0.6630  Txt2Img: 0.7051\n",
            "test Loss: 0.0145 Img2Txt: 0.6630  Txt2Img: 0.7051\n",
            "\n",
            "Epoch 21/500\n",
            "--------------------\n",
            "train Loss: 0.0105 Img2Txt: 0.6805  Txt2Img: 0.7078\n",
            "test Loss: 0.0144 Img2Txt: 0.6805  Txt2Img: 0.7078\n",
            "\n",
            "Epoch 22/500\n",
            "--------------------\n",
            "train Loss: 0.0104 Img2Txt: 0.6785  Txt2Img: 0.7021\n",
            "test Loss: 0.0141 Img2Txt: 0.6785  Txt2Img: 0.7021\n",
            "\n",
            "Epoch 23/500\n",
            "--------------------\n",
            "train Loss: 0.0102 Img2Txt: 0.6947  Txt2Img: 0.7100\n",
            "test Loss: 0.0141 Img2Txt: 0.6947  Txt2Img: 0.7100\n",
            "\n",
            "Epoch 24/500\n",
            "--------------------\n",
            "train Loss: 0.0100 Img2Txt: 0.6649  Txt2Img: 0.6981\n",
            "test Loss: 0.0140 Img2Txt: 0.6649  Txt2Img: 0.6981\n",
            "\n",
            "Epoch 25/500\n",
            "--------------------\n",
            "train Loss: 0.0099 Img2Txt: 0.6690  Txt2Img: 0.7083\n",
            "test Loss: 0.0142 Img2Txt: 0.6690  Txt2Img: 0.7083\n",
            "\n",
            "Epoch 26/500\n",
            "--------------------\n",
            "train Loss: 0.0098 Img2Txt: 0.6654  Txt2Img: 0.6994\n",
            "test Loss: 0.0142 Img2Txt: 0.6654  Txt2Img: 0.6994\n",
            "\n",
            "Epoch 27/500\n",
            "--------------------\n",
            "train Loss: 0.0096 Img2Txt: 0.6643  Txt2Img: 0.6990\n",
            "test Loss: 0.0138 Img2Txt: 0.6643  Txt2Img: 0.6990\n",
            "\n",
            "Epoch 28/500\n",
            "--------------------\n",
            "train Loss: 0.0095 Img2Txt: 0.6888  Txt2Img: 0.7041\n",
            "test Loss: 0.0138 Img2Txt: 0.6888  Txt2Img: 0.7041\n",
            "\n",
            "Epoch 29/500\n",
            "--------------------\n",
            "train Loss: 0.0093 Img2Txt: 0.6700  Txt2Img: 0.7053\n",
            "test Loss: 0.0136 Img2Txt: 0.6700  Txt2Img: 0.7053\n",
            "\n",
            "Epoch 30/500\n",
            "--------------------\n",
            "train Loss: 0.0092 Img2Txt: 0.6714  Txt2Img: 0.7076\n",
            "test Loss: 0.0139 Img2Txt: 0.6714  Txt2Img: 0.7076\n",
            "\n",
            "Epoch 31/500\n",
            "--------------------\n",
            "train Loss: 0.0092 Img2Txt: 0.6674  Txt2Img: 0.7046\n",
            "test Loss: 0.0139 Img2Txt: 0.6674  Txt2Img: 0.7046\n",
            "\n",
            "Epoch 32/500\n",
            "--------------------\n",
            "train Loss: 0.0091 Img2Txt: 0.6472  Txt2Img: 0.6954\n",
            "test Loss: 0.0137 Img2Txt: 0.6472  Txt2Img: 0.6954\n",
            "\n",
            "Epoch 33/500\n",
            "--------------------\n",
            "train Loss: 0.0086 Img2Txt: 0.6878  Txt2Img: 0.6993\n",
            "test Loss: 0.0137 Img2Txt: 0.6878  Txt2Img: 0.6993\n",
            "\n",
            "Epoch 34/500\n",
            "--------------------\n",
            "train Loss: 0.0087 Img2Txt: 0.6543  Txt2Img: 0.6975\n",
            "test Loss: 0.0135 Img2Txt: 0.6543  Txt2Img: 0.6975\n",
            "\n",
            "Epoch 35/500\n",
            "--------------------\n",
            "train Loss: 0.0088 Img2Txt: 0.6737  Txt2Img: 0.6990\n",
            "test Loss: 0.0136 Img2Txt: 0.6737  Txt2Img: 0.6990\n",
            "\n",
            "Epoch 36/500\n",
            "--------------------\n",
            "train Loss: 0.0086 Img2Txt: 0.6552  Txt2Img: 0.6899\n",
            "test Loss: 0.0138 Img2Txt: 0.6552  Txt2Img: 0.6899\n",
            "\n",
            "Epoch 37/500\n",
            "--------------------\n",
            "train Loss: 0.0084 Img2Txt: 0.6593  Txt2Img: 0.6958\n",
            "test Loss: 0.0137 Img2Txt: 0.6593  Txt2Img: 0.6958\n",
            "\n",
            "Epoch 38/500\n",
            "--------------------\n",
            "train Loss: 0.0084 Img2Txt: 0.6394  Txt2Img: 0.6928\n",
            "test Loss: 0.0135 Img2Txt: 0.6394  Txt2Img: 0.6928\n",
            "\n",
            "Epoch 39/500\n",
            "--------------------\n",
            "train Loss: 0.0083 Img2Txt: 0.6828  Txt2Img: 0.6968\n",
            "test Loss: 0.0134 Img2Txt: 0.6828  Txt2Img: 0.6968\n",
            "\n",
            "Epoch 40/500\n",
            "--------------------\n",
            "train Loss: 0.0085 Img2Txt: 0.6639  Txt2Img: 0.6946\n",
            "test Loss: 0.0137 Img2Txt: 0.6639  Txt2Img: 0.6946\n",
            "\n",
            "Epoch 41/500\n",
            "--------------------\n",
            "train Loss: 0.0084 Img2Txt: 0.6677  Txt2Img: 0.6892\n",
            "test Loss: 0.0133 Img2Txt: 0.6677  Txt2Img: 0.6892\n",
            "\n",
            "Epoch 42/500\n",
            "--------------------\n",
            "train Loss: 0.0078 Img2Txt: 0.6612  Txt2Img: 0.6858\n",
            "test Loss: 0.0137 Img2Txt: 0.6612  Txt2Img: 0.6858\n",
            "\n",
            "Epoch 43/500\n",
            "--------------------\n",
            "train Loss: 0.0084 Img2Txt: 0.6698  Txt2Img: 0.6941\n",
            "test Loss: 0.0134 Img2Txt: 0.6698  Txt2Img: 0.6941\n",
            "\n",
            "Epoch 44/500\n",
            "--------------------\n",
            "train Loss: 0.0078 Img2Txt: 0.6708  Txt2Img: 0.6920\n",
            "test Loss: 0.0135 Img2Txt: 0.6708  Txt2Img: 0.6920\n",
            "\n",
            "Epoch 45/500\n",
            "--------------------\n",
            "train Loss: 0.0085 Img2Txt: 0.6421  Txt2Img: 0.6789\n",
            "test Loss: 0.0136 Img2Txt: 0.6421  Txt2Img: 0.6789\n",
            "\n",
            "Epoch 46/500\n",
            "--------------------\n",
            "train Loss: 0.0081 Img2Txt: 0.6530  Txt2Img: 0.6937\n",
            "test Loss: 0.0136 Img2Txt: 0.6530  Txt2Img: 0.6937\n",
            "\n",
            "Epoch 47/500\n",
            "--------------------\n",
            "train Loss: 0.0081 Img2Txt: 0.6705  Txt2Img: 0.7004\n",
            "test Loss: 0.0135 Img2Txt: 0.6705  Txt2Img: 0.7004\n",
            "\n",
            "Epoch 48/500\n",
            "--------------------\n",
            "train Loss: 0.0079 Img2Txt: 0.6713  Txt2Img: 0.6864\n",
            "test Loss: 0.0135 Img2Txt: 0.6713  Txt2Img: 0.6864\n",
            "\n",
            "Epoch 49/500\n",
            "--------------------\n",
            "train Loss: 0.0080 Img2Txt: 0.6698  Txt2Img: 0.6930\n",
            "test Loss: 0.0131 Img2Txt: 0.6698  Txt2Img: 0.6930\n",
            "\n",
            "Epoch 50/500\n",
            "--------------------\n",
            "train Loss: 0.0075 Img2Txt: 0.6656  Txt2Img: 0.6937\n",
            "test Loss: 0.0136 Img2Txt: 0.6656  Txt2Img: 0.6937\n",
            "\n",
            "Epoch 51/500\n",
            "--------------------\n",
            "train Loss: 0.0078 Img2Txt: 0.6594  Txt2Img: 0.6791\n",
            "test Loss: 0.0134 Img2Txt: 0.6594  Txt2Img: 0.6791\n",
            "\n",
            "Epoch 52/500\n",
            "--------------------\n",
            "train Loss: 0.0076 Img2Txt: 0.6638  Txt2Img: 0.6910\n",
            "test Loss: 0.0133 Img2Txt: 0.6638  Txt2Img: 0.6910\n",
            "\n",
            "Epoch 53/500\n",
            "--------------------\n",
            "train Loss: 0.0075 Img2Txt: 0.6566  Txt2Img: 0.6831\n",
            "test Loss: 0.0134 Img2Txt: 0.6566  Txt2Img: 0.6831\n",
            "\n",
            "Epoch 54/500\n",
            "--------------------\n",
            "train Loss: 0.0078 Img2Txt: 0.6731  Txt2Img: 0.6931\n",
            "test Loss: 0.0134 Img2Txt: 0.6731  Txt2Img: 0.6931\n",
            "\n",
            "Epoch 55/500\n",
            "--------------------\n",
            "train Loss: 0.0077 Img2Txt: 0.6318  Txt2Img: 0.6714\n",
            "test Loss: 0.0134 Img2Txt: 0.6318  Txt2Img: 0.6714\n",
            "\n",
            "Epoch 56/500\n",
            "--------------------\n",
            "train Loss: 0.0075 Img2Txt: 0.6737  Txt2Img: 0.6997\n",
            "test Loss: 0.0134 Img2Txt: 0.6737  Txt2Img: 0.6997\n",
            "\n",
            "Epoch 57/500\n",
            "--------------------\n",
            "train Loss: 0.0075 Img2Txt: 0.6535  Txt2Img: 0.6814\n",
            "test Loss: 0.0131 Img2Txt: 0.6535  Txt2Img: 0.6814\n",
            "\n",
            "Epoch 58/500\n",
            "--------------------\n",
            "train Loss: 0.0075 Img2Txt: 0.6660  Txt2Img: 0.6843\n",
            "test Loss: 0.0135 Img2Txt: 0.6660  Txt2Img: 0.6843\n",
            "\n",
            "Epoch 59/500\n",
            "--------------------\n",
            "train Loss: 0.0074 Img2Txt: 0.6420  Txt2Img: 0.6922\n",
            "test Loss: 0.0131 Img2Txt: 0.6420  Txt2Img: 0.6922\n",
            "\n",
            "Epoch 60/500\n",
            "--------------------\n",
            "train Loss: 0.0074 Img2Txt: 0.6615  Txt2Img: 0.6908\n",
            "test Loss: 0.0134 Img2Txt: 0.6615  Txt2Img: 0.6908\n",
            "\n",
            "Epoch 61/500\n",
            "--------------------\n",
            "train Loss: 0.0076 Img2Txt: 0.6510  Txt2Img: 0.6787\n",
            "test Loss: 0.0133 Img2Txt: 0.6510  Txt2Img: 0.6787\n",
            "\n",
            "Epoch 62/500\n",
            "--------------------\n",
            "train Loss: 0.0073 Img2Txt: 0.6412  Txt2Img: 0.6814\n",
            "test Loss: 0.0133 Img2Txt: 0.6412  Txt2Img: 0.6814\n",
            "\n",
            "Epoch 63/500\n",
            "--------------------\n",
            "train Loss: 0.0073 Img2Txt: 0.6643  Txt2Img: 0.6764\n",
            "test Loss: 0.0132 Img2Txt: 0.6643  Txt2Img: 0.6764\n",
            "\n",
            "Epoch 64/500\n",
            "--------------------\n",
            "train Loss: 0.0073 Img2Txt: 0.6559  Txt2Img: 0.6974\n",
            "test Loss: 0.0133 Img2Txt: 0.6559  Txt2Img: 0.6974\n",
            "\n",
            "Epoch 65/500\n",
            "--------------------\n",
            "train Loss: 0.0072 Img2Txt: 0.6482  Txt2Img: 0.6850\n",
            "test Loss: 0.0132 Img2Txt: 0.6482  Txt2Img: 0.6850\n",
            "\n",
            "Epoch 66/500\n",
            "--------------------\n",
            "train Loss: 0.0070 Img2Txt: 0.6538  Txt2Img: 0.6753\n",
            "test Loss: 0.0131 Img2Txt: 0.6538  Txt2Img: 0.6753\n",
            "\n",
            "Epoch 67/500\n",
            "--------------------\n",
            "train Loss: 0.0071 Img2Txt: 0.6660  Txt2Img: 0.6875\n",
            "test Loss: 0.0130 Img2Txt: 0.6660  Txt2Img: 0.6875\n",
            "\n",
            "Epoch 68/500\n",
            "--------------------\n",
            "train Loss: 0.0070 Img2Txt: 0.6374  Txt2Img: 0.6839\n",
            "test Loss: 0.0134 Img2Txt: 0.6374  Txt2Img: 0.6839\n",
            "\n",
            "Epoch 69/500\n",
            "--------------------\n",
            "train Loss: 0.0072 Img2Txt: 0.6592  Txt2Img: 0.6894\n",
            "test Loss: 0.0134 Img2Txt: 0.6592  Txt2Img: 0.6894\n",
            "\n",
            "Epoch 70/500\n",
            "--------------------\n",
            "train Loss: 0.0072 Img2Txt: 0.6517  Txt2Img: 0.6859\n",
            "test Loss: 0.0131 Img2Txt: 0.6517  Txt2Img: 0.6859\n",
            "\n",
            "Epoch 71/500\n",
            "--------------------\n",
            "train Loss: 0.0068 Img2Txt: 0.6598  Txt2Img: 0.6691\n",
            "test Loss: 0.0131 Img2Txt: 0.6598  Txt2Img: 0.6691\n",
            "\n",
            "Epoch 72/500\n",
            "--------------------\n",
            "train Loss: 0.0073 Img2Txt: 0.6496  Txt2Img: 0.6957\n",
            "test Loss: 0.0131 Img2Txt: 0.6496  Txt2Img: 0.6957\n",
            "\n",
            "Epoch 73/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6487  Txt2Img: 0.6859\n",
            "test Loss: 0.0133 Img2Txt: 0.6487  Txt2Img: 0.6859\n",
            "\n",
            "Epoch 74/500\n",
            "--------------------\n",
            "train Loss: 0.0071 Img2Txt: 0.6181  Txt2Img: 0.6819\n",
            "test Loss: 0.0133 Img2Txt: 0.6181  Txt2Img: 0.6819\n",
            "\n",
            "Epoch 75/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6598  Txt2Img: 0.6840\n",
            "test Loss: 0.0130 Img2Txt: 0.6598  Txt2Img: 0.6840\n",
            "\n",
            "Epoch 76/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6457  Txt2Img: 0.6936\n",
            "test Loss: 0.0131 Img2Txt: 0.6457  Txt2Img: 0.6936\n",
            "\n",
            "Epoch 77/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6440  Txt2Img: 0.6808\n",
            "test Loss: 0.0133 Img2Txt: 0.6440  Txt2Img: 0.6808\n",
            "\n",
            "Epoch 78/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6517  Txt2Img: 0.6862\n",
            "test Loss: 0.0131 Img2Txt: 0.6517  Txt2Img: 0.6862\n",
            "\n",
            "Epoch 79/500\n",
            "--------------------\n",
            "train Loss: 0.0068 Img2Txt: 0.6280  Txt2Img: 0.6800\n",
            "test Loss: 0.0130 Img2Txt: 0.6280  Txt2Img: 0.6800\n",
            "\n",
            "Epoch 80/500\n",
            "--------------------\n",
            "train Loss: 0.0070 Img2Txt: 0.6344  Txt2Img: 0.6720\n",
            "test Loss: 0.0132 Img2Txt: 0.6344  Txt2Img: 0.6720\n",
            "\n",
            "Epoch 81/500\n",
            "--------------------\n",
            "train Loss: 0.0072 Img2Txt: 0.6573  Txt2Img: 0.6899\n",
            "test Loss: 0.0132 Img2Txt: 0.6573  Txt2Img: 0.6899\n",
            "\n",
            "Epoch 82/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6423  Txt2Img: 0.6884\n",
            "test Loss: 0.0129 Img2Txt: 0.6423  Txt2Img: 0.6884\n",
            "\n",
            "Epoch 83/500\n",
            "--------------------\n",
            "train Loss: 0.0068 Img2Txt: 0.6209  Txt2Img: 0.6685\n",
            "test Loss: 0.0131 Img2Txt: 0.6209  Txt2Img: 0.6685\n",
            "\n",
            "Epoch 84/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6470  Txt2Img: 0.6840\n",
            "test Loss: 0.0133 Img2Txt: 0.6470  Txt2Img: 0.6840\n",
            "\n",
            "Epoch 85/500\n",
            "--------------------\n",
            "train Loss: 0.0068 Img2Txt: 0.6463  Txt2Img: 0.6797\n",
            "test Loss: 0.0130 Img2Txt: 0.6463  Txt2Img: 0.6797\n",
            "\n",
            "Epoch 86/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6239  Txt2Img: 0.6772\n",
            "test Loss: 0.0132 Img2Txt: 0.6239  Txt2Img: 0.6772\n",
            "\n",
            "Epoch 87/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6271  Txt2Img: 0.6711\n",
            "test Loss: 0.0131 Img2Txt: 0.6271  Txt2Img: 0.6711\n",
            "\n",
            "Epoch 88/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6276  Txt2Img: 0.6851\n",
            "test Loss: 0.0132 Img2Txt: 0.6276  Txt2Img: 0.6851\n",
            "\n",
            "Epoch 89/500\n",
            "--------------------\n",
            "train Loss: 0.0068 Img2Txt: 0.6444  Txt2Img: 0.6765\n",
            "test Loss: 0.0131 Img2Txt: 0.6444  Txt2Img: 0.6765\n",
            "\n",
            "Epoch 90/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6577  Txt2Img: 0.6880\n",
            "test Loss: 0.0129 Img2Txt: 0.6577  Txt2Img: 0.6880\n",
            "\n",
            "Epoch 91/500\n",
            "--------------------\n",
            "train Loss: 0.0066 Img2Txt: 0.6224  Txt2Img: 0.6743\n",
            "test Loss: 0.0130 Img2Txt: 0.6224  Txt2Img: 0.6743\n",
            "\n",
            "Epoch 92/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6481  Txt2Img: 0.6852\n",
            "test Loss: 0.0132 Img2Txt: 0.6481  Txt2Img: 0.6852\n",
            "\n",
            "Epoch 93/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6382  Txt2Img: 0.6722\n",
            "test Loss: 0.0130 Img2Txt: 0.6382  Txt2Img: 0.6722\n",
            "\n",
            "Epoch 94/500\n",
            "--------------------\n",
            "train Loss: 0.0066 Img2Txt: 0.6441  Txt2Img: 0.6810\n",
            "test Loss: 0.0131 Img2Txt: 0.6441  Txt2Img: 0.6810\n",
            "\n",
            "Epoch 95/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6272  Txt2Img: 0.6825\n",
            "test Loss: 0.0131 Img2Txt: 0.6272  Txt2Img: 0.6825\n",
            "\n",
            "Epoch 96/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6349  Txt2Img: 0.6797\n",
            "test Loss: 0.0133 Img2Txt: 0.6349  Txt2Img: 0.6797\n",
            "\n",
            "Epoch 97/500\n",
            "--------------------\n",
            "train Loss: 0.0069 Img2Txt: 0.6440  Txt2Img: 0.6797\n",
            "test Loss: 0.0129 Img2Txt: 0.6440  Txt2Img: 0.6797\n",
            "\n",
            "Epoch 98/500\n",
            "--------------------\n",
            "train Loss: 0.0064 Img2Txt: 0.6599  Txt2Img: 0.6826\n",
            "test Loss: 0.0129 Img2Txt: 0.6599  Txt2Img: 0.6826\n",
            "\n",
            "Epoch 99/500\n",
            "--------------------\n",
            "train Loss: 0.0063 Img2Txt: 0.6325  Txt2Img: 0.6806\n",
            "test Loss: 0.0130 Img2Txt: 0.6325  Txt2Img: 0.6806\n",
            "\n",
            "Epoch 100/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6486  Txt2Img: 0.6777\n",
            "test Loss: 0.0130 Img2Txt: 0.6486  Txt2Img: 0.6777\n",
            "\n",
            "Epoch 101/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6552  Txt2Img: 0.6777\n",
            "test Loss: 0.0130 Img2Txt: 0.6552  Txt2Img: 0.6777\n",
            "\n",
            "Epoch 102/500\n",
            "--------------------\n",
            "train Loss: 0.0064 Img2Txt: 0.6406  Txt2Img: 0.6820\n",
            "test Loss: 0.0130 Img2Txt: 0.6406  Txt2Img: 0.6820\n",
            "\n",
            "Epoch 103/500\n",
            "--------------------\n",
            "train Loss: 0.0070 Img2Txt: 0.6560  Txt2Img: 0.6905\n",
            "test Loss: 0.0129 Img2Txt: 0.6560  Txt2Img: 0.6905\n",
            "\n",
            "Epoch 104/500\n",
            "--------------------\n",
            "train Loss: 0.0062 Img2Txt: 0.6267  Txt2Img: 0.6771\n",
            "test Loss: 0.0131 Img2Txt: 0.6267  Txt2Img: 0.6771\n",
            "\n",
            "Epoch 105/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6493  Txt2Img: 0.6763\n",
            "test Loss: 0.0130 Img2Txt: 0.6493  Txt2Img: 0.6763\n",
            "\n",
            "Epoch 106/500\n",
            "--------------------\n",
            "train Loss: 0.0062 Img2Txt: 0.6414  Txt2Img: 0.6774\n",
            "test Loss: 0.0129 Img2Txt: 0.6414  Txt2Img: 0.6774\n",
            "\n",
            "Epoch 107/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6470  Txt2Img: 0.6826\n",
            "test Loss: 0.0129 Img2Txt: 0.6470  Txt2Img: 0.6826\n",
            "\n",
            "Epoch 108/500\n",
            "--------------------\n",
            "train Loss: 0.0062 Img2Txt: 0.6258  Txt2Img: 0.6738\n",
            "test Loss: 0.0131 Img2Txt: 0.6258  Txt2Img: 0.6738\n",
            "\n",
            "Epoch 109/500\n",
            "--------------------\n",
            "train Loss: 0.0070 Img2Txt: 0.6532  Txt2Img: 0.6807\n",
            "test Loss: 0.0129 Img2Txt: 0.6532  Txt2Img: 0.6807\n",
            "\n",
            "Epoch 110/500\n",
            "--------------------\n",
            "train Loss: 0.0060 Img2Txt: 0.6341  Txt2Img: 0.6663\n",
            "test Loss: 0.0129 Img2Txt: 0.6341  Txt2Img: 0.6663\n",
            "\n",
            "Epoch 111/500\n",
            "--------------------\n",
            "train Loss: 0.0066 Img2Txt: 0.6565  Txt2Img: 0.6825\n",
            "test Loss: 0.0129 Img2Txt: 0.6565  Txt2Img: 0.6825\n",
            "\n",
            "Epoch 112/500\n",
            "--------------------\n",
            "train Loss: 0.0061 Img2Txt: 0.6296  Txt2Img: 0.6787\n",
            "test Loss: 0.0130 Img2Txt: 0.6296  Txt2Img: 0.6787\n",
            "\n",
            "Epoch 113/500\n",
            "--------------------\n",
            "train Loss: 0.0067 Img2Txt: 0.6499  Txt2Img: 0.6755\n",
            "test Loss: 0.0129 Img2Txt: 0.6499  Txt2Img: 0.6755\n",
            "\n",
            "Epoch 114/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6382  Txt2Img: 0.6700\n",
            "test Loss: 0.0129 Img2Txt: 0.6382  Txt2Img: 0.6700\n",
            "\n",
            "Epoch 115/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6521  Txt2Img: 0.6857\n",
            "test Loss: 0.0129 Img2Txt: 0.6521  Txt2Img: 0.6857\n",
            "\n",
            "Epoch 116/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6379  Txt2Img: 0.6855\n",
            "test Loss: 0.0129 Img2Txt: 0.6379  Txt2Img: 0.6855\n",
            "\n",
            "Epoch 117/500\n",
            "--------------------\n",
            "train Loss: 0.0065 Img2Txt: 0.6503  Txt2Img: 0.6824\n",
            "test Loss: 0.0128 Img2Txt: 0.6503  Txt2Img: 0.6824\n",
            "\n",
            "Epoch 118/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6513  Txt2Img: 0.6784\n",
            "test Loss: 0.0128 Img2Txt: 0.6513  Txt2Img: 0.6784\n",
            "\n",
            "Epoch 119/500\n",
            "--------------------\n",
            "train Loss: 0.0063 Img2Txt: 0.6477  Txt2Img: 0.6886\n",
            "test Loss: 0.0129 Img2Txt: 0.6477  Txt2Img: 0.6886\n",
            "\n",
            "Epoch 120/500\n",
            "--------------------\n",
            "train Loss: 0.0061 Img2Txt: 0.6418  Txt2Img: 0.6799\n",
            "test Loss: 0.0128 Img2Txt: 0.6418  Txt2Img: 0.6799\n",
            "\n",
            "Epoch 121/500\n",
            "--------------------\n",
            "train Loss: 0.0062 Img2Txt: 0.6530  Txt2Img: 0.6781\n",
            "test Loss: 0.0129 Img2Txt: 0.6530  Txt2Img: 0.6781\n",
            "\n",
            "Epoch 122/500\n",
            "--------------------\n",
            "train Loss: 0.0060 Img2Txt: 0.6597  Txt2Img: 0.6817\n",
            "test Loss: 0.0128 Img2Txt: 0.6597  Txt2Img: 0.6817\n",
            "\n",
            "Epoch 123/500\n",
            "--------------------\n",
            "train Loss: 0.0061 Img2Txt: 0.6449  Txt2Img: 0.6836\n",
            "test Loss: 0.0129 Img2Txt: 0.6449  Txt2Img: 0.6836\n",
            "\n",
            "Epoch 124/500\n",
            "--------------------\n",
            "train Loss: 0.0061 Img2Txt: 0.6523  Txt2Img: 0.6805\n",
            "test Loss: 0.0127 Img2Txt: 0.6523  Txt2Img: 0.6805\n",
            "\n",
            "Epoch 125/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6528  Txt2Img: 0.6839\n",
            "test Loss: 0.0128 Img2Txt: 0.6528  Txt2Img: 0.6839\n",
            "\n",
            "Epoch 126/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6619  Txt2Img: 0.6886\n",
            "test Loss: 0.0127 Img2Txt: 0.6619  Txt2Img: 0.6886\n",
            "\n",
            "Epoch 127/500\n",
            "--------------------\n",
            "train Loss: 0.0060 Img2Txt: 0.6441  Txt2Img: 0.6792\n",
            "test Loss: 0.0128 Img2Txt: 0.6441  Txt2Img: 0.6792\n",
            "\n",
            "Epoch 128/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6594  Txt2Img: 0.6848\n",
            "test Loss: 0.0127 Img2Txt: 0.6594  Txt2Img: 0.6848\n",
            "\n",
            "Epoch 129/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6448  Txt2Img: 0.6834\n",
            "test Loss: 0.0128 Img2Txt: 0.6448  Txt2Img: 0.6834\n",
            "\n",
            "Epoch 130/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6636  Txt2Img: 0.6928\n",
            "test Loss: 0.0127 Img2Txt: 0.6636  Txt2Img: 0.6928\n",
            "\n",
            "Epoch 131/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6383  Txt2Img: 0.6800\n",
            "test Loss: 0.0126 Img2Txt: 0.6383  Txt2Img: 0.6800\n",
            "\n",
            "Epoch 132/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6563  Txt2Img: 0.6808\n",
            "test Loss: 0.0128 Img2Txt: 0.6563  Txt2Img: 0.6808\n",
            "\n",
            "Epoch 133/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6504  Txt2Img: 0.6864\n",
            "test Loss: 0.0127 Img2Txt: 0.6504  Txt2Img: 0.6864\n",
            "\n",
            "Epoch 134/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6617  Txt2Img: 0.6790\n",
            "test Loss: 0.0126 Img2Txt: 0.6617  Txt2Img: 0.6790\n",
            "\n",
            "Epoch 135/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6376  Txt2Img: 0.6800\n",
            "test Loss: 0.0128 Img2Txt: 0.6376  Txt2Img: 0.6800\n",
            "\n",
            "Epoch 136/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6509  Txt2Img: 0.6823\n",
            "test Loss: 0.0127 Img2Txt: 0.6509  Txt2Img: 0.6823\n",
            "\n",
            "Epoch 137/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6255  Txt2Img: 0.6783\n",
            "test Loss: 0.0128 Img2Txt: 0.6255  Txt2Img: 0.6783\n",
            "\n",
            "Epoch 138/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6432  Txt2Img: 0.6820\n",
            "test Loss: 0.0127 Img2Txt: 0.6432  Txt2Img: 0.6820\n",
            "\n",
            "Epoch 139/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6504  Txt2Img: 0.6855\n",
            "test Loss: 0.0127 Img2Txt: 0.6504  Txt2Img: 0.6855\n",
            "\n",
            "Epoch 140/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6481  Txt2Img: 0.6795\n",
            "test Loss: 0.0129 Img2Txt: 0.6481  Txt2Img: 0.6795\n",
            "\n",
            "Epoch 141/500\n",
            "--------------------\n",
            "train Loss: 0.0059 Img2Txt: 0.6251  Txt2Img: 0.6758\n",
            "test Loss: 0.0126 Img2Txt: 0.6251  Txt2Img: 0.6758\n",
            "\n",
            "Epoch 142/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6460  Txt2Img: 0.6773\n",
            "test Loss: 0.0129 Img2Txt: 0.6460  Txt2Img: 0.6773\n",
            "\n",
            "Epoch 143/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6480  Txt2Img: 0.6800\n",
            "test Loss: 0.0127 Img2Txt: 0.6480  Txt2Img: 0.6800\n",
            "\n",
            "Epoch 144/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6438  Txt2Img: 0.6775\n",
            "test Loss: 0.0128 Img2Txt: 0.6438  Txt2Img: 0.6775\n",
            "\n",
            "Epoch 145/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6333  Txt2Img: 0.6737\n",
            "test Loss: 0.0127 Img2Txt: 0.6333  Txt2Img: 0.6737\n",
            "\n",
            "Epoch 146/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6453  Txt2Img: 0.6827\n",
            "test Loss: 0.0128 Img2Txt: 0.6453  Txt2Img: 0.6827\n",
            "\n",
            "Epoch 147/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6487  Txt2Img: 0.6817\n",
            "test Loss: 0.0127 Img2Txt: 0.6487  Txt2Img: 0.6817\n",
            "\n",
            "Epoch 148/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6349  Txt2Img: 0.6697\n",
            "test Loss: 0.0129 Img2Txt: 0.6349  Txt2Img: 0.6697\n",
            "\n",
            "Epoch 149/500\n",
            "--------------------\n",
            "train Loss: 0.0058 Img2Txt: 0.6337  Txt2Img: 0.6765\n",
            "test Loss: 0.0127 Img2Txt: 0.6337  Txt2Img: 0.6765\n",
            "\n",
            "Epoch 150/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6420  Txt2Img: 0.6779\n",
            "test Loss: 0.0127 Img2Txt: 0.6420  Txt2Img: 0.6779\n",
            "\n",
            "Epoch 151/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6438  Txt2Img: 0.6825\n",
            "test Loss: 0.0126 Img2Txt: 0.6438  Txt2Img: 0.6825\n",
            "\n",
            "Epoch 152/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6403  Txt2Img: 0.6719\n",
            "test Loss: 0.0129 Img2Txt: 0.6403  Txt2Img: 0.6719\n",
            "\n",
            "Epoch 153/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6473  Txt2Img: 0.6764\n",
            "test Loss: 0.0125 Img2Txt: 0.6473  Txt2Img: 0.6764\n",
            "\n",
            "Epoch 154/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6391  Txt2Img: 0.6803\n",
            "test Loss: 0.0127 Img2Txt: 0.6391  Txt2Img: 0.6803\n",
            "\n",
            "Epoch 155/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6347  Txt2Img: 0.6732\n",
            "test Loss: 0.0126 Img2Txt: 0.6347  Txt2Img: 0.6732\n",
            "\n",
            "Epoch 156/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6486  Txt2Img: 0.6770\n",
            "test Loss: 0.0128 Img2Txt: 0.6486  Txt2Img: 0.6770\n",
            "\n",
            "Epoch 157/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6432  Txt2Img: 0.6776\n",
            "test Loss: 0.0126 Img2Txt: 0.6432  Txt2Img: 0.6776\n",
            "\n",
            "Epoch 158/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6320  Txt2Img: 0.6763\n",
            "test Loss: 0.0128 Img2Txt: 0.6320  Txt2Img: 0.6763\n",
            "\n",
            "Epoch 159/500\n",
            "--------------------\n",
            "train Loss: 0.0057 Img2Txt: 0.6370  Txt2Img: 0.6742\n",
            "test Loss: 0.0126 Img2Txt: 0.6370  Txt2Img: 0.6742\n",
            "\n",
            "Epoch 160/500\n",
            "--------------------\n",
            "train Loss: 0.0055 Img2Txt: 0.6513  Txt2Img: 0.6769\n",
            "test Loss: 0.0127 Img2Txt: 0.6513  Txt2Img: 0.6769\n",
            "\n",
            "Epoch 161/500\n",
            "--------------------\n",
            "train Loss: 0.0055 Img2Txt: 0.6446  Txt2Img: 0.6802\n",
            "test Loss: 0.0126 Img2Txt: 0.6446  Txt2Img: 0.6802\n",
            "\n",
            "Epoch 162/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6341  Txt2Img: 0.6724\n",
            "test Loss: 0.0128 Img2Txt: 0.6341  Txt2Img: 0.6724\n",
            "\n",
            "Epoch 163/500\n",
            "--------------------\n",
            "train Loss: 0.0055 Img2Txt: 0.6368  Txt2Img: 0.6722\n",
            "test Loss: 0.0126 Img2Txt: 0.6368  Txt2Img: 0.6722\n",
            "\n",
            "Epoch 164/500\n",
            "--------------------\n",
            "train Loss: 0.0055 Img2Txt: 0.6513  Txt2Img: 0.6754\n",
            "test Loss: 0.0126 Img2Txt: 0.6513  Txt2Img: 0.6754\n",
            "\n",
            "Epoch 165/500\n",
            "--------------------\n",
            "train Loss: 0.0053 Img2Txt: 0.6457  Txt2Img: 0.6789\n",
            "test Loss: 0.0126 Img2Txt: 0.6457  Txt2Img: 0.6789\n",
            "\n",
            "Epoch 166/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6430  Txt2Img: 0.6745\n",
            "test Loss: 0.0127 Img2Txt: 0.6430  Txt2Img: 0.6745\n",
            "\n",
            "Epoch 167/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6286  Txt2Img: 0.6695\n",
            "test Loss: 0.0126 Img2Txt: 0.6286  Txt2Img: 0.6695\n",
            "\n",
            "Epoch 168/500\n",
            "--------------------\n",
            "train Loss: 0.0056 Img2Txt: 0.6467  Txt2Img: 0.6752\n",
            "test Loss: 0.0127 Img2Txt: 0.6467  Txt2Img: 0.6752\n",
            "\n",
            "Epoch 169/500\n",
            "--------------------\n",
            "train Loss: 0.0053 Img2Txt: 0.6425  Txt2Img: 0.6749\n",
            "test Loss: 0.0125 Img2Txt: 0.6425  Txt2Img: 0.6749\n",
            "\n",
            "Epoch 170/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6504  Txt2Img: 0.6750\n",
            "test Loss: 0.0126 Img2Txt: 0.6504  Txt2Img: 0.6750\n",
            "\n",
            "Epoch 171/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6291  Txt2Img: 0.6721\n",
            "test Loss: 0.0126 Img2Txt: 0.6291  Txt2Img: 0.6721\n",
            "\n",
            "Epoch 172/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6482  Txt2Img: 0.6751\n",
            "test Loss: 0.0126 Img2Txt: 0.6482  Txt2Img: 0.6751\n",
            "\n",
            "Epoch 173/500\n",
            "--------------------\n",
            "train Loss: 0.0053 Img2Txt: 0.6377  Txt2Img: 0.6781\n",
            "test Loss: 0.0125 Img2Txt: 0.6377  Txt2Img: 0.6781\n",
            "\n",
            "Epoch 174/500\n",
            "--------------------\n",
            "train Loss: 0.0054 Img2Txt: 0.6515  Txt2Img: 0.6733\n",
            "test Loss: 0.0126 Img2Txt: 0.6515  Txt2Img: 0.6733\n",
            "\n",
            "Epoch 175/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6464  Txt2Img: 0.6840\n",
            "test Loss: 0.0125 Img2Txt: 0.6464  Txt2Img: 0.6840\n",
            "\n",
            "Epoch 176/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6491  Txt2Img: 0.6773\n",
            "test Loss: 0.0126 Img2Txt: 0.6491  Txt2Img: 0.6773\n",
            "\n",
            "Epoch 177/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6340  Txt2Img: 0.6752\n",
            "test Loss: 0.0125 Img2Txt: 0.6340  Txt2Img: 0.6752\n",
            "\n",
            "Epoch 178/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6429  Txt2Img: 0.6774\n",
            "test Loss: 0.0126 Img2Txt: 0.6429  Txt2Img: 0.6774\n",
            "\n",
            "Epoch 179/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6477  Txt2Img: 0.6768\n",
            "test Loss: 0.0125 Img2Txt: 0.6477  Txt2Img: 0.6768\n",
            "\n",
            "Epoch 180/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6438  Txt2Img: 0.6757\n",
            "test Loss: 0.0126 Img2Txt: 0.6438  Txt2Img: 0.6757\n",
            "\n",
            "Epoch 181/500\n",
            "--------------------\n",
            "train Loss: 0.0053 Img2Txt: 0.6380  Txt2Img: 0.6749\n",
            "test Loss: 0.0125 Img2Txt: 0.6380  Txt2Img: 0.6749\n",
            "\n",
            "Epoch 182/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6407  Txt2Img: 0.6789\n",
            "test Loss: 0.0126 Img2Txt: 0.6407  Txt2Img: 0.6789\n",
            "\n",
            "Epoch 183/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6388  Txt2Img: 0.6760\n",
            "test Loss: 0.0124 Img2Txt: 0.6388  Txt2Img: 0.6760\n",
            "\n",
            "Epoch 184/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6440  Txt2Img: 0.6738\n",
            "test Loss: 0.0126 Img2Txt: 0.6440  Txt2Img: 0.6738\n",
            "\n",
            "Epoch 185/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6437  Txt2Img: 0.6784\n",
            "test Loss: 0.0125 Img2Txt: 0.6437  Txt2Img: 0.6784\n",
            "\n",
            "Epoch 186/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6506  Txt2Img: 0.6749\n",
            "test Loss: 0.0127 Img2Txt: 0.6506  Txt2Img: 0.6749\n",
            "\n",
            "Epoch 187/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6282  Txt2Img: 0.6743\n",
            "test Loss: 0.0124 Img2Txt: 0.6282  Txt2Img: 0.6743\n",
            "\n",
            "Epoch 188/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6430  Txt2Img: 0.6773\n",
            "test Loss: 0.0126 Img2Txt: 0.6430  Txt2Img: 0.6773\n",
            "\n",
            "Epoch 189/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6401  Txt2Img: 0.6718\n",
            "test Loss: 0.0126 Img2Txt: 0.6401  Txt2Img: 0.6718\n",
            "\n",
            "Epoch 190/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6472  Txt2Img: 0.6755\n",
            "test Loss: 0.0127 Img2Txt: 0.6472  Txt2Img: 0.6755\n",
            "\n",
            "Epoch 191/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6385  Txt2Img: 0.6776\n",
            "test Loss: 0.0124 Img2Txt: 0.6385  Txt2Img: 0.6776\n",
            "\n",
            "Epoch 192/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6474  Txt2Img: 0.6792\n",
            "test Loss: 0.0125 Img2Txt: 0.6474  Txt2Img: 0.6792\n",
            "\n",
            "Epoch 193/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6309  Txt2Img: 0.6732\n",
            "test Loss: 0.0126 Img2Txt: 0.6309  Txt2Img: 0.6732\n",
            "\n",
            "Epoch 194/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6339  Txt2Img: 0.6749\n",
            "test Loss: 0.0126 Img2Txt: 0.6339  Txt2Img: 0.6749\n",
            "\n",
            "Epoch 195/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6631  Txt2Img: 0.6851\n",
            "test Loss: 0.0125 Img2Txt: 0.6631  Txt2Img: 0.6851\n",
            "\n",
            "Epoch 196/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6325  Txt2Img: 0.6756\n",
            "test Loss: 0.0126 Img2Txt: 0.6325  Txt2Img: 0.6756\n",
            "\n",
            "Epoch 197/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6592  Txt2Img: 0.6796\n",
            "test Loss: 0.0125 Img2Txt: 0.6592  Txt2Img: 0.6796\n",
            "\n",
            "Epoch 198/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6323  Txt2Img: 0.6738\n",
            "test Loss: 0.0125 Img2Txt: 0.6323  Txt2Img: 0.6738\n",
            "\n",
            "Epoch 199/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6582  Txt2Img: 0.6865\n",
            "test Loss: 0.0125 Img2Txt: 0.6582  Txt2Img: 0.6865\n",
            "\n",
            "Epoch 200/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6301  Txt2Img: 0.6668\n",
            "test Loss: 0.0126 Img2Txt: 0.6301  Txt2Img: 0.6668\n",
            "\n",
            "Epoch 201/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6595  Txt2Img: 0.6811\n",
            "test Loss: 0.0125 Img2Txt: 0.6595  Txt2Img: 0.6811\n",
            "\n",
            "Epoch 202/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6423  Txt2Img: 0.6753\n",
            "test Loss: 0.0125 Img2Txt: 0.6423  Txt2Img: 0.6753\n",
            "\n",
            "Epoch 203/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6560  Txt2Img: 0.6847\n",
            "test Loss: 0.0126 Img2Txt: 0.6560  Txt2Img: 0.6847\n",
            "\n",
            "Epoch 204/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6420  Txt2Img: 0.6727\n",
            "test Loss: 0.0125 Img2Txt: 0.6420  Txt2Img: 0.6727\n",
            "\n",
            "Epoch 205/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6592  Txt2Img: 0.6817\n",
            "test Loss: 0.0126 Img2Txt: 0.6592  Txt2Img: 0.6817\n",
            "\n",
            "Epoch 206/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6439  Txt2Img: 0.6756\n",
            "test Loss: 0.0124 Img2Txt: 0.6439  Txt2Img: 0.6756\n",
            "\n",
            "Epoch 207/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6514  Txt2Img: 0.6863\n",
            "test Loss: 0.0126 Img2Txt: 0.6514  Txt2Img: 0.6863\n",
            "\n",
            "Epoch 208/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6393  Txt2Img: 0.6718\n",
            "test Loss: 0.0124 Img2Txt: 0.6393  Txt2Img: 0.6718\n",
            "\n",
            "Epoch 209/500\n",
            "--------------------\n",
            "train Loss: 0.0052 Img2Txt: 0.6523  Txt2Img: 0.6798\n",
            "test Loss: 0.0126 Img2Txt: 0.6523  Txt2Img: 0.6798\n",
            "\n",
            "Epoch 210/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6439  Txt2Img: 0.6749\n",
            "test Loss: 0.0124 Img2Txt: 0.6439  Txt2Img: 0.6749\n",
            "\n",
            "Epoch 211/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6592  Txt2Img: 0.6790\n",
            "test Loss: 0.0127 Img2Txt: 0.6592  Txt2Img: 0.6790\n",
            "\n",
            "Epoch 212/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6441  Txt2Img: 0.6721\n",
            "test Loss: 0.0124 Img2Txt: 0.6441  Txt2Img: 0.6721\n",
            "\n",
            "Epoch 213/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6593  Txt2Img: 0.6904\n",
            "test Loss: 0.0124 Img2Txt: 0.6593  Txt2Img: 0.6904\n",
            "\n",
            "Epoch 214/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6414  Txt2Img: 0.6701\n",
            "test Loss: 0.0124 Img2Txt: 0.6414  Txt2Img: 0.6701\n",
            "\n",
            "Epoch 215/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6600  Txt2Img: 0.6854\n",
            "test Loss: 0.0125 Img2Txt: 0.6600  Txt2Img: 0.6854\n",
            "\n",
            "Epoch 216/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6403  Txt2Img: 0.6687\n",
            "test Loss: 0.0126 Img2Txt: 0.6403  Txt2Img: 0.6687\n",
            "\n",
            "Epoch 217/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6609  Txt2Img: 0.6852\n",
            "test Loss: 0.0124 Img2Txt: 0.6609  Txt2Img: 0.6852\n",
            "\n",
            "Epoch 218/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6422  Txt2Img: 0.6746\n",
            "test Loss: 0.0124 Img2Txt: 0.6422  Txt2Img: 0.6746\n",
            "\n",
            "Epoch 219/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6663  Txt2Img: 0.6853\n",
            "test Loss: 0.0125 Img2Txt: 0.6663  Txt2Img: 0.6853\n",
            "\n",
            "Epoch 220/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6368  Txt2Img: 0.6699\n",
            "test Loss: 0.0124 Img2Txt: 0.6368  Txt2Img: 0.6699\n",
            "\n",
            "Epoch 221/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6570  Txt2Img: 0.6818\n",
            "test Loss: 0.0125 Img2Txt: 0.6570  Txt2Img: 0.6818\n",
            "\n",
            "Epoch 222/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6393  Txt2Img: 0.6768\n",
            "test Loss: 0.0125 Img2Txt: 0.6393  Txt2Img: 0.6768\n",
            "\n",
            "Epoch 223/500\n",
            "--------------------\n",
            "train Loss: 0.0051 Img2Txt: 0.6595  Txt2Img: 0.6823\n",
            "test Loss: 0.0125 Img2Txt: 0.6595  Txt2Img: 0.6823\n",
            "\n",
            "Epoch 224/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6377  Txt2Img: 0.6755\n",
            "test Loss: 0.0124 Img2Txt: 0.6377  Txt2Img: 0.6755\n",
            "\n",
            "Epoch 225/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6627  Txt2Img: 0.6865\n",
            "test Loss: 0.0124 Img2Txt: 0.6627  Txt2Img: 0.6865\n",
            "\n",
            "Epoch 226/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6384  Txt2Img: 0.6741\n",
            "test Loss: 0.0125 Img2Txt: 0.6384  Txt2Img: 0.6741\n",
            "\n",
            "Epoch 227/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6553  Txt2Img: 0.6833\n",
            "test Loss: 0.0125 Img2Txt: 0.6553  Txt2Img: 0.6833\n",
            "\n",
            "Epoch 228/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6419  Txt2Img: 0.6735\n",
            "test Loss: 0.0124 Img2Txt: 0.6419  Txt2Img: 0.6735\n",
            "\n",
            "Epoch 229/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6601  Txt2Img: 0.6864\n",
            "test Loss: 0.0124 Img2Txt: 0.6601  Txt2Img: 0.6864\n",
            "\n",
            "Epoch 230/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6404  Txt2Img: 0.6717\n",
            "test Loss: 0.0125 Img2Txt: 0.6404  Txt2Img: 0.6717\n",
            "\n",
            "Epoch 231/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6530  Txt2Img: 0.6831\n",
            "test Loss: 0.0125 Img2Txt: 0.6530  Txt2Img: 0.6831\n",
            "\n",
            "Epoch 232/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6375  Txt2Img: 0.6757\n",
            "test Loss: 0.0125 Img2Txt: 0.6375  Txt2Img: 0.6757\n",
            "\n",
            "Epoch 233/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6586  Txt2Img: 0.6789\n",
            "test Loss: 0.0125 Img2Txt: 0.6586  Txt2Img: 0.6789\n",
            "\n",
            "Epoch 234/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6393  Txt2Img: 0.6804\n",
            "test Loss: 0.0124 Img2Txt: 0.6393  Txt2Img: 0.6804\n",
            "\n",
            "Epoch 235/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6530  Txt2Img: 0.6824\n",
            "test Loss: 0.0124 Img2Txt: 0.6530  Txt2Img: 0.6824\n",
            "\n",
            "Epoch 236/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6361  Txt2Img: 0.6755\n",
            "test Loss: 0.0124 Img2Txt: 0.6361  Txt2Img: 0.6755\n",
            "\n",
            "Epoch 237/500\n",
            "--------------------\n",
            "train Loss: 0.0049 Img2Txt: 0.6543  Txt2Img: 0.6754\n",
            "test Loss: 0.0126 Img2Txt: 0.6543  Txt2Img: 0.6754\n",
            "\n",
            "Epoch 238/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6363  Txt2Img: 0.6759\n",
            "test Loss: 0.0125 Img2Txt: 0.6363  Txt2Img: 0.6759\n",
            "\n",
            "Epoch 239/500\n",
            "--------------------\n",
            "train Loss: 0.0050 Img2Txt: 0.6616  Txt2Img: 0.6868\n",
            "test Loss: 0.0124 Img2Txt: 0.6616  Txt2Img: 0.6868\n",
            "\n",
            "Epoch 240/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6371  Txt2Img: 0.6760\n",
            "test Loss: 0.0124 Img2Txt: 0.6371  Txt2Img: 0.6760\n",
            "\n",
            "Epoch 241/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6573  Txt2Img: 0.6777\n",
            "test Loss: 0.0125 Img2Txt: 0.6573  Txt2Img: 0.6777\n",
            "\n",
            "Epoch 242/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6395  Txt2Img: 0.6692\n",
            "test Loss: 0.0125 Img2Txt: 0.6395  Txt2Img: 0.6692\n",
            "\n",
            "Epoch 243/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6625  Txt2Img: 0.6888\n",
            "test Loss: 0.0123 Img2Txt: 0.6625  Txt2Img: 0.6888\n",
            "\n",
            "Epoch 244/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6397  Txt2Img: 0.6801\n",
            "test Loss: 0.0124 Img2Txt: 0.6397  Txt2Img: 0.6801\n",
            "\n",
            "Epoch 245/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6631  Txt2Img: 0.6776\n",
            "test Loss: 0.0124 Img2Txt: 0.6631  Txt2Img: 0.6776\n",
            "\n",
            "Epoch 246/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6377  Txt2Img: 0.6703\n",
            "test Loss: 0.0124 Img2Txt: 0.6377  Txt2Img: 0.6703\n",
            "\n",
            "Epoch 247/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6576  Txt2Img: 0.6817\n",
            "test Loss: 0.0124 Img2Txt: 0.6576  Txt2Img: 0.6817\n",
            "\n",
            "Epoch 248/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6409  Txt2Img: 0.6814\n",
            "test Loss: 0.0124 Img2Txt: 0.6409  Txt2Img: 0.6814\n",
            "\n",
            "Epoch 249/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6651  Txt2Img: 0.6792\n",
            "test Loss: 0.0124 Img2Txt: 0.6651  Txt2Img: 0.6792\n",
            "\n",
            "Epoch 250/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6346  Txt2Img: 0.6700\n",
            "test Loss: 0.0124 Img2Txt: 0.6346  Txt2Img: 0.6700\n",
            "\n",
            "Epoch 251/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6555  Txt2Img: 0.6757\n",
            "test Loss: 0.0125 Img2Txt: 0.6555  Txt2Img: 0.6757\n",
            "\n",
            "Epoch 252/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6400  Txt2Img: 0.6737\n",
            "test Loss: 0.0124 Img2Txt: 0.6400  Txt2Img: 0.6737\n",
            "\n",
            "Epoch 253/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6623  Txt2Img: 0.6861\n",
            "test Loss: 0.0123 Img2Txt: 0.6623  Txt2Img: 0.6861\n",
            "\n",
            "Epoch 254/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6356  Txt2Img: 0.6715\n",
            "test Loss: 0.0124 Img2Txt: 0.6356  Txt2Img: 0.6715\n",
            "\n",
            "Epoch 255/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6602  Txt2Img: 0.6775\n",
            "test Loss: 0.0124 Img2Txt: 0.6602  Txt2Img: 0.6775\n",
            "\n",
            "Epoch 256/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6442  Txt2Img: 0.6708\n",
            "test Loss: 0.0125 Img2Txt: 0.6442  Txt2Img: 0.6708\n",
            "\n",
            "Epoch 257/500\n",
            "--------------------\n",
            "train Loss: 0.0047 Img2Txt: 0.6604  Txt2Img: 0.6840\n",
            "test Loss: 0.0123 Img2Txt: 0.6604  Txt2Img: 0.6840\n",
            "\n",
            "Epoch 258/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6370  Txt2Img: 0.6695\n",
            "test Loss: 0.0124 Img2Txt: 0.6370  Txt2Img: 0.6695\n",
            "\n",
            "Epoch 259/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6652  Txt2Img: 0.6786\n",
            "test Loss: 0.0124 Img2Txt: 0.6652  Txt2Img: 0.6786\n",
            "\n",
            "Epoch 260/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6378  Txt2Img: 0.6765\n",
            "test Loss: 0.0124 Img2Txt: 0.6378  Txt2Img: 0.6765\n",
            "\n",
            "Epoch 261/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6568  Txt2Img: 0.6739\n",
            "test Loss: 0.0123 Img2Txt: 0.6568  Txt2Img: 0.6739\n",
            "\n",
            "Epoch 262/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6418  Txt2Img: 0.6759\n",
            "test Loss: 0.0124 Img2Txt: 0.6418  Txt2Img: 0.6759\n",
            "\n",
            "Epoch 263/500\n",
            "--------------------\n",
            "train Loss: 0.0048 Img2Txt: 0.6627  Txt2Img: 0.6763\n",
            "test Loss: 0.0124 Img2Txt: 0.6627  Txt2Img: 0.6763\n",
            "\n",
            "Epoch 264/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6436  Txt2Img: 0.6733\n",
            "test Loss: 0.0123 Img2Txt: 0.6436  Txt2Img: 0.6733\n",
            "\n",
            "Epoch 265/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6610  Txt2Img: 0.6803\n",
            "test Loss: 0.0123 Img2Txt: 0.6610  Txt2Img: 0.6803\n",
            "\n",
            "Epoch 266/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6424  Txt2Img: 0.6724\n",
            "test Loss: 0.0124 Img2Txt: 0.6424  Txt2Img: 0.6724\n",
            "\n",
            "Epoch 267/500\n",
            "--------------------\n",
            "train Loss: 0.0046 Img2Txt: 0.6675  Txt2Img: 0.6827\n",
            "test Loss: 0.0123 Img2Txt: 0.6675  Txt2Img: 0.6827\n",
            "\n",
            "Epoch 268/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6414  Txt2Img: 0.6795\n",
            "test Loss: 0.0123 Img2Txt: 0.6414  Txt2Img: 0.6795\n",
            "\n",
            "Epoch 269/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6621  Txt2Img: 0.6773\n",
            "test Loss: 0.0123 Img2Txt: 0.6621  Txt2Img: 0.6773\n",
            "\n",
            "Epoch 270/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6439  Txt2Img: 0.6695\n",
            "test Loss: 0.0124 Img2Txt: 0.6439  Txt2Img: 0.6695\n",
            "\n",
            "Epoch 271/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6636  Txt2Img: 0.6768\n",
            "test Loss: 0.0123 Img2Txt: 0.6636  Txt2Img: 0.6768\n",
            "\n",
            "Epoch 272/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6402  Txt2Img: 0.6756\n",
            "test Loss: 0.0123 Img2Txt: 0.6402  Txt2Img: 0.6756\n",
            "\n",
            "Epoch 273/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6652  Txt2Img: 0.6766\n",
            "test Loss: 0.0124 Img2Txt: 0.6652  Txt2Img: 0.6766\n",
            "\n",
            "Epoch 274/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6407  Txt2Img: 0.6762\n",
            "test Loss: 0.0123 Img2Txt: 0.6407  Txt2Img: 0.6762\n",
            "\n",
            "Epoch 275/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6605  Txt2Img: 0.6779\n",
            "test Loss: 0.0124 Img2Txt: 0.6605  Txt2Img: 0.6779\n",
            "\n",
            "Epoch 276/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6408  Txt2Img: 0.6713\n",
            "test Loss: 0.0123 Img2Txt: 0.6408  Txt2Img: 0.6713\n",
            "\n",
            "Epoch 277/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6622  Txt2Img: 0.6744\n",
            "test Loss: 0.0123 Img2Txt: 0.6622  Txt2Img: 0.6744\n",
            "\n",
            "Epoch 278/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6434  Txt2Img: 0.6780\n",
            "test Loss: 0.0123 Img2Txt: 0.6434  Txt2Img: 0.6780\n",
            "\n",
            "Epoch 279/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6600  Txt2Img: 0.6806\n",
            "test Loss: 0.0123 Img2Txt: 0.6600  Txt2Img: 0.6806\n",
            "\n",
            "Epoch 280/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6436  Txt2Img: 0.6691\n",
            "test Loss: 0.0124 Img2Txt: 0.6436  Txt2Img: 0.6691\n",
            "\n",
            "Epoch 281/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6588  Txt2Img: 0.6788\n",
            "test Loss: 0.0123 Img2Txt: 0.6588  Txt2Img: 0.6788\n",
            "\n",
            "Epoch 282/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6429  Txt2Img: 0.6746\n",
            "test Loss: 0.0123 Img2Txt: 0.6429  Txt2Img: 0.6746\n",
            "\n",
            "Epoch 283/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6536  Txt2Img: 0.6748\n",
            "test Loss: 0.0123 Img2Txt: 0.6536  Txt2Img: 0.6748\n",
            "\n",
            "Epoch 284/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6430  Txt2Img: 0.6739\n",
            "test Loss: 0.0124 Img2Txt: 0.6430  Txt2Img: 0.6739\n",
            "\n",
            "Epoch 285/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6606  Txt2Img: 0.6756\n",
            "test Loss: 0.0122 Img2Txt: 0.6606  Txt2Img: 0.6756\n",
            "\n",
            "Epoch 286/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6362  Txt2Img: 0.6763\n",
            "test Loss: 0.0123 Img2Txt: 0.6362  Txt2Img: 0.6763\n",
            "\n",
            "Epoch 287/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6592  Txt2Img: 0.6746\n",
            "test Loss: 0.0124 Img2Txt: 0.6592  Txt2Img: 0.6746\n",
            "\n",
            "Epoch 288/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6399  Txt2Img: 0.6677\n",
            "test Loss: 0.0123 Img2Txt: 0.6399  Txt2Img: 0.6677\n",
            "\n",
            "Epoch 289/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6534  Txt2Img: 0.6722\n",
            "test Loss: 0.0123 Img2Txt: 0.6534  Txt2Img: 0.6722\n",
            "\n",
            "Epoch 290/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6405  Txt2Img: 0.6792\n",
            "test Loss: 0.0124 Img2Txt: 0.6405  Txt2Img: 0.6792\n",
            "\n",
            "Epoch 291/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6712  Txt2Img: 0.6785\n",
            "test Loss: 0.0123 Img2Txt: 0.6712  Txt2Img: 0.6785\n",
            "\n",
            "Epoch 292/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6372  Txt2Img: 0.6736\n",
            "test Loss: 0.0123 Img2Txt: 0.6372  Txt2Img: 0.6736\n",
            "\n",
            "Epoch 293/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6572  Txt2Img: 0.6781\n",
            "test Loss: 0.0123 Img2Txt: 0.6572  Txt2Img: 0.6781\n",
            "\n",
            "Epoch 294/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6390  Txt2Img: 0.6717\n",
            "test Loss: 0.0124 Img2Txt: 0.6390  Txt2Img: 0.6717\n",
            "\n",
            "Epoch 295/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6683  Txt2Img: 0.6819\n",
            "test Loss: 0.0123 Img2Txt: 0.6683  Txt2Img: 0.6819\n",
            "\n",
            "Epoch 296/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6376  Txt2Img: 0.6736\n",
            "test Loss: 0.0124 Img2Txt: 0.6376  Txt2Img: 0.6736\n",
            "\n",
            "Epoch 297/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6657  Txt2Img: 0.6795\n",
            "test Loss: 0.0123 Img2Txt: 0.6657  Txt2Img: 0.6795\n",
            "\n",
            "Epoch 298/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6356  Txt2Img: 0.6691\n",
            "test Loss: 0.0125 Img2Txt: 0.6356  Txt2Img: 0.6691\n",
            "\n",
            "Epoch 299/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6626  Txt2Img: 0.6766\n",
            "test Loss: 0.0124 Img2Txt: 0.6626  Txt2Img: 0.6766\n",
            "\n",
            "Epoch 300/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6391  Txt2Img: 0.6754\n",
            "test Loss: 0.0124 Img2Txt: 0.6391  Txt2Img: 0.6754\n",
            "\n",
            "Epoch 301/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6639  Txt2Img: 0.6741\n",
            "test Loss: 0.0123 Img2Txt: 0.6639  Txt2Img: 0.6741\n",
            "\n",
            "Epoch 302/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6302  Txt2Img: 0.6700\n",
            "test Loss: 0.0125 Img2Txt: 0.6302  Txt2Img: 0.6700\n",
            "\n",
            "Epoch 303/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6640  Txt2Img: 0.6801\n",
            "test Loss: 0.0123 Img2Txt: 0.6640  Txt2Img: 0.6801\n",
            "\n",
            "Epoch 304/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6409  Txt2Img: 0.6729\n",
            "test Loss: 0.0125 Img2Txt: 0.6409  Txt2Img: 0.6729\n",
            "\n",
            "Epoch 305/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6548  Txt2Img: 0.6716\n",
            "test Loss: 0.0123 Img2Txt: 0.6548  Txt2Img: 0.6716\n",
            "\n",
            "Epoch 306/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6368  Txt2Img: 0.6719\n",
            "test Loss: 0.0125 Img2Txt: 0.6368  Txt2Img: 0.6719\n",
            "\n",
            "Epoch 307/500\n",
            "--------------------\n",
            "train Loss: 0.0045 Img2Txt: 0.6664  Txt2Img: 0.6778\n",
            "test Loss: 0.0121 Img2Txt: 0.6664  Txt2Img: 0.6778\n",
            "\n",
            "Epoch 308/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6386  Txt2Img: 0.6691\n",
            "test Loss: 0.0123 Img2Txt: 0.6386  Txt2Img: 0.6691\n",
            "\n",
            "Epoch 309/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6544  Txt2Img: 0.6778\n",
            "test Loss: 0.0121 Img2Txt: 0.6544  Txt2Img: 0.6778\n",
            "\n",
            "Epoch 310/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6389  Txt2Img: 0.6668\n",
            "test Loss: 0.0125 Img2Txt: 0.6389  Txt2Img: 0.6668\n",
            "\n",
            "Epoch 311/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6644  Txt2Img: 0.6762\n",
            "test Loss: 0.0122 Img2Txt: 0.6644  Txt2Img: 0.6762\n",
            "\n",
            "Epoch 312/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6353  Txt2Img: 0.6685\n",
            "test Loss: 0.0123 Img2Txt: 0.6353  Txt2Img: 0.6685\n",
            "\n",
            "Epoch 313/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6549  Txt2Img: 0.6743\n",
            "test Loss: 0.0123 Img2Txt: 0.6549  Txt2Img: 0.6743\n",
            "\n",
            "Epoch 314/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6324  Txt2Img: 0.6669\n",
            "test Loss: 0.0124 Img2Txt: 0.6324  Txt2Img: 0.6669\n",
            "\n",
            "Epoch 315/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6577  Txt2Img: 0.6809\n",
            "test Loss: 0.0123 Img2Txt: 0.6577  Txt2Img: 0.6809\n",
            "\n",
            "Epoch 316/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6325  Txt2Img: 0.6750\n",
            "test Loss: 0.0124 Img2Txt: 0.6325  Txt2Img: 0.6750\n",
            "\n",
            "Epoch 317/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6649  Txt2Img: 0.6751\n",
            "test Loss: 0.0123 Img2Txt: 0.6649  Txt2Img: 0.6751\n",
            "\n",
            "Epoch 318/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6311  Txt2Img: 0.6609\n",
            "test Loss: 0.0124 Img2Txt: 0.6311  Txt2Img: 0.6609\n",
            "\n",
            "Epoch 319/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6510  Txt2Img: 0.6813\n",
            "test Loss: 0.0123 Img2Txt: 0.6510  Txt2Img: 0.6813\n",
            "\n",
            "Epoch 320/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6310  Txt2Img: 0.6739\n",
            "test Loss: 0.0124 Img2Txt: 0.6310  Txt2Img: 0.6739\n",
            "\n",
            "Epoch 321/500\n",
            "--------------------\n",
            "train Loss: 0.0044 Img2Txt: 0.6567  Txt2Img: 0.6674\n",
            "test Loss: 0.0124 Img2Txt: 0.6567  Txt2Img: 0.6674\n",
            "\n",
            "Epoch 322/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6335  Txt2Img: 0.6659\n",
            "test Loss: 0.0124 Img2Txt: 0.6335  Txt2Img: 0.6659\n",
            "\n",
            "Epoch 323/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6607  Txt2Img: 0.6836\n",
            "test Loss: 0.0122 Img2Txt: 0.6607  Txt2Img: 0.6836\n",
            "\n",
            "Epoch 324/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6304  Txt2Img: 0.6623\n",
            "test Loss: 0.0125 Img2Txt: 0.6304  Txt2Img: 0.6623\n",
            "\n",
            "Epoch 325/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6531  Txt2Img: 0.6711\n",
            "test Loss: 0.0124 Img2Txt: 0.6531  Txt2Img: 0.6711\n",
            "\n",
            "Epoch 326/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6343  Txt2Img: 0.6721\n",
            "test Loss: 0.0124 Img2Txt: 0.6343  Txt2Img: 0.6721\n",
            "\n",
            "Epoch 327/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6586  Txt2Img: 0.6752\n",
            "test Loss: 0.0123 Img2Txt: 0.6586  Txt2Img: 0.6752\n",
            "\n",
            "Epoch 328/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6356  Txt2Img: 0.6642\n",
            "test Loss: 0.0124 Img2Txt: 0.6356  Txt2Img: 0.6642\n",
            "\n",
            "Epoch 329/500\n",
            "--------------------\n",
            "train Loss: 0.0043 Img2Txt: 0.6603  Txt2Img: 0.6745\n",
            "test Loss: 0.0122 Img2Txt: 0.6603  Txt2Img: 0.6745\n",
            "\n",
            "Epoch 330/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6288  Txt2Img: 0.6712\n",
            "test Loss: 0.0123 Img2Txt: 0.6288  Txt2Img: 0.6712\n",
            "\n",
            "Epoch 331/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6548  Txt2Img: 0.6694\n",
            "test Loss: 0.0123 Img2Txt: 0.6548  Txt2Img: 0.6694\n",
            "\n",
            "Epoch 332/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6315  Txt2Img: 0.6602\n",
            "test Loss: 0.0124 Img2Txt: 0.6315  Txt2Img: 0.6602\n",
            "\n",
            "Epoch 333/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6602  Txt2Img: 0.6772\n",
            "test Loss: 0.0123 Img2Txt: 0.6602  Txt2Img: 0.6772\n",
            "\n",
            "Epoch 334/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6300  Txt2Img: 0.6738\n",
            "test Loss: 0.0123 Img2Txt: 0.6300  Txt2Img: 0.6738\n",
            "\n",
            "Epoch 335/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6517  Txt2Img: 0.6665\n",
            "test Loss: 0.0124 Img2Txt: 0.6517  Txt2Img: 0.6665\n",
            "\n",
            "Epoch 336/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6304  Txt2Img: 0.6648\n",
            "test Loss: 0.0123 Img2Txt: 0.6304  Txt2Img: 0.6648\n",
            "\n",
            "Epoch 337/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6596  Txt2Img: 0.6741\n",
            "test Loss: 0.0123 Img2Txt: 0.6596  Txt2Img: 0.6741\n",
            "\n",
            "Epoch 338/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6254  Txt2Img: 0.6675\n",
            "test Loss: 0.0122 Img2Txt: 0.6254  Txt2Img: 0.6675\n",
            "\n",
            "Epoch 339/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6509  Txt2Img: 0.6677\n",
            "test Loss: 0.0124 Img2Txt: 0.6509  Txt2Img: 0.6677\n",
            "\n",
            "Epoch 340/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6337  Txt2Img: 0.6671\n",
            "test Loss: 0.0123 Img2Txt: 0.6337  Txt2Img: 0.6671\n",
            "\n",
            "Epoch 341/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6583  Txt2Img: 0.6713\n",
            "test Loss: 0.0123 Img2Txt: 0.6583  Txt2Img: 0.6713\n",
            "\n",
            "Epoch 342/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6296  Txt2Img: 0.6663\n",
            "test Loss: 0.0123 Img2Txt: 0.6296  Txt2Img: 0.6663\n",
            "\n",
            "Epoch 343/500\n",
            "--------------------\n",
            "train Loss: 0.0042 Img2Txt: 0.6603  Txt2Img: 0.6743\n",
            "test Loss: 0.0122 Img2Txt: 0.6603  Txt2Img: 0.6743\n",
            "\n",
            "Epoch 344/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6311  Txt2Img: 0.6647\n",
            "test Loss: 0.0124 Img2Txt: 0.6311  Txt2Img: 0.6647\n",
            "\n",
            "Epoch 345/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6538  Txt2Img: 0.6728\n",
            "test Loss: 0.0123 Img2Txt: 0.6538  Txt2Img: 0.6728\n",
            "\n",
            "Epoch 346/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6384  Txt2Img: 0.6692\n",
            "test Loss: 0.0123 Img2Txt: 0.6384  Txt2Img: 0.6692\n",
            "\n",
            "Epoch 347/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6612  Txt2Img: 0.6681\n",
            "test Loss: 0.0123 Img2Txt: 0.6612  Txt2Img: 0.6681\n",
            "\n",
            "Epoch 348/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6313  Txt2Img: 0.6682\n",
            "test Loss: 0.0124 Img2Txt: 0.6313  Txt2Img: 0.6682\n",
            "\n",
            "Epoch 349/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6505  Txt2Img: 0.6660\n",
            "test Loss: 0.0123 Img2Txt: 0.6505  Txt2Img: 0.6660\n",
            "\n",
            "Epoch 350/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6341  Txt2Img: 0.6720\n",
            "test Loss: 0.0123 Img2Txt: 0.6341  Txt2Img: 0.6720\n",
            "\n",
            "Epoch 351/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6574  Txt2Img: 0.6664\n",
            "test Loss: 0.0123 Img2Txt: 0.6574  Txt2Img: 0.6664\n",
            "\n",
            "Epoch 352/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6408  Txt2Img: 0.6648\n",
            "test Loss: 0.0124 Img2Txt: 0.6408  Txt2Img: 0.6648\n",
            "\n",
            "Epoch 353/500\n",
            "--------------------\n",
            "train Loss: 0.0040 Img2Txt: 0.6515  Txt2Img: 0.6751\n",
            "test Loss: 0.0122 Img2Txt: 0.6515  Txt2Img: 0.6751\n",
            "\n",
            "Epoch 354/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6401  Txt2Img: 0.6725\n",
            "test Loss: 0.0123 Img2Txt: 0.6401  Txt2Img: 0.6725\n",
            "\n",
            "Epoch 355/500\n",
            "--------------------\n",
            "train Loss: 0.0041 Img2Txt: 0.6504  Txt2Img: 0.6658\n",
            "test Loss: 0.0123 Img2Txt: 0.6504  Txt2Img: 0.6658\n",
            "\n",
            "Epoch 356/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6434  Txt2Img: 0.6788\n",
            "test Loss: 0.0122 Img2Txt: 0.6434  Txt2Img: 0.6788\n",
            "\n",
            "Epoch 357/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6552  Txt2Img: 0.6734\n",
            "test Loss: 0.0122 Img2Txt: 0.6552  Txt2Img: 0.6734\n",
            "\n",
            "Epoch 358/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6308  Txt2Img: 0.6697\n",
            "test Loss: 0.0123 Img2Txt: 0.6308  Txt2Img: 0.6697\n",
            "\n",
            "Epoch 359/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6489  Txt2Img: 0.6662\n",
            "test Loss: 0.0124 Img2Txt: 0.6489  Txt2Img: 0.6662\n",
            "\n",
            "Epoch 360/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6354  Txt2Img: 0.6652\n",
            "test Loss: 0.0123 Img2Txt: 0.6354  Txt2Img: 0.6652\n",
            "\n",
            "Epoch 361/500\n",
            "--------------------\n",
            "train Loss: 0.0039 Img2Txt: 0.6549  Txt2Img: 0.6723\n",
            "test Loss: 0.0123 Img2Txt: 0.6549  Txt2Img: 0.6723\n",
            "\n",
            "Epoch 362/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6388  Txt2Img: 0.6659\n",
            "test Loss: 0.0122 Img2Txt: 0.6388  Txt2Img: 0.6659\n",
            "\n",
            "Epoch 363/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6570  Txt2Img: 0.6671\n",
            "test Loss: 0.0123 Img2Txt: 0.6570  Txt2Img: 0.6671\n",
            "\n",
            "Epoch 364/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6285  Txt2Img: 0.6610\n",
            "test Loss: 0.0124 Img2Txt: 0.6285  Txt2Img: 0.6610\n",
            "\n",
            "Epoch 365/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6448  Txt2Img: 0.6681\n",
            "test Loss: 0.0123 Img2Txt: 0.6448  Txt2Img: 0.6681\n",
            "\n",
            "Epoch 366/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6395  Txt2Img: 0.6656\n",
            "test Loss: 0.0123 Img2Txt: 0.6395  Txt2Img: 0.6656\n",
            "\n",
            "Epoch 367/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6575  Txt2Img: 0.6635\n",
            "test Loss: 0.0123 Img2Txt: 0.6575  Txt2Img: 0.6635\n",
            "\n",
            "Epoch 368/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6318  Txt2Img: 0.6639\n",
            "test Loss: 0.0123 Img2Txt: 0.6318  Txt2Img: 0.6639\n",
            "\n",
            "Epoch 369/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6506  Txt2Img: 0.6712\n",
            "test Loss: 0.0123 Img2Txt: 0.6506  Txt2Img: 0.6712\n",
            "\n",
            "Epoch 370/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6426  Txt2Img: 0.6605\n",
            "test Loss: 0.0123 Img2Txt: 0.6426  Txt2Img: 0.6605\n",
            "\n",
            "Epoch 371/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6537  Txt2Img: 0.6661\n",
            "test Loss: 0.0123 Img2Txt: 0.6537  Txt2Img: 0.6661\n",
            "\n",
            "Epoch 372/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6358  Txt2Img: 0.6699\n",
            "test Loss: 0.0122 Img2Txt: 0.6358  Txt2Img: 0.6699\n",
            "\n",
            "Epoch 373/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6472  Txt2Img: 0.6731\n",
            "test Loss: 0.0122 Img2Txt: 0.6472  Txt2Img: 0.6731\n",
            "\n",
            "Epoch 374/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6371  Txt2Img: 0.6617\n",
            "test Loss: 0.0124 Img2Txt: 0.6371  Txt2Img: 0.6617\n",
            "\n",
            "Epoch 375/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6552  Txt2Img: 0.6635\n",
            "test Loss: 0.0122 Img2Txt: 0.6552  Txt2Img: 0.6635\n",
            "\n",
            "Epoch 376/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6340  Txt2Img: 0.6667\n",
            "test Loss: 0.0122 Img2Txt: 0.6340  Txt2Img: 0.6667\n",
            "\n",
            "Epoch 377/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6492  Txt2Img: 0.6679\n",
            "test Loss: 0.0122 Img2Txt: 0.6492  Txt2Img: 0.6679\n",
            "\n",
            "Epoch 378/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6389  Txt2Img: 0.6660\n",
            "test Loss: 0.0123 Img2Txt: 0.6389  Txt2Img: 0.6660\n",
            "\n",
            "Epoch 379/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6555  Txt2Img: 0.6680\n",
            "test Loss: 0.0122 Img2Txt: 0.6555  Txt2Img: 0.6680\n",
            "\n",
            "Epoch 380/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6330  Txt2Img: 0.6665\n",
            "test Loss: 0.0122 Img2Txt: 0.6330  Txt2Img: 0.6665\n",
            "\n",
            "Epoch 381/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6466  Txt2Img: 0.6652\n",
            "test Loss: 0.0122 Img2Txt: 0.6466  Txt2Img: 0.6652\n",
            "\n",
            "Epoch 382/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6393  Txt2Img: 0.6658\n",
            "test Loss: 0.0123 Img2Txt: 0.6393  Txt2Img: 0.6658\n",
            "\n",
            "Epoch 383/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6547  Txt2Img: 0.6721\n",
            "test Loss: 0.0122 Img2Txt: 0.6547  Txt2Img: 0.6721\n",
            "\n",
            "Epoch 384/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6302  Txt2Img: 0.6586\n",
            "test Loss: 0.0124 Img2Txt: 0.6302  Txt2Img: 0.6586\n",
            "\n",
            "Epoch 385/500\n",
            "--------------------\n",
            "train Loss: 0.0038 Img2Txt: 0.6502  Txt2Img: 0.6706\n",
            "test Loss: 0.0122 Img2Txt: 0.6502  Txt2Img: 0.6706\n",
            "\n",
            "Epoch 386/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6390  Txt2Img: 0.6703\n",
            "test Loss: 0.0122 Img2Txt: 0.6390  Txt2Img: 0.6703\n",
            "\n",
            "Epoch 387/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6526  Txt2Img: 0.6720\n",
            "test Loss: 0.0122 Img2Txt: 0.6526  Txt2Img: 0.6720\n",
            "\n",
            "Epoch 388/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6395  Txt2Img: 0.6647\n",
            "test Loss: 0.0123 Img2Txt: 0.6395  Txt2Img: 0.6647\n",
            "\n",
            "Epoch 389/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6479  Txt2Img: 0.6700\n",
            "test Loss: 0.0122 Img2Txt: 0.6479  Txt2Img: 0.6700\n",
            "\n",
            "Epoch 390/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6375  Txt2Img: 0.6658\n",
            "test Loss: 0.0123 Img2Txt: 0.6375  Txt2Img: 0.6658\n",
            "\n",
            "Epoch 391/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6459  Txt2Img: 0.6694\n",
            "test Loss: 0.0123 Img2Txt: 0.6459  Txt2Img: 0.6694\n",
            "\n",
            "Epoch 392/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6399  Txt2Img: 0.6743\n",
            "test Loss: 0.0121 Img2Txt: 0.6399  Txt2Img: 0.6743\n",
            "\n",
            "Epoch 393/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6582  Txt2Img: 0.6750\n",
            "test Loss: 0.0121 Img2Txt: 0.6582  Txt2Img: 0.6750\n",
            "\n",
            "Epoch 394/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6393  Txt2Img: 0.6621\n",
            "test Loss: 0.0123 Img2Txt: 0.6393  Txt2Img: 0.6621\n",
            "\n",
            "Epoch 395/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6465  Txt2Img: 0.6710\n",
            "test Loss: 0.0123 Img2Txt: 0.6465  Txt2Img: 0.6710\n",
            "\n",
            "Epoch 396/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6357  Txt2Img: 0.6710\n",
            "test Loss: 0.0122 Img2Txt: 0.6357  Txt2Img: 0.6710\n",
            "\n",
            "Epoch 397/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6540  Txt2Img: 0.6732\n",
            "test Loss: 0.0122 Img2Txt: 0.6540  Txt2Img: 0.6732\n",
            "\n",
            "Epoch 398/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6383  Txt2Img: 0.6590\n",
            "test Loss: 0.0124 Img2Txt: 0.6383  Txt2Img: 0.6590\n",
            "\n",
            "Epoch 399/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6493  Txt2Img: 0.6758\n",
            "test Loss: 0.0122 Img2Txt: 0.6493  Txt2Img: 0.6758\n",
            "\n",
            "Epoch 400/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6380  Txt2Img: 0.6716\n",
            "test Loss: 0.0122 Img2Txt: 0.6380  Txt2Img: 0.6716\n",
            "\n",
            "Epoch 401/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6463  Txt2Img: 0.6716\n",
            "test Loss: 0.0122 Img2Txt: 0.6463  Txt2Img: 0.6716\n",
            "\n",
            "Epoch 402/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6444  Txt2Img: 0.6707\n",
            "test Loss: 0.0123 Img2Txt: 0.6444  Txt2Img: 0.6707\n",
            "\n",
            "Epoch 403/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6568  Txt2Img: 0.6708\n",
            "test Loss: 0.0122 Img2Txt: 0.6568  Txt2Img: 0.6708\n",
            "\n",
            "Epoch 404/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6328  Txt2Img: 0.6650\n",
            "test Loss: 0.0123 Img2Txt: 0.6328  Txt2Img: 0.6650\n",
            "\n",
            "Epoch 405/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6460  Txt2Img: 0.6745\n",
            "test Loss: 0.0123 Img2Txt: 0.6460  Txt2Img: 0.6745\n",
            "\n",
            "Epoch 406/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6407  Txt2Img: 0.6713\n",
            "test Loss: 0.0123 Img2Txt: 0.6407  Txt2Img: 0.6713\n",
            "\n",
            "Epoch 407/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6460  Txt2Img: 0.6723\n",
            "test Loss: 0.0122 Img2Txt: 0.6460  Txt2Img: 0.6723\n",
            "\n",
            "Epoch 408/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6403  Txt2Img: 0.6670\n",
            "test Loss: 0.0123 Img2Txt: 0.6403  Txt2Img: 0.6670\n",
            "\n",
            "Epoch 409/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6492  Txt2Img: 0.6700\n",
            "test Loss: 0.0123 Img2Txt: 0.6492  Txt2Img: 0.6700\n",
            "\n",
            "Epoch 410/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6395  Txt2Img: 0.6666\n",
            "test Loss: 0.0123 Img2Txt: 0.6395  Txt2Img: 0.6666\n",
            "\n",
            "Epoch 411/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6365  Txt2Img: 0.6670\n",
            "test Loss: 0.0123 Img2Txt: 0.6365  Txt2Img: 0.6670\n",
            "\n",
            "Epoch 412/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6440  Txt2Img: 0.6631\n",
            "test Loss: 0.0123 Img2Txt: 0.6440  Txt2Img: 0.6631\n",
            "\n",
            "Epoch 413/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6495  Txt2Img: 0.6734\n",
            "test Loss: 0.0123 Img2Txt: 0.6495  Txt2Img: 0.6734\n",
            "\n",
            "Epoch 414/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6384  Txt2Img: 0.6661\n",
            "test Loss: 0.0121 Img2Txt: 0.6384  Txt2Img: 0.6661\n",
            "\n",
            "Epoch 415/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6444  Txt2Img: 0.6695\n",
            "test Loss: 0.0122 Img2Txt: 0.6444  Txt2Img: 0.6695\n",
            "\n",
            "Epoch 416/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6408  Txt2Img: 0.6601\n",
            "test Loss: 0.0124 Img2Txt: 0.6408  Txt2Img: 0.6601\n",
            "\n",
            "Epoch 417/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6429  Txt2Img: 0.6702\n",
            "test Loss: 0.0123 Img2Txt: 0.6429  Txt2Img: 0.6702\n",
            "\n",
            "Epoch 418/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6372  Txt2Img: 0.6689\n",
            "test Loss: 0.0122 Img2Txt: 0.6372  Txt2Img: 0.6689\n",
            "\n",
            "Epoch 419/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6522  Txt2Img: 0.6711\n",
            "test Loss: 0.0122 Img2Txt: 0.6522  Txt2Img: 0.6711\n",
            "\n",
            "Epoch 420/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6365  Txt2Img: 0.6619\n",
            "test Loss: 0.0123 Img2Txt: 0.6365  Txt2Img: 0.6619\n",
            "\n",
            "Epoch 421/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6487  Txt2Img: 0.6675\n",
            "test Loss: 0.0124 Img2Txt: 0.6487  Txt2Img: 0.6675\n",
            "\n",
            "Epoch 422/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6354  Txt2Img: 0.6715\n",
            "test Loss: 0.0122 Img2Txt: 0.6354  Txt2Img: 0.6715\n",
            "\n",
            "Epoch 423/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6507  Txt2Img: 0.6651\n",
            "test Loss: 0.0123 Img2Txt: 0.6507  Txt2Img: 0.6651\n",
            "\n",
            "Epoch 424/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6422  Txt2Img: 0.6625\n",
            "test Loss: 0.0122 Img2Txt: 0.6422  Txt2Img: 0.6625\n",
            "\n",
            "Epoch 425/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6521  Txt2Img: 0.6732\n",
            "test Loss: 0.0123 Img2Txt: 0.6521  Txt2Img: 0.6732\n",
            "\n",
            "Epoch 426/500\n",
            "--------------------\n",
            "train Loss: 0.0037 Img2Txt: 0.6425  Txt2Img: 0.6550\n",
            "test Loss: 0.0124 Img2Txt: 0.6425  Txt2Img: 0.6550\n",
            "\n",
            "Epoch 427/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6413  Txt2Img: 0.6683\n",
            "test Loss: 0.0122 Img2Txt: 0.6413  Txt2Img: 0.6683\n",
            "\n",
            "Epoch 428/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6426  Txt2Img: 0.6604\n",
            "test Loss: 0.0122 Img2Txt: 0.6426  Txt2Img: 0.6604\n",
            "\n",
            "Epoch 429/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6492  Txt2Img: 0.6769\n",
            "test Loss: 0.0122 Img2Txt: 0.6492  Txt2Img: 0.6769\n",
            "\n",
            "Epoch 430/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6383  Txt2Img: 0.6595\n",
            "test Loss: 0.0123 Img2Txt: 0.6383  Txt2Img: 0.6595\n",
            "\n",
            "Epoch 431/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6456  Txt2Img: 0.6746\n",
            "test Loss: 0.0122 Img2Txt: 0.6456  Txt2Img: 0.6746\n",
            "\n",
            "Epoch 432/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6394  Txt2Img: 0.6621\n",
            "test Loss: 0.0122 Img2Txt: 0.6394  Txt2Img: 0.6621\n",
            "\n",
            "Epoch 433/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6535  Txt2Img: 0.6725\n",
            "test Loss: 0.0123 Img2Txt: 0.6535  Txt2Img: 0.6725\n",
            "\n",
            "Epoch 434/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6347  Txt2Img: 0.6641\n",
            "test Loss: 0.0122 Img2Txt: 0.6347  Txt2Img: 0.6641\n",
            "\n",
            "Epoch 435/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6568  Txt2Img: 0.6718\n",
            "test Loss: 0.0122 Img2Txt: 0.6568  Txt2Img: 0.6718\n",
            "\n",
            "Epoch 436/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6437  Txt2Img: 0.6646\n",
            "test Loss: 0.0122 Img2Txt: 0.6437  Txt2Img: 0.6646\n",
            "\n",
            "Epoch 437/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6459  Txt2Img: 0.6690\n",
            "test Loss: 0.0123 Img2Txt: 0.6459  Txt2Img: 0.6690\n",
            "\n",
            "Epoch 438/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6374  Txt2Img: 0.6593\n",
            "test Loss: 0.0122 Img2Txt: 0.6374  Txt2Img: 0.6593\n",
            "\n",
            "Epoch 439/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6536  Txt2Img: 0.6739\n",
            "test Loss: 0.0122 Img2Txt: 0.6536  Txt2Img: 0.6739\n",
            "\n",
            "Epoch 440/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6392  Txt2Img: 0.6568\n",
            "test Loss: 0.0122 Img2Txt: 0.6392  Txt2Img: 0.6568\n",
            "\n",
            "Epoch 441/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6414  Txt2Img: 0.6754\n",
            "test Loss: 0.0122 Img2Txt: 0.6414  Txt2Img: 0.6754\n",
            "\n",
            "Epoch 442/500\n",
            "--------------------\n",
            "train Loss: 0.0036 Img2Txt: 0.6404  Txt2Img: 0.6544\n",
            "test Loss: 0.0123 Img2Txt: 0.6404  Txt2Img: 0.6544\n",
            "\n",
            "Epoch 443/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6543  Txt2Img: 0.6707\n",
            "test Loss: 0.0124 Img2Txt: 0.6543  Txt2Img: 0.6707\n",
            "\n",
            "Epoch 444/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6395  Txt2Img: 0.6600\n",
            "test Loss: 0.0121 Img2Txt: 0.6395  Txt2Img: 0.6600\n",
            "\n",
            "Epoch 445/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6560  Txt2Img: 0.6747\n",
            "test Loss: 0.0122 Img2Txt: 0.6560  Txt2Img: 0.6747\n",
            "\n",
            "Epoch 446/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6430  Txt2Img: 0.6613\n",
            "test Loss: 0.0122 Img2Txt: 0.6430  Txt2Img: 0.6613\n",
            "\n",
            "Epoch 447/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6536  Txt2Img: 0.6695\n",
            "test Loss: 0.0123 Img2Txt: 0.6536  Txt2Img: 0.6695\n",
            "\n",
            "Epoch 448/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6383  Txt2Img: 0.6640\n",
            "test Loss: 0.0121 Img2Txt: 0.6383  Txt2Img: 0.6640\n",
            "\n",
            "Epoch 449/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6552  Txt2Img: 0.6649\n",
            "test Loss: 0.0123 Img2Txt: 0.6552  Txt2Img: 0.6649\n",
            "\n",
            "Epoch 450/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6367  Txt2Img: 0.6614\n",
            "test Loss: 0.0122 Img2Txt: 0.6367  Txt2Img: 0.6614\n",
            "\n",
            "Epoch 451/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6561  Txt2Img: 0.6760\n",
            "test Loss: 0.0122 Img2Txt: 0.6561  Txt2Img: 0.6760\n",
            "\n",
            "Epoch 452/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6326  Txt2Img: 0.6624\n",
            "test Loss: 0.0121 Img2Txt: 0.6326  Txt2Img: 0.6624\n",
            "\n",
            "Epoch 453/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6504  Txt2Img: 0.6642\n",
            "test Loss: 0.0123 Img2Txt: 0.6504  Txt2Img: 0.6642\n",
            "\n",
            "Epoch 454/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6386  Txt2Img: 0.6557\n",
            "test Loss: 0.0122 Img2Txt: 0.6386  Txt2Img: 0.6557\n",
            "\n",
            "Epoch 455/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6552  Txt2Img: 0.6725\n",
            "test Loss: 0.0122 Img2Txt: 0.6552  Txt2Img: 0.6725\n",
            "\n",
            "Epoch 456/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6301  Txt2Img: 0.6641\n",
            "test Loss: 0.0121 Img2Txt: 0.6301  Txt2Img: 0.6641\n",
            "\n",
            "Epoch 457/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6546  Txt2Img: 0.6692\n",
            "test Loss: 0.0123 Img2Txt: 0.6546  Txt2Img: 0.6692\n",
            "\n",
            "Epoch 458/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6394  Txt2Img: 0.6618\n",
            "test Loss: 0.0122 Img2Txt: 0.6394  Txt2Img: 0.6618\n",
            "\n",
            "Epoch 459/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6502  Txt2Img: 0.6702\n",
            "test Loss: 0.0122 Img2Txt: 0.6502  Txt2Img: 0.6702\n",
            "\n",
            "Epoch 460/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6350  Txt2Img: 0.6621\n",
            "test Loss: 0.0121 Img2Txt: 0.6350  Txt2Img: 0.6621\n",
            "\n",
            "Epoch 461/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6527  Txt2Img: 0.6620\n",
            "test Loss: 0.0123 Img2Txt: 0.6527  Txt2Img: 0.6620\n",
            "\n",
            "Epoch 462/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6343  Txt2Img: 0.6577\n",
            "test Loss: 0.0121 Img2Txt: 0.6343  Txt2Img: 0.6577\n",
            "\n",
            "Epoch 463/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6516  Txt2Img: 0.6734\n",
            "test Loss: 0.0122 Img2Txt: 0.6516  Txt2Img: 0.6734\n",
            "\n",
            "Epoch 464/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6318  Txt2Img: 0.6601\n",
            "test Loss: 0.0121 Img2Txt: 0.6318  Txt2Img: 0.6601\n",
            "\n",
            "Epoch 465/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6487  Txt2Img: 0.6607\n",
            "test Loss: 0.0124 Img2Txt: 0.6487  Txt2Img: 0.6607\n",
            "\n",
            "Epoch 466/500\n",
            "--------------------\n",
            "train Loss: 0.0035 Img2Txt: 0.6315  Txt2Img: 0.6594\n",
            "test Loss: 0.0121 Img2Txt: 0.6315  Txt2Img: 0.6594\n",
            "\n",
            "Epoch 467/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6542  Txt2Img: 0.6738\n",
            "test Loss: 0.0122 Img2Txt: 0.6542  Txt2Img: 0.6738\n",
            "\n",
            "Epoch 468/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6400  Txt2Img: 0.6591\n",
            "test Loss: 0.0122 Img2Txt: 0.6400  Txt2Img: 0.6591\n",
            "\n",
            "Epoch 469/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6408  Txt2Img: 0.6683\n",
            "test Loss: 0.0122 Img2Txt: 0.6408  Txt2Img: 0.6683\n",
            "\n",
            "Epoch 470/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6429  Txt2Img: 0.6637\n",
            "test Loss: 0.0121 Img2Txt: 0.6429  Txt2Img: 0.6637\n",
            "\n",
            "Epoch 471/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6431  Txt2Img: 0.6698\n",
            "test Loss: 0.0121 Img2Txt: 0.6431  Txt2Img: 0.6698\n",
            "\n",
            "Epoch 472/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6413  Txt2Img: 0.6582\n",
            "test Loss: 0.0122 Img2Txt: 0.6413  Txt2Img: 0.6582\n",
            "\n",
            "Epoch 473/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6472  Txt2Img: 0.6711\n",
            "test Loss: 0.0122 Img2Txt: 0.6472  Txt2Img: 0.6711\n",
            "\n",
            "Epoch 474/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6363  Txt2Img: 0.6663\n",
            "test Loss: 0.0122 Img2Txt: 0.6363  Txt2Img: 0.6663\n",
            "\n",
            "Epoch 475/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6448  Txt2Img: 0.6686\n",
            "test Loss: 0.0121 Img2Txt: 0.6448  Txt2Img: 0.6686\n",
            "\n",
            "Epoch 476/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6396  Txt2Img: 0.6597\n",
            "test Loss: 0.0122 Img2Txt: 0.6396  Txt2Img: 0.6597\n",
            "\n",
            "Epoch 477/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6505  Txt2Img: 0.6697\n",
            "test Loss: 0.0121 Img2Txt: 0.6505  Txt2Img: 0.6697\n",
            "\n",
            "Epoch 478/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6354  Txt2Img: 0.6572\n",
            "test Loss: 0.0122 Img2Txt: 0.6354  Txt2Img: 0.6572\n",
            "\n",
            "Epoch 479/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6384  Txt2Img: 0.6648\n",
            "test Loss: 0.0121 Img2Txt: 0.6384  Txt2Img: 0.6648\n",
            "\n",
            "Epoch 480/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6460  Txt2Img: 0.6614\n",
            "test Loss: 0.0123 Img2Txt: 0.6460  Txt2Img: 0.6614\n",
            "\n",
            "Epoch 481/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6425  Txt2Img: 0.6723\n",
            "test Loss: 0.0121 Img2Txt: 0.6425  Txt2Img: 0.6723\n",
            "\n",
            "Epoch 482/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6396  Txt2Img: 0.6648\n",
            "test Loss: 0.0121 Img2Txt: 0.6396  Txt2Img: 0.6648\n",
            "\n",
            "Epoch 483/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6399  Txt2Img: 0.6669\n",
            "test Loss: 0.0121 Img2Txt: 0.6399  Txt2Img: 0.6669\n",
            "\n",
            "Epoch 484/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6427  Txt2Img: 0.6658\n",
            "test Loss: 0.0123 Img2Txt: 0.6427  Txt2Img: 0.6658\n",
            "\n",
            "Epoch 485/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6456  Txt2Img: 0.6723\n",
            "test Loss: 0.0121 Img2Txt: 0.6456  Txt2Img: 0.6723\n",
            "\n",
            "Epoch 486/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6321  Txt2Img: 0.6628\n",
            "test Loss: 0.0122 Img2Txt: 0.6321  Txt2Img: 0.6628\n",
            "\n",
            "Epoch 487/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6462  Txt2Img: 0.6646\n",
            "test Loss: 0.0122 Img2Txt: 0.6462  Txt2Img: 0.6646\n",
            "\n",
            "Epoch 488/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6388  Txt2Img: 0.6588\n",
            "test Loss: 0.0123 Img2Txt: 0.6388  Txt2Img: 0.6588\n",
            "\n",
            "Epoch 489/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6413  Txt2Img: 0.6682\n",
            "test Loss: 0.0121 Img2Txt: 0.6413  Txt2Img: 0.6682\n",
            "\n",
            "Epoch 490/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6399  Txt2Img: 0.6676\n",
            "test Loss: 0.0122 Img2Txt: 0.6399  Txt2Img: 0.6676\n",
            "\n",
            "Epoch 491/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6435  Txt2Img: 0.6665\n",
            "test Loss: 0.0122 Img2Txt: 0.6435  Txt2Img: 0.6665\n",
            "\n",
            "Epoch 492/500\n",
            "--------------------\n",
            "train Loss: 0.0032 Img2Txt: 0.6399  Txt2Img: 0.6698\n",
            "test Loss: 0.0122 Img2Txt: 0.6399  Txt2Img: 0.6698\n",
            "\n",
            "Epoch 493/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6434  Txt2Img: 0.6663\n",
            "test Loss: 0.0121 Img2Txt: 0.6434  Txt2Img: 0.6663\n",
            "\n",
            "Epoch 494/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6415  Txt2Img: 0.6697\n",
            "test Loss: 0.0122 Img2Txt: 0.6415  Txt2Img: 0.6697\n",
            "\n",
            "Epoch 495/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6387  Txt2Img: 0.6643\n",
            "test Loss: 0.0123 Img2Txt: 0.6387  Txt2Img: 0.6643\n",
            "\n",
            "Epoch 496/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6394  Txt2Img: 0.6691\n",
            "test Loss: 0.0122 Img2Txt: 0.6394  Txt2Img: 0.6691\n",
            "\n",
            "Epoch 497/500\n",
            "--------------------\n",
            "train Loss: 0.0033 Img2Txt: 0.6466  Txt2Img: 0.6697\n",
            "test Loss: 0.0120 Img2Txt: 0.6466  Txt2Img: 0.6697\n",
            "\n",
            "Epoch 498/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6385  Txt2Img: 0.6582\n",
            "test Loss: 0.0123 Img2Txt: 0.6385  Txt2Img: 0.6582\n",
            "\n",
            "Epoch 499/500\n",
            "--------------------\n",
            "train Loss: 0.0034 Img2Txt: 0.6421  Txt2Img: 0.6692\n",
            "test Loss: 0.0121 Img2Txt: 0.6421  Txt2Img: 0.6692\n",
            "\n",
            "Epoch 500/500\n",
            "--------------------\n",
            "train Loss: 0.0031 Img2Txt: 0.6422  Txt2Img: 0.6690\n",
            "test Loss: 0.0122 Img2Txt: 0.6422  Txt2Img: 0.6690\n",
            "\n",
            "Training complete in 1m 38s\n",
            "Best average ACC: 0.702351\n",
            "...Training is completed...\n",
            "...Evaluation on testing data...\n",
            "...Image to Text MAP = 0.6947012621304236\n",
            "...Text to Image MAP = 0.7100000824457635\n",
            "...Average MAP = 0.7023506722880936\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}