prophetnet_seq2seqtrainer_finetuning_experiment.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "prophetnet_seq2seqtrainer_finetuning_experiment.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyOJOMTAti+M0BmRVcB3PTRZ",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/forest1988/colaboratory/blob/main/prophetnet_seq2seqtrainer_finetuning_experiment.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KUopihAwK5MV"
      },
      "source": [
        "# ProphetNet fine-tuning with Seq2Seq Trainer\n",
        "\n",
        "- Fine-tuning experiment using ProphetNet in HuggingFace Transformer with Seq2Seq Trainer\n",
        "\n",
        "- Fine-tuning for summarization task\n",
        "  - dataset: xsum or cnn-dm"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "V1mtBgB2nBwy",
        "outputId": "e8685a55-b63d-4785-cb20-23576a6cea6f"
      },
      "source": [
        "!nvidia-smi"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Tue Nov 17 14:56:12 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 455.38       Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   63C    P8    11W /  70W |      0MiB / 15079MiB |      0%      Default |\n",
            "|                               |                      |                 ERR! |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ond9VTc5Zt8d"
      },
      "source": [
        "import torch"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "RQpqDmBWaESu",
        "outputId": "0fca7dda-e096-435a-8f41-ba9ae3bebfc4"
      },
      "source": [
        "torch.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'1.7.0+cu101'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xzxYczZ0aF97",
        "outputId": "bfc8592b-1a6f-4699-cd2e-c114c28d9371"
      },
      "source": [
        "!git clone https://github.com/huggingface/transformers.git"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "fatal: destination path 'transformers' already exists and is not an empty directory.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9khHo2A9b1p5",
        "outputId": "d5fa36bf-8740-4f43-de4f-d006e9d261c4"
      },
      "source": [
        "%cd transformers/"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/transformers\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2QGowDHBYuyA"
      },
      "source": [
        "# !git log | grep \"prepare_seq2seq_batch function to ProphetNet\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "O8aOcq_2RJDV",
        "outputId": "02dd7d84-2d78-45a6-fa76-d3122605ade0"
      },
      "source": [
        "!pip install -e ."
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Obtaining file:///content/transformers\n",
            "  Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n",
            "  Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n",
            "    Preparing wheel metadata ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (3.0.12)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (20.4)\n",
            "Requirement already satisfied: dataclasses; python_version < \"3.7\" in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (0.7)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (1.18.5)\n",
            "Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (2019.12.20)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (2.23.0)\n",
            "Requirement already satisfied: tqdm>=4.27 in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (4.41.1)\n",
            "Requirement already satisfied: tokenizers==0.9.4 in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (0.9.4)\n",
            "Requirement already satisfied: sacremoses in /usr/local/lib/python3.6/dist-packages (from transformers==4.0.0.dev0) (0.0.43)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from packaging->transformers==4.0.0.dev0) (1.15.0)\n",
            "Requirement already satisfied: pyparsing>=2.0.2 in /usr/local/lib/python3.6/dist-packages (from packaging->transformers==4.0.0.dev0) (2.4.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->transformers==4.0.0.dev0) (2020.6.20)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->transformers==4.0.0.dev0) (2.10)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->transformers==4.0.0.dev0) (3.0.4)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->transformers==4.0.0.dev0) (1.24.3)\n",
            "Requirement already satisfied: joblib in /usr/local/lib/python3.6/dist-packages (from sacremoses->transformers==4.0.0.dev0) (0.17.0)\n",
            "Requirement already satisfied: click in /usr/local/lib/python3.6/dist-packages (from sacremoses->transformers==4.0.0.dev0) (7.1.2)\n",
            "Installing collected packages: transformers\n",
            "  Found existing installation: transformers 4.0.0.dev0\n",
            "    Can't uninstall 'transformers'. No files were found to uninstall.\n",
            "  Running setup.py develop for transformers\n",
            "Successfully installed transformers\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7wpt5LV-MMlG",
        "outputId": "3f675fb3-1b89-49b2-e917-12eacb92c93b"
      },
      "source": [
        "%cd /content/transformers/examples/seq2seq/"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/transformers/examples/seq2seq\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YKqwk3OnZZfr",
        "outputId": "88306cbc-ec24-4b87-9b46-7584134a18d9"
      },
      "source": [
        "# to avoid the \"ModuleNotFoundError\" \n",
        "import sys\n",
        "print(sys.path)\n",
        "sys.path.append('/content/transformers/src')\n",
        "print(sys.path)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['', '/env/python', '/usr/lib/python36.zip', '/usr/lib/python3.6', '/usr/lib/python3.6/lib-dynload', '/usr/local/lib/python3.6/dist-packages', '/content/transformers/src', '/usr/lib/python3/dist-packages', '/usr/local/lib/python3.6/dist-packages/IPython/extensions', '/root/.ipython']\n",
            "['', '/env/python', '/usr/lib/python36.zip', '/usr/lib/python3.6', '/usr/lib/python3.6/lib-dynload', '/usr/local/lib/python3.6/dist-packages', '/content/transformers/src', '/usr/lib/python3/dist-packages', '/usr/local/lib/python3.6/dist-packages/IPython/extensions', '/root/.ipython', '/content/transformers/src']\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gmKBOl0cg7yK"
      },
      "source": [
        "# If \"ModuleNotFoundError\" occurs in this cell, please re-run the runtime. \n",
        "# (The first time Colab runtime runs, it seems the sys.path is not correctly updated when we use `pip install -e .`)\n",
        "\n",
        "import transformers"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "9iK_fsLQg_dP",
        "outputId": "5bfbd4a5-5bcb-4e02-f88d-389a5fd8c405"
      },
      "source": [
        "transformers.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'4.0.0-dev'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xwqGSr7fcMKp",
        "outputId": "232f6029-2860-45c3-8674-f880fc2467ae"
      },
      "source": [
        "!pip install cookiecutter\n",
        "!transformers-cli env"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: cookiecutter in /usr/local/lib/python3.6/dist-packages (1.7.2)\n",
            "Requirement already satisfied: requests>=2.23.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (2.23.0)\n",
            "Requirement already satisfied: python-slugify>=4.0.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (4.0.1)\n",
            "Requirement already satisfied: MarkupSafe<2.0.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (1.1.1)\n",
            "Requirement already satisfied: poyo>=0.5.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (0.5.0)\n",
            "Requirement already satisfied: Jinja2<3.0.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (2.11.2)\n",
            "Requirement already satisfied: click>=7.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (7.1.2)\n",
            "Requirement already satisfied: six>=1.10 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (1.15.0)\n",
            "Requirement already satisfied: jinja2-time>=0.2.0 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (0.2.0)\n",
            "Requirement already satisfied: binaryornot>=0.4.4 in /usr/local/lib/python3.6/dist-packages (from cookiecutter) (0.4.4)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests>=2.23.0->cookiecutter) (2020.6.20)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests>=2.23.0->cookiecutter) (2.10)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests>=2.23.0->cookiecutter) (3.0.4)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests>=2.23.0->cookiecutter) (1.24.3)\n",
            "Requirement already satisfied: text-unidecode>=1.3 in /usr/local/lib/python3.6/dist-packages (from python-slugify>=4.0.0->cookiecutter) (1.3)\n",
            "Requirement already satisfied: arrow in /usr/local/lib/python3.6/dist-packages (from jinja2-time>=0.2.0->cookiecutter) (0.17.0)\n",
            "Requirement already satisfied: python-dateutil>=2.7.0 in /usr/local/lib/python3.6/dist-packages (from arrow->jinja2-time>=0.2.0->cookiecutter) (2.8.1)\n",
            "2020-11-17 14:56:31.362609: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.1\n",
            "WARNING:tensorflow:From /content/transformers/src/transformers/commands/env.py:36: is_gpu_available (from tensorflow.python.framework.test_util) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use `tf.config.list_physical_devices('GPU')` instead.\n",
            "2020-11-17 14:56:33.235745: I tensorflow/core/platform/profile_utils/cpu_utils.cc:104] CPU Frequency: 2200000000 Hz\n",
            "2020-11-17 14:56:33.235991: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7bbafc0 initialized for platform Host (this does not guarantee that XLA will be used). Devices:\n",
            "2020-11-17 14:56:33.236027: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version\n",
            "2020-11-17 14:56:33.243232: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcuda.so.1\n",
            "2020-11-17 14:56:33.353054: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:33.353802: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7bbb180 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:\n",
            "2020-11-17 14:56:33.353836: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Tesla T4, Compute Capability 7.5\n",
            "2020-11-17 14:56:33.354024: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:33.354578: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1716] Found device 0 with properties: \n",
            "pciBusID: 0000:00:04.0 name: Tesla T4 computeCapability: 7.5\n",
            "coreClock: 1.59GHz coreCount: 40 deviceMemorySize: 14.73GiB deviceMemoryBandwidth: 298.08GiB/s\n",
            "2020-11-17 14:56:33.354622: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.1\n",
            "2020-11-17 14:56:33.541206: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcublas.so.10\n",
            "2020-11-17 14:56:33.662095: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcufft.so.10\n",
            "2020-11-17 14:56:33.711191: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcurand.so.10\n",
            "2020-11-17 14:56:33.954796: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcusolver.so.10\n",
            "2020-11-17 14:56:34.011210: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcusparse.so.10\n",
            "2020-11-17 14:56:34.470928: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudnn.so.7\n",
            "2020-11-17 14:56:34.471184: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:34.471870: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:34.472423: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1858] Adding visible gpu devices: 0\n",
            "2020-11-17 14:56:34.474987: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.1\n",
            "2020-11-17 14:56:39.334385: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1257] Device interconnect StreamExecutor with strength 1 edge matrix:\n",
            "2020-11-17 14:56:39.334454: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1263]      0 \n",
            "2020-11-17 14:56:39.334467: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1276] 0:   N \n",
            "2020-11-17 14:56:39.341039: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:39.341797: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
            "2020-11-17 14:56:39.342373: W tensorflow/core/common_runtime/gpu/gpu_bfc_allocator.cc:39] Overriding allow_growth setting because the TF_FORCE_GPU_ALLOW_GROWTH environment variable is set. Original config value was 0.\n",
            "2020-11-17 14:56:39.342423: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1402] Created TensorFlow device (/device:GPU:0 with 13176 MB memory) -> physical GPU (device: 0, name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5)\n",
            "\n",
            "Copy-and-paste the text below in your GitHub issue and FILL OUT the two last points.\n",
            "\n",
            "- `transformers` version: 4.0.0-dev\n",
            "- Platform: Linux-4.19.112+-x86_64-with-Ubuntu-18.04-bionic\n",
            "- Python version: 3.6.9\n",
            "- PyTorch version (GPU?): 1.7.0+cu101 (True)\n",
            "- Tensorflow version (GPU?): 2.3.0 (True)\n",
            "- Using GPU in script?: <fill in>\n",
            "- Using distributed or parallel set-up in script?: <fill in>\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iawDc5qpcSpj"
      },
      "source": [
        "## dataset "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MwGCB7HucRhf",
        "outputId": "f5c9965a-1f65-40e1-e2df-50e31c1aab98"
      },
      "source": [
        "!wget https://cdn-datasets.huggingface.co/summarization/xsum.tar.gz\n",
        "!tar -xzvf xsum.tar.gz"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-11-17 14:56:40--  https://cdn-datasets.huggingface.co/summarization/xsum.tar.gz\n",
            "Resolving cdn-datasets.huggingface.co (cdn-datasets.huggingface.co)... 99.84.181.104, 99.84.181.99, 99.84.181.50, ...\n",
            "Connecting to cdn-datasets.huggingface.co (cdn-datasets.huggingface.co)|99.84.181.104|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 204844092 (195M) [application/x-tar]\n",
            "Saving to: ‘xsum.tar.gz.1’\n",
            "\n",
            "xsum.tar.gz.1       100%[===================>] 195.35M   234MB/s    in 0.8s    \n",
            "\n",
            "2020-11-17 14:56:41 (234 MB/s) - ‘xsum.tar.gz.1’ saved [204844092/204844092]\n",
            "\n",
            "xsum/\n",
            "xsum/train.target\n",
            "xsum/train.source\n",
            "xsum/val.source\n",
            "xsum/val.target\n",
            "xsum/test.source\n",
            "xsum/test.target\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HlDoDzJMfVQy"
      },
      "source": [
        "import os\n",
        "os.environ['XSUM_DIR']=os.path.join(os.getcwd(), 'xsum')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wqUue_HQgFQt",
        "outputId": "890d4cfd-9423-4f42-e121-57d93a380920"
      },
      "source": [
        "!echo $XSUM_DIR"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/transformers/examples/seq2seq/xsum\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LtDHrviaNM7R",
        "outputId": "2faa9122-c36a-4bd1-9017-ef530c80cf39"
      },
      "source": [
        "!ls $XSUM_DIR"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "test.source  test.target  train.source\ttrain.target  val.source  val.target\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "76OEnEKVsVou",
        "outputId": "71acf81a-c3d6-4cee-d36b-abcf7b4dd116"
      },
      "source": [
        "!wc -l $XSUM_DIR/*"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "    11333 /content/transformers/examples/seq2seq/xsum/test.source\n",
            "    11333 /content/transformers/examples/seq2seq/xsum/test.target\n",
            "   204017 /content/transformers/examples/seq2seq/xsum/train.source\n",
            "   204017 /content/transformers/examples/seq2seq/xsum/train.target\n",
            "    11327 /content/transformers/examples/seq2seq/xsum/val.source\n",
            "    11327 /content/transformers/examples/seq2seq/xsum/val.target\n",
            "   453354 total\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "24O2LGrLqNDP",
        "outputId": "0edfe402-e5ff-424e-9d5f-5e60fcf0388f"
      },
      "source": [
        "!wget https://cdn-datasets.huggingface.co/summarization/cnn_dm_v2.tgz\n",
        "!tar -xzvf cnn_dm_v2.tgz  # empty lines removed\n",
        "!mv cnn_cln cnn_dm\n",
        "\n",
        "os.environ['CNN_DIR']=os.path.join(os.getcwd(), 'cnn_dm')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-11-17 14:56:47--  https://cdn-datasets.huggingface.co/summarization/cnn_dm_v2.tgz\n",
            "Resolving cdn-datasets.huggingface.co (cdn-datasets.huggingface.co)... 99.84.181.99, 99.84.181.83, 99.84.181.50, ...\n",
            "Connecting to cdn-datasets.huggingface.co (cdn-datasets.huggingface.co)|99.84.181.99|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 514271123 (490M) [application/x-tar]\n",
            "Saving to: ‘cnn_dm_v2.tgz.1’\n",
            "\n",
            "cnn_dm_v2.tgz.1     100%[===================>] 490.45M   274MB/s    in 1.8s    \n",
            "\n",
            "2020-11-17 14:56:49 (274 MB/s) - ‘cnn_dm_v2.tgz.1’ saved [514271123/514271123]\n",
            "\n",
            "cnn_cln/\n",
            "cnn_cln/train.source\n",
            "cnn_cln/train.target\n",
            "cnn_cln/test.target\n",
            "cnn_cln/test.source\n",
            "cnn_cln/val.source\n",
            "cnn_cln/val.target\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v5wCLhnyg3FD"
      },
      "source": [
        "## Install required package"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ryygCHtEdXdY"
      },
      "source": [
        "%%capture\n",
        "!pip install gitpython\n",
        "!pip install rouge_score\n",
        "!pip install sacrebleu"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "40hjr62hgfSi"
      },
      "source": [
        "## Fine-tune ProphetNet\n",
        "\n",
        "- Due to Colab's time limit (12 hours), about one-tenth of the dataset is used here."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "background_save": true,
          "base_uri": "https://localhost:8080/"
        },
        "id": "lJB3GanOfF0N",
        "outputId": "3b5b0f81-f7dd-4040-ba35-3a949d198b27"
      },
      "source": [
        "!python finetune_trainer.py \\\n",
        "    --learning_rate=3e-5 \\\n",
        "    --fp16 \\\n",
        "    --do_train --do_eval --do_predict --evaluate_during_training \\\n",
        "    --predict_with_generate \\\n",
        "    --max_source_length 510 \\\n",
        "    --per_device_train_batch_size 2 \\\n",
        "    --per_device_eval_batch_size 2 \\\n",
        "    --n_train 20400 \\\n",
        "    --n_val 100 \\\n",
        "    --model_name_or_path microsoft/prophetnet-large-uncased \\\n",
        "    --data_dir $XSUM_DIR \\\n",
        "    --output_dir prophetnet_large_uncased_xsum \\\n",
        "    --save_steps 10000 \\\n",
        "    --save_total_limit 5 \\\n",
        "    --overwrite_output_dir "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2020-11-17 14:57:23.833868: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.1\n",
            "/content/transformers/src/transformers/training_args.py:347: FutureWarning: The `evaluate_during_training` argument is deprecated in favor of `evaluation_strategy` (which has more options)\n",
            "  FutureWarning,\n",
            "11/17/2020 14:57:25 - WARNING - __main__ -   Process rank: -1, device: cuda:0, n_gpu: 1, distributed training: False, 16-bits training: True\n",
            "11/17/2020 14:57:25 - INFO - __main__ -   Training/evaluation parameters Seq2SeqTrainingArguments(output_dir='prophetnet_large_uncased_xsum', overwrite_output_dir=True, do_train=True, do_eval=True, do_predict=True, evaluate_during_training=True, evaluation_strategy=<EvaluationStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=2, per_device_eval_batch_size=2, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=1, eval_accumulation_steps=None, learning_rate=3e-05, weight_decay=0.0, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, warmup_steps=0, logging_dir='runs/Nov17_14-57-25_db0ad700327b', logging_first_step=False, logging_steps=500, save_steps=10000, save_total_limit=5, no_cuda=False, seed=42, fp16=True, fp16_opt_level='O1', local_rank=-1, tpu_num_cores=None, tpu_metrics_debug=False, debug=False, dataloader_drop_last=False, eval_steps=500, dataloader_num_workers=0, past_index=-1, run_name='prophetnet_large_uncased_xsum', disable_tqdm=False, remove_unused_columns=True, label_names=None, load_best_model_at_end=False, metric_for_best_model=None, greater_is_better=None, label_smoothing=0.0, sortish_sampler=False, predict_with_generate=True, adafactor=False, encoder_layerdrop=None, decoder_layerdrop=None, dropout=None, attention_dropout=None, lr_scheduler='linear')\n",
            "[INFO|configuration_utils.py:412] 2020-11-17 14:57:25,931 >> loading configuration file https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/config.json from cache at /root/.cache/torch/transformers/5af1cdf83ab91efb16c4b84e954521820508cf6de67834aba5c8d360056ebd81.007d7f4c78db8e2eed8b0afcf7fbeb3cd1ec12a204131bf92811fa10c45947b4\n",
            "[INFO|configuration_utils.py:448] 2020-11-17 14:57:25,932 >> Model config ProphetNetConfig {\n",
            "  \"activation_dropout\": 0.1,\n",
            "  \"activation_function\": \"gelu\",\n",
            "  \"add_cross_attention\": true,\n",
            "  \"architectures\": [\n",
            "    \"ProphetNetForConditionalGeneration\"\n",
            "  ],\n",
            "  \"attention_dropout\": 0.1,\n",
            "  \"bos_token_id\": 102,\n",
            "  \"decoder_ffn_dim\": 4096,\n",
            "  \"decoder_layerdrop\": 0.0,\n",
            "  \"decoder_max_position_embeddings\": 514,\n",
            "  \"decoder_start_token_id\": 102,\n",
            "  \"disable_ngram_loss\": false,\n",
            "  \"dropout\": 0.1,\n",
            "  \"encoder_ffn_dim\": 4096,\n",
            "  \"encoder_layerdrop\": 0.0,\n",
            "  \"encoder_max_position_embeddings\": 513,\n",
            "  \"eos_token_id\": 102,\n",
            "  \"eps\": 0.0,\n",
            "  \"hidden_size\": 1024,\n",
            "  \"id2label\": {\n",
            "    \"0\": \"LABEL_0\",\n",
            "    \"1\": \"LABEL_1\",\n",
            "    \"2\": \"LABEL_2\"\n",
            "  },\n",
            "  \"init_std\": 0.02,\n",
            "  \"is_encoder_decoder\": true,\n",
            "  \"label2id\": {\n",
            "    \"LABEL_0\": 0,\n",
            "    \"LABEL_1\": 1,\n",
            "    \"LABEL_2\": 2\n",
            "  },\n",
            "  \"max_position_embeddings\": 512,\n",
            "  \"model_type\": \"prophetnet\",\n",
            "  \"ngram\": 2,\n",
            "  \"num_buckets\": 32,\n",
            "  \"num_decoder_attention_heads\": 16,\n",
            "  \"num_decoder_layers\": 12,\n",
            "  \"num_encoder_attention_heads\": 16,\n",
            "  \"num_encoder_layers\": 12,\n",
            "  \"output_past\": false,\n",
            "  \"pad_token_id\": 0,\n",
            "  \"prefix\": \" \",\n",
            "  \"relative_max_distance\": 128,\n",
            "  \"task_specific_params\": {\n",
            "    \"summarization\": {\n",
            "      \"early_stopping\": true,\n",
            "      \"length_penalty\": 2.0,\n",
            "      \"no_repeat_ngram_size\": 3,\n",
            "      \"num_beams\": 4\n",
            "    }\n",
            "  },\n",
            "  \"vocab_size\": 30522\n",
            "}\n",
            "\n",
            "[INFO|configuration_utils.py:412] 2020-11-17 14:57:26,097 >> loading configuration file https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/config.json from cache at /root/.cache/torch/transformers/5af1cdf83ab91efb16c4b84e954521820508cf6de67834aba5c8d360056ebd81.007d7f4c78db8e2eed8b0afcf7fbeb3cd1ec12a204131bf92811fa10c45947b4\n",
            "[INFO|configuration_utils.py:448] 2020-11-17 14:57:26,097 >> Model config ProphetNetConfig {\n",
            "  \"activation_dropout\": 0.1,\n",
            "  \"activation_function\": \"gelu\",\n",
            "  \"add_cross_attention\": true,\n",
            "  \"architectures\": [\n",
            "    \"ProphetNetForConditionalGeneration\"\n",
            "  ],\n",
            "  \"attention_dropout\": 0.1,\n",
            "  \"bos_token_id\": 102,\n",
            "  \"decoder_ffn_dim\": 4096,\n",
            "  \"decoder_layerdrop\": 0.0,\n",
            "  \"decoder_max_position_embeddings\": 514,\n",
            "  \"decoder_start_token_id\": 102,\n",
            "  \"disable_ngram_loss\": false,\n",
            "  \"dropout\": 0.1,\n",
            "  \"encoder_ffn_dim\": 4096,\n",
            "  \"encoder_layerdrop\": 0.0,\n",
            "  \"encoder_max_position_embeddings\": 513,\n",
            "  \"eos_token_id\": 102,\n",
            "  \"eps\": 0.0,\n",
            "  \"hidden_size\": 1024,\n",
            "  \"id2label\": {\n",
            "    \"0\": \"LABEL_0\",\n",
            "    \"1\": \"LABEL_1\",\n",
            "    \"2\": \"LABEL_2\"\n",
            "  },\n",
            "  \"init_std\": 0.02,\n",
            "  \"is_encoder_decoder\": true,\n",
            "  \"label2id\": {\n",
            "    \"LABEL_0\": 0,\n",
            "    \"LABEL_1\": 1,\n",
            "    \"LABEL_2\": 2\n",
            "  },\n",
            "  \"max_position_embeddings\": 512,\n",
            "  \"model_type\": \"prophetnet\",\n",
            "  \"ngram\": 2,\n",
            "  \"num_buckets\": 32,\n",
            "  \"num_decoder_attention_heads\": 16,\n",
            "  \"num_decoder_layers\": 12,\n",
            "  \"num_encoder_attention_heads\": 16,\n",
            "  \"num_encoder_layers\": 12,\n",
            "  \"output_past\": false,\n",
            "  \"pad_token_id\": 0,\n",
            "  \"prefix\": \" \",\n",
            "  \"relative_max_distance\": 128,\n",
            "  \"task_specific_params\": {\n",
            "    \"summarization\": {\n",
            "      \"early_stopping\": true,\n",
            "      \"length_penalty\": 2.0,\n",
            "      \"no_repeat_ngram_size\": 3,\n",
            "      \"num_beams\": 4\n",
            "    }\n",
            "  },\n",
            "  \"vocab_size\": 30522\n",
            "}\n",
            "\n",
            "[INFO|tokenization_utils_base.py:1780] 2020-11-17 14:57:26,180 >> loading file https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/prophetnet.tokenizer from cache at /root/.cache/torch/transformers/3b5ecfc96b12e126f8589d998af18d8e17970efd33f511b9b1d212cb418a7aa6.bfccaaa55dc78254b9fdcf3c43f1799da93d9e66f1cead1d46d464130f9e9649\n",
            "11/17/2020 14:57:26 - INFO - filelock -   Lock 140294570376160 acquired on /root/.cache/torch/transformers/3d0f611b9f59504ec6dc841d79edeeef72f013b046b9aeec430882d215480a27.ab2e9c0807ddfec82dd02c95d416e2e2db96bc2a83f2b8a3c45d729831a78030.lock\n",
            "[INFO|file_utils.py:1194] 2020-11-17 14:57:26,283 >> https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/pytorch_model.bin not found in cache or force_download set to True, downloading to /root/.cache/torch/transformers/tmp7uy_5m0s\n",
            "Downloading: 100% 1.57G/1.57G [00:26<00:00, 60.0MB/s]\n",
            "[INFO|file_utils.py:1198] 2020-11-17 14:57:52,491 >> storing https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/pytorch_model.bin in cache at /root/.cache/torch/transformers/3d0f611b9f59504ec6dc841d79edeeef72f013b046b9aeec430882d215480a27.ab2e9c0807ddfec82dd02c95d416e2e2db96bc2a83f2b8a3c45d729831a78030\n",
            "[INFO|file_utils.py:1201] 2020-11-17 14:57:52,491 >> creating metadata file for /root/.cache/torch/transformers/3d0f611b9f59504ec6dc841d79edeeef72f013b046b9aeec430882d215480a27.ab2e9c0807ddfec82dd02c95d416e2e2db96bc2a83f2b8a3c45d729831a78030\n",
            "11/17/2020 14:57:52 - INFO - filelock -   Lock 140294570376160 released on /root/.cache/torch/transformers/3d0f611b9f59504ec6dc841d79edeeef72f013b046b9aeec430882d215480a27.ab2e9c0807ddfec82dd02c95d416e2e2db96bc2a83f2b8a3c45d729831a78030.lock\n",
            "[INFO|modeling_utils.py:939] 2020-11-17 14:57:52,492 >> loading weights file https://huggingface.co/microsoft/prophetnet-large-uncased/resolve/main/pytorch_model.bin from cache at /root/.cache/torch/transformers/3d0f611b9f59504ec6dc841d79edeeef72f013b046b9aeec430882d215480a27.ab2e9c0807ddfec82dd02c95d416e2e2db96bc2a83f2b8a3c45d729831a78030\n",
            "[INFO|modeling_utils.py:1055] 2020-11-17 14:58:15,129 >> All model checkpoint weights were used when initializing ProphetNetForConditionalGeneration.\n",
            "\n",
            "[INFO|modeling_utils.py:1064] 2020-11-17 14:58:15,130 >> All the weights of ProphetNetForConditionalGeneration were initialized from the model checkpoint at microsoft/prophetnet-large-uncased.\n",
            "If your task is similar to the task the model of the checkpoint was trained on, you can already use ProphetNetForConditionalGeneration for predictions without further training.\n",
            "11/17/2020 14:58:15 - INFO - utils -   using task specific params for summarization: {'early_stopping': True, 'length_penalty': 2.0, 'no_repeat_ngram_size': 3, 'num_beams': 4}\n",
            "[INFO|trainer.py:693] 2020-11-17 14:58:34,617 >> ***** Running training *****\n",
            "[INFO|trainer.py:694] 2020-11-17 14:58:34,617 >>   Num examples = 20400\n",
            "[INFO|trainer.py:695] 2020-11-17 14:58:34,618 >>   Num Epochs = 3\n",
            "[INFO|trainer.py:696] 2020-11-17 14:58:34,618 >>   Instantaneous batch size per device = 2\n",
            "[INFO|trainer.py:697] 2020-11-17 14:58:34,618 >>   Total train batch size (w. parallel, distributed & accumulation) = 2\n",
            "[INFO|trainer.py:698] 2020-11-17 14:58:34,618 >>   Gradient Accumulation steps = 1\n",
            "[INFO|trainer.py:699] 2020-11-17 14:58:34,618 >>   Total optimization steps = 30600\n",
            "  0% 0/30600 [00:00<?, ?it/s]/usr/local/lib/python3.6/dist-packages/torch/optim/lr_scheduler.py:136: UserWarning: Detected call of `lr_scheduler.step()` before `optimizer.step()`. In PyTorch 1.1.0 and later, you should call them in the opposite order: `optimizer.step()` before `lr_scheduler.step()`.  Failure to do this will result in PyTorch skipping the first value of the learning rate schedule. See more details at https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate\n",
            "  \"https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate\", UserWarning)\n",
            "{'loss': 2.514990478515625, 'learning_rate': 2.9509803921568626e-05, 'epoch': 0.049019607843137254}\n",
            "  2% 500/30600 [04:46<4:49:11,  1.73it/s][INFO|trainer.py:1387] 2020-11-17 15:03:21,104 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:03:21,104 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:03:21,104 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.24it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:46,  1.02it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:52,  1.13s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:57,  1.27s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:58,  1.34s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:56,  1.34s/it]\u001b[A\n",
            " 18% 9/50 [00:11<00:57,  1.41s/it]\u001b[A\n",
            " 20% 10/50 [00:13<01:00,  1.51s/it]\u001b[A\n",
            " 22% 11/50 [00:14<00:56,  1.46s/it]\u001b[A\n",
            " 24% 12/50 [00:16<00:53,  1.42s/it]\u001b[A\n",
            " 26% 13/50 [00:17<00:51,  1.39s/it]\u001b[A\n",
            " 28% 14/50 [00:19<00:50,  1.41s/it]\u001b[A\n",
            " 30% 15/50 [00:20<00:48,  1.38s/it]\u001b[A\n",
            " 32% 16/50 [00:21<00:44,  1.30s/it]\u001b[A\n",
            " 34% 17/50 [00:22<00:44,  1.34s/it]\u001b[A\n",
            " 36% 18/50 [00:24<00:45,  1.42s/it]\u001b[A\n",
            " 38% 19/50 [00:26<00:45,  1.48s/it]\u001b[A\n",
            " 40% 20/50 [00:27<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:29<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:30<00:42,  1.51s/it]\u001b[A\n",
            " 46% 23/50 [00:32<00:38,  1.41s/it]\u001b[A\n",
            " 48% 24/50 [00:33<00:36,  1.41s/it]\u001b[A\n",
            " 50% 25/50 [00:34<00:34,  1.37s/it]\u001b[A\n",
            " 52% 26/50 [00:36<00:34,  1.42s/it]\u001b[A\n",
            " 54% 27/50 [00:37<00:31,  1.38s/it]\u001b[A\n",
            " 56% 28/50 [00:39<00:31,  1.42s/it]\u001b[A\n",
            " 58% 29/50 [00:40<00:29,  1.42s/it]\u001b[A\n",
            " 60% 30/50 [00:41<00:27,  1.38s/it]\u001b[A\n",
            " 62% 31/50 [00:43<00:26,  1.42s/it]\u001b[A\n",
            " 64% 32/50 [00:45<00:27,  1.53s/it]\u001b[A\n",
            " 66% 33/50 [00:46<00:25,  1.52s/it]\u001b[A\n",
            " 68% 34/50 [00:47<00:23,  1.45s/it]\u001b[A\n",
            " 70% 35/50 [00:49<00:21,  1.42s/it]\u001b[A\n",
            " 72% 36/50 [00:50<00:20,  1.45s/it]\u001b[A\n",
            " 74% 37/50 [00:52<00:20,  1.54s/it]\u001b[A\n",
            " 76% 38/50 [00:54<00:19,  1.65s/it]\u001b[A\n",
            " 78% 39/50 [00:55<00:16,  1.54s/it]\u001b[A\n",
            " 80% 40/50 [00:57<00:15,  1.58s/it]\u001b[A\n",
            " 82% 41/50 [00:58<00:13,  1.52s/it]\u001b[A\n",
            " 84% 42/50 [01:00<00:12,  1.58s/it]\u001b[A\n",
            " 86% 43/50 [01:02<00:10,  1.55s/it]\u001b[A\n",
            " 88% 44/50 [01:03<00:08,  1.48s/it]\u001b[A\n",
            " 90% 45/50 [01:04<00:07,  1.47s/it]\u001b[A\n",
            " 92% 46/50 [01:06<00:06,  1.55s/it]\u001b[A\n",
            " 94% 47/50 [01:08<00:04,  1.55s/it]\u001b[A\n",
            " 96% 48/50 [01:09<00:03,  1.51s/it]\u001b[A\n",
            " 98% 49/50 [01:11<00:01,  1.56s/it]\u001b[A\n",
            "                                         \n",
            "\u001b[A{'eval_loss': 2.161533832550049, 'eval_rouge1': 35.7163, 'eval_rouge2': 12.6353, 'eval_rougeL': 27.8907, 'eval_rougeLsum': 27.9211, 'eval_gen_len': 23.4, 'epoch': 0.049019607843137254}\n",
            "  2% 500/30600 [06:01<4:49:11,  1.73it/s]\n",
            "100% 50/50 [01:13<00:00,  1.57s/it]\u001b[A\n",
            "{'loss': 2.411956787109375, 'learning_rate': 2.9019607843137255e-05, 'epoch': 0.09803921568627451}\n",
            "  3% 1000/30600 [10:54<4:50:41,  1.70it/s][INFO|trainer.py:1387] 2020-11-17 15:09:29,567 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:09:29,568 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:09:29,568 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.18it/s]\u001b[A\n",
            "  6% 3/50 [00:03<01:00,  1.28s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.35s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:03,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:05,  1.53s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:09,  1.66s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:08,  1.66s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.64s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:03,  1.64s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:05,  1.73s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:02,  1.70s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:02,  1.73s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:57,  1.65s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:50,  1.48s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:52,  1.60s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:49,  1.55s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:53,  1.73s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.65s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:50,  1.73s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:43,  1.56s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:41,  1.53s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:45,  1.73s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:44,  1.76s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:43,  1.82s/it]\u001b[A\n",
            " 54% 27/50 [00:44<00:41,  1.78s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:37,  1.73s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:35,  1.68s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:33,  1.66s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:35,  1.87s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:31,  1.78s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:29,  1.74s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:27,  1.74s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:26,  1.77s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:25,  1.82s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:22,  1.74s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:21,  1.80s/it]\u001b[A\n",
            " 78% 39/50 [01:05<00:18,  1.72s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:16,  1.68s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.63s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.72s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.65s/it]\u001b[A\n",
            " 88% 44/50 [01:14<00:12,  2.05s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:09,  1.92s/it]\u001b[A\n",
            " 92% 46/50 [01:18<00:07,  1.87s/it]\u001b[A\n",
            " 94% 47/50 [01:20<00:06,  2.05s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.81s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.83s/it]\u001b[A\n",
            "                                          \n",
            "\u001b[A{'eval_loss': 2.35081148147583, 'eval_rouge1': 32.4986, 'eval_rouge2': 10.7431, 'eval_rougeL': 24.9911, 'eval_rougeLsum': 25.1585, 'eval_gen_len': 26.2, 'epoch': 0.09803921568627451}\n",
            "  3% 1000/30600 [12:22<4:50:41,  1.70it/s]\n",
            "100% 50/50 [01:25<00:00,  1.78s/it]\u001b[A\n",
            "{'loss': 2.35504931640625, 'learning_rate': 2.8529411764705883e-05, 'epoch': 0.14705882352941177}\n",
            "  5% 1500/30600 [17:14<4:53:13,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 15:15:49,638 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:15:49,638 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:15:49,638 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.25it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.09s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:04,  1.39s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:05,  1.46s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:08,  1.55s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:07,  1.58s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:06,  1.59s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:08,  1.66s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:06,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:07,  1.72s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:06,  1.76s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:01,  1.67s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:01,  1.70s/it]\u001b[A\n",
            " 30% 15/50 [00:24<00:57,  1.63s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:53,  1.57s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:53,  1.61s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.56s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:50,  1.63s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:48,  1.67s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:44,  1.59s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:42,  1.56s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:38,  1.54s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:36,  1.53s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:38,  1.75s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:36,  1.74s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:33,  1.69s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:32,  1.73s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:35,  1.97s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:31,  1.86s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.80s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:26,  1.76s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:24,  1.72s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:21,  1.78s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:18,  1.67s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:17,  1.76s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:15,  1.73s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:14,  1.78s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:12,  1.80s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.77s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.71s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.74s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.82s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.68s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.61s/it]\u001b[A\n",
            "                                          \n",
            "\u001b[A{'eval_loss': 2.110572099685669, 'eval_rouge1': 35.1291, 'eval_rouge2': 13.2243, 'eval_rougeL': 28.3829, 'eval_rougeLsum': 28.3851, 'eval_gen_len': 26.5, 'epoch': 0.14705882352941177}\n",
            "  5% 1500/30600 [18:40<4:53:13,  1.65it/s]\n",
            "100% 50/50 [01:23<00:00,  1.61s/it]\u001b[A\n",
            "{'loss': 2.31517919921875, 'learning_rate': 2.8039215686274512e-05, 'epoch': 0.19607843137254902}\n",
            "  7% 2000/30600 [23:32<4:43:29,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 15:22:07,015 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:22:07,015 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:22:07,015 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:34,  1.38it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:57,  1.25s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.36s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.46s/it]\u001b[A\n",
            " 16% 8/50 [00:11<00:59,  1.42s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.61s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:04,  1.60s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.57s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:59,  1.56s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:58,  1.57s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:58,  1.63s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:56,  1.61s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:50,  1.50s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.49s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:51,  1.61s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:49,  1.59s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:51,  1.70s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:49,  1.72s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:43,  1.57s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:40,  1.51s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:39,  1.51s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:36,  1.45s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:35,  1.49s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:32,  1.40s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:32,  1.47s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:34,  1.63s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:31,  1.55s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:29,  1.55s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:29,  1.62s/it]\u001b[A\n",
            " 66% 33/50 [00:50<00:27,  1.64s/it]\u001b[A\n",
            " 68% 34/50 [00:52<00:25,  1.60s/it]\u001b[A\n",
            " 70% 35/50 [00:53<00:24,  1.60s/it]\u001b[A\n",
            " 72% 36/50 [00:55<00:23,  1.67s/it]\u001b[A\n",
            " 74% 37/50 [00:57<00:20,  1.60s/it]\u001b[A\n",
            " 76% 38/50 [00:59<00:20,  1.72s/it]\u001b[A\n",
            " 78% 39/50 [01:00<00:17,  1.62s/it]\u001b[A\n",
            " 80% 40/50 [01:02<00:16,  1.60s/it]\u001b[A\n",
            " 82% 41/50 [01:03<00:13,  1.50s/it]\u001b[A\n",
            " 84% 42/50 [01:05<00:12,  1.60s/it]\u001b[A\n",
            " 86% 43/50 [01:06<00:10,  1.55s/it]\u001b[A\n",
            " 88% 44/50 [01:08<00:09,  1.51s/it]\u001b[A\n",
            " 90% 45/50 [01:09<00:07,  1.47s/it]\u001b[A\n",
            " 92% 46/50 [01:11<00:06,  1.54s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:04,  1.60s/it]\u001b[A\n",
            " 96% 48/50 [01:14<00:03,  1.54s/it]\u001b[A\n",
            " 98% 49/50 [01:15<00:01,  1.53s/it]\u001b[A\n",
            "                                          \n",
            "\u001b[A{'eval_loss': 2.1514880657196045, 'eval_rouge1': 34.2593, 'eval_rouge2': 11.3961, 'eval_rougeL': 26.9222, 'eval_rougeLsum': 26.9119, 'eval_gen_len': 24.4, 'epoch': 0.19607843137254902}\n",
            "  7% 2000/30600 [24:52<4:43:29,  1.68it/s]\n",
            "100% 50/50 [01:18<00:00,  1.59s/it]\u001b[A\n",
            "{'loss': 2.303767578125, 'learning_rate': 2.7549019607843137e-05, 'epoch': 0.24509803921568626}\n",
            "  8% 2500/30600 [29:43<4:43:59,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 15:28:17,844 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:28:17,845 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:28:17,845 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.27it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.09s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.23s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:58,  1.30s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:59,  1.36s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:57,  1.35s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:57,  1.38s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:59,  1.46s/it]\u001b[A\n",
            " 20% 10/50 [00:13<00:55,  1.39s/it]\u001b[A\n",
            " 22% 11/50 [00:14<00:52,  1.35s/it]\u001b[A\n",
            " 24% 12/50 [00:16<00:51,  1.34s/it]\u001b[A\n",
            " 26% 13/50 [00:17<00:50,  1.35s/it]\u001b[A\n",
            " 28% 14/50 [00:18<00:49,  1.36s/it]\u001b[A\n",
            " 30% 15/50 [00:20<00:46,  1.34s/it]\u001b[A\n",
            " 32% 16/50 [00:21<00:48,  1.42s/it]\u001b[A\n",
            " 34% 17/50 [00:23<00:46,  1.42s/it]\u001b[A\n",
            " 36% 18/50 [00:24<00:44,  1.40s/it]\u001b[A\n",
            " 38% 19/50 [00:26<00:45,  1.48s/it]\u001b[A\n",
            " 40% 20/50 [00:27<00:45,  1.50s/it]\u001b[A\n",
            " 42% 21/50 [00:29<00:44,  1.55s/it]\u001b[A\n",
            " 44% 22/50 [00:30<00:40,  1.45s/it]\u001b[A\n",
            " 46% 23/50 [00:31<00:37,  1.40s/it]\u001b[A\n",
            " 48% 24/50 [00:33<00:38,  1.50s/it]\u001b[A\n",
            " 50% 25/50 [00:35<00:38,  1.53s/it]\u001b[A\n",
            " 52% 26/50 [00:36<00:35,  1.46s/it]\u001b[A\n",
            " 54% 27/50 [00:38<00:33,  1.48s/it]\u001b[A\n",
            " 56% 28/50 [00:39<00:31,  1.44s/it]\u001b[A\n",
            " 58% 29/50 [00:40<00:30,  1.46s/it]\u001b[A\n",
            " 60% 30/50 [00:42<00:29,  1.45s/it]\u001b[A\n",
            " 62% 31/50 [00:43<00:28,  1.48s/it]\u001b[A\n",
            " 64% 32/50 [00:45<00:28,  1.57s/it]\u001b[A\n",
            " 66% 33/50 [00:47<00:26,  1.57s/it]\u001b[A\n",
            " 68% 34/50 [00:48<00:24,  1.52s/it]\u001b[A\n",
            " 70% 35/50 [00:50<00:22,  1.52s/it]\u001b[A\n",
            " 72% 36/50 [00:51<00:22,  1.58s/it]\u001b[A\n",
            " 74% 37/50 [00:53<00:20,  1.55s/it]\u001b[A\n",
            " 76% 38/50 [00:54<00:17,  1.50s/it]\u001b[A\n",
            " 78% 39/50 [00:56<00:16,  1.52s/it]\u001b[A\n",
            " 80% 40/50 [00:57<00:14,  1.50s/it]\u001b[A\n",
            " 82% 41/50 [00:59<00:13,  1.45s/it]\u001b[A\n",
            " 84% 42/50 [01:00<00:12,  1.54s/it]\u001b[A\n",
            " 86% 43/50 [01:02<00:10,  1.51s/it]\u001b[A\n",
            " 88% 44/50 [01:03<00:09,  1.51s/it]\u001b[A\n",
            " 90% 45/50 [01:05<00:07,  1.48s/it]\u001b[A\n",
            " 92% 46/50 [01:06<00:05,  1.49s/it]\u001b[A\n",
            " 94% 47/50 [01:08<00:04,  1.54s/it]\u001b[A\n",
            " 96% 48/50 [01:09<00:02,  1.45s/it]\u001b[A\n",
            " 98% 49/50 [01:11<00:01,  1.47s/it]\u001b[A\n",
            "100% 50/50 [01:12<00:00,  1.47s/it]\u001b[A\n",
            "{'eval_loss': 2.1534409523010254, 'eval_rouge1': 35.1626, 'eval_rouge2': 12.2013, 'eval_rougeL': 27.1744, 'eval_rougeLsum': 27.188, 'eval_gen_len': 23.0, 'epoch': 0.24509803921568626}\n",
            "\n",
            "  8% 2500/30600 [30:58<4:43:59,  1.65it/s]\n",
            "{'loss': 2.3356748046875, 'learning_rate': 2.7058823529411766e-05, 'epoch': 0.29411764705882354}\n",
            " 10% 3000/30600 [35:47<4:30:37,  1.70it/s][INFO|trainer.py:1387] 2020-11-17 15:34:22,110 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:34:22,110 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:34:22,110 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.24it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.23s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:02,  1.39s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.46s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.41s/it]\u001b[A\n",
            " 16% 8/50 [00:11<00:59,  1.41s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.47s/it]\u001b[A\n",
            " 20% 10/50 [00:14<00:59,  1.48s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:59,  1.51s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:55,  1.45s/it]\u001b[A\n",
            " 26% 13/50 [00:18<00:57,  1.57s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:55,  1.55s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:50,  1.45s/it]\u001b[A\n",
            " 32% 16/50 [00:22<00:46,  1.38s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.43s/it]\u001b[A\n",
            " 36% 18/50 [00:25<00:45,  1.43s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:46,  1.51s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:46,  1.53s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:44,  1.52s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:41,  1.49s/it]\u001b[A\n",
            " 46% 23/50 [00:33<00:39,  1.46s/it]\u001b[A\n",
            " 48% 24/50 [00:34<00:37,  1.45s/it]\u001b[A\n",
            " 50% 25/50 [00:36<00:35,  1.41s/it]\u001b[A\n",
            " 52% 26/50 [00:37<00:32,  1.35s/it]\u001b[A\n",
            " 54% 27/50 [00:38<00:32,  1.40s/it]\u001b[A\n",
            " 56% 28/50 [00:40<00:33,  1.51s/it]\u001b[A\n",
            " 58% 29/50 [00:42<00:32,  1.57s/it]\u001b[A\n",
            " 60% 30/50 [00:44<00:32,  1.61s/it]\u001b[A\n",
            " 62% 31/50 [00:45<00:29,  1.58s/it]\u001b[A\n",
            " 64% 32/50 [00:47<00:28,  1.59s/it]\u001b[A\n",
            " 66% 33/50 [00:49<00:28,  1.66s/it]\u001b[A\n",
            " 68% 34/50 [00:50<00:24,  1.55s/it]\u001b[A\n",
            " 70% 35/50 [00:51<00:23,  1.58s/it]\u001b[A\n",
            " 72% 36/50 [00:53<00:23,  1.65s/it]\u001b[A\n",
            " 74% 37/50 [00:55<00:21,  1.63s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:20,  1.71s/it]\u001b[A\n",
            " 78% 39/50 [00:58<00:17,  1.59s/it]\u001b[A\n",
            " 80% 40/50 [01:00<00:15,  1.59s/it]\u001b[A\n",
            " 82% 41/50 [01:01<00:13,  1.48s/it]\u001b[A\n",
            " 84% 42/50 [01:02<00:11,  1.48s/it]\u001b[A\n",
            " 86% 43/50 [01:04<00:10,  1.53s/it]\u001b[A\n",
            " 88% 44/50 [01:06<00:09,  1.54s/it]\u001b[A\n",
            " 90% 45/50 [01:07<00:07,  1.54s/it]\u001b[A\n",
            " 92% 46/50 [01:09<00:06,  1.56s/it]\u001b[A\n",
            " 94% 47/50 [01:10<00:04,  1.57s/it]\u001b[A\n",
            " 96% 48/50 [01:12<00:03,  1.51s/it]\u001b[A\n",
            " 98% 49/50 [01:13<00:01,  1.56s/it]\u001b[A\n",
            "100% 50/50 [01:15<00:00,  1.50s/it]\u001b[A\n",
            "{'eval_loss': 2.1197800636291504, 'eval_rouge1': 35.1159, 'eval_rouge2': 13.7567, 'eval_rougeL': 27.7594, 'eval_rougeLsum': 27.7901, 'eval_gen_len': 23.7, 'epoch': 0.29411764705882354}\n",
            "\n",
            " 10% 3000/30600 [37:04<4:30:37,  1.70it/s]\n",
            "{'loss': 2.2800302734375, 'learning_rate': 2.656862745098039e-05, 'epoch': 0.3431372549019608}\n",
            " 11% 3500/30600 [41:56<5:29:25,  1.37it/s][INFO|trainer.py:1387] 2020-11-17 15:40:31,536 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:40:31,536 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:40:31,536 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.22it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:52,  1.11s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:58,  1.28s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:04,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.41s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:11<00:59,  1.42s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.47s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:02,  1.56s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:03,  1.62s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:56,  1.49s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:57,  1.55s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:55,  1.55s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:53,  1.52s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:47,  1.40s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.44s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:45,  1.44s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:45,  1.47s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:43,  1.46s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:44,  1.52s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:41,  1.47s/it]\u001b[A\n",
            " 46% 23/50 [00:33<00:41,  1.54s/it]\u001b[A\n",
            " 48% 24/50 [00:35<00:39,  1.53s/it]\u001b[A\n",
            " 50% 25/50 [00:36<00:37,  1.50s/it]\u001b[A\n",
            " 52% 26/50 [00:38<00:36,  1.52s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:41<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:34,  1.63s/it]\u001b[A\n",
            " 60% 30/50 [00:44<00:31,  1.59s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:29,  1.57s/it]\u001b[A\n",
            " 64% 32/50 [00:47<00:28,  1.59s/it]\u001b[A\n",
            " 66% 33/50 [00:49<00:28,  1.67s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:26,  1.63s/it]\u001b[A\n",
            " 70% 35/50 [00:53<00:24,  1.63s/it]\u001b[A\n",
            " 72% 36/50 [00:54<00:23,  1.68s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:21,  1.64s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:19,  1.61s/it]\u001b[A\n",
            " 78% 39/50 [00:59<00:18,  1.65s/it]\u001b[A\n",
            " 80% 40/50 [01:01<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:14,  1.61s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:13,  1.67s/it]\u001b[A\n",
            " 86% 43/50 [01:05<00:11,  1.58s/it]\u001b[A\n",
            " 88% 44/50 [01:07<00:09,  1.53s/it]\u001b[A\n",
            " 90% 45/50 [01:09<00:08,  1.62s/it]\u001b[A\n",
            " 92% 46/50 [01:10<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:05,  1.76s/it]\u001b[A\n",
            " 96% 48/50 [01:14<00:03,  1.73s/it]\u001b[A\n",
            " 98% 49/50 [01:16<00:01,  1.68s/it]\u001b[A\n",
            "                                          \n",
            "\u001b[A{'eval_loss': 2.102212905883789, 'eval_rouge1': 34.9543, 'eval_rouge2': 13.0656, 'eval_rougeL': 27.3549, 'eval_rougeLsum': 27.4297, 'eval_gen_len': 24.6, 'epoch': 0.3431372549019608}\n",
            " 11% 3500/30600 [43:17<5:29:25,  1.37it/s]\n",
            "100% 50/50 [01:18<00:00,  1.67s/it]\u001b[A\n",
            "{'loss': 2.262490234375, 'learning_rate': 2.607843137254902e-05, 'epoch': 0.39215686274509803}\n",
            " 13% 4000/30600 [48:09<4:23:43,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 15:46:44,166 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:46:44,166 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:46:44,166 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.25it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.11s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.22s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.37s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.45s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.42s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:59,  1.41s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:59,  1.46s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:01,  1.53s/it]\u001b[A\n",
            " 22% 11/50 [00:15<01:00,  1.56s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:58,  1.54s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:01,  1.67s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:59,  1.64s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:53,  1.53s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:47,  1.41s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:46,  1.40s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:46,  1.45s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:47,  1.53s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:47,  1.60s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:46,  1.59s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:42,  1.54s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:41,  1.55s/it]\u001b[A\n",
            " 48% 24/50 [00:35<00:38,  1.47s/it]\u001b[A\n",
            " 50% 25/50 [00:36<00:35,  1.43s/it]\u001b[A\n",
            " 52% 26/50 [00:38<00:33,  1.41s/it]\u001b[A\n",
            " 54% 27/50 [00:39<00:33,  1.48s/it]\u001b[A\n",
            " 56% 28/50 [00:41<00:34,  1.59s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:35,  1.69s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:33,  1.69s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:31,  1.67s/it]\u001b[A\n",
            " 64% 32/50 [00:48<00:30,  1.70s/it]\u001b[A\n",
            " 66% 33/50 [00:50<00:29,  1.73s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:26,  1.64s/it]\u001b[A\n",
            " 70% 35/50 [00:53<00:23,  1.55s/it]\u001b[A\n",
            " 72% 36/50 [00:55<00:23,  1.65s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:20,  1.61s/it]\u001b[A\n",
            " 76% 38/50 [00:58<00:19,  1.60s/it]\u001b[A\n",
            " 78% 39/50 [01:00<00:18,  1.67s/it]\u001b[A\n",
            " 80% 40/50 [01:01<00:16,  1.69s/it]\u001b[A\n",
            " 82% 41/50 [01:03<00:15,  1.77s/it]\u001b[A\n",
            " 84% 42/50 [01:05<00:14,  1.77s/it]\u001b[A\n",
            " 86% 43/50 [01:07<00:12,  1.72s/it]\u001b[A\n",
            " 88% 44/50 [01:08<00:10,  1.67s/it]\u001b[A\n",
            " 90% 45/50 [01:10<00:08,  1.70s/it]\u001b[A\n",
            " 92% 46/50 [01:12<00:06,  1.66s/it]\u001b[A\n",
            " 94% 47/50 [01:13<00:05,  1.73s/it]\u001b[A\n",
            " 96% 48/50 [01:15<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:16<00:01,  1.59s/it]\u001b[A\n",
            "                                          \n",
            "\u001b[A{'eval_loss': 2.111318826675415, 'eval_rouge1': 37.0742, 'eval_rouge2': 13.7096, 'eval_rougeL': 28.5197, 'eval_rougeLsum': 28.5479, 'eval_gen_len': 24.5, 'epoch': 0.39215686274509803}\n",
            " 13% 4000/30600 [49:30<4:23:43,  1.68it/s]\n",
            "100% 50/50 [01:19<00:00,  1.61s/it]\u001b[A\n",
            "{'loss': 2.247935546875, 'learning_rate': 2.5588235294117648e-05, 'epoch': 0.4411764705882353}\n",
            " 15% 4500/30600 [54:24<4:18:42,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 15:52:58,741 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:52:58,741 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:52:58,741 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.29it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.07s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:55,  1.21s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:59,  1.33s/it]\u001b[A\n",
            " 12% 6/50 [00:07<01:00,  1.37s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.38s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:59,  1.41s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:59,  1.46s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:00,  1.52s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:04,  1.66s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:02,  1.65s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:04,  1.75s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:59,  1.64s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:57,  1.63s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:51,  1.51s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.51s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:47,  1.49s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:48,  1.56s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:47,  1.57s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:46,  1.59s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:42,  1.51s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:42,  1.59s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:37,  1.49s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:36,  1.53s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:34,  1.51s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:31,  1.45s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:32,  1.54s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:30,  1.55s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:28,  1.51s/it]\u001b[A\n",
            " 64% 32/50 [00:48<00:28,  1.57s/it]\u001b[A\n",
            " 66% 33/50 [00:49<00:25,  1.51s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:23,  1.49s/it]\u001b[A\n",
            " 70% 35/50 [00:53<00:23,  1.55s/it]\u001b[A\n",
            " 72% 36/50 [00:54<00:21,  1.56s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:20,  1.57s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:18,  1.56s/it]\u001b[A\n",
            " 78% 39/50 [00:59<00:17,  1.55s/it]\u001b[A\n",
            " 80% 40/50 [01:00<00:15,  1.55s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:13,  1.50s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:12,  1.58s/it]\u001b[A\n",
            " 86% 43/50 [01:05<00:10,  1.56s/it]\u001b[A\n",
            " 88% 44/50 [01:07<00:09,  1.54s/it]\u001b[A\n",
            " 90% 45/50 [01:08<00:07,  1.48s/it]\u001b[A\n",
            " 92% 46/50 [01:10<00:06,  1.54s/it]\u001b[A\n",
            " 94% 47/50 [01:11<00:04,  1.63s/it]\u001b[A\n",
            " 96% 48/50 [01:13<00:03,  1.55s/it]\u001b[A\n",
            " 98% 49/50 [01:14<00:01,  1.53s/it]\u001b[A\n",
            "100% 50/50 [01:16<00:00,  1.63s/it]\u001b[A\n",
            "{'eval_loss': 2.0960121154785156, 'eval_rouge1': 34.5841, 'eval_rouge2': 12.7286, 'eval_rougeL': 27.4372, 'eval_rougeLsum': 27.4592, 'eval_gen_len': 23.4, 'epoch': 0.4411764705882353}\n",
            "\n",
            " 15% 4500/30600 [55:43<4:18:42,  1.68it/s]\n",
            "{'loss': 2.28320703125, 'learning_rate': 2.5098039215686277e-05, 'epoch': 0.49019607843137253}\n",
            " 16% 5000/30600 [1:00:35<4:02:49,  1.76it/s][INFO|trainer.py:1387] 2020-11-17 15:59:09,954 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 15:59:09,954 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 15:59:09,954 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:35,  1.34it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:59,  1.30s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.36s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:05,  1.48s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:03,  1.48s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:00,  1.45s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:58,  1.42s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:01,  1.55s/it]\u001b[A\n",
            " 22% 11/50 [00:16<00:59,  1.53s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:56,  1.50s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:58,  1.57s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:56,  1.57s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:54,  1.55s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:48,  1.43s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.43s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:48,  1.52s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:48,  1.56s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:52,  1.75s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:52,  1.82s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:47,  1.69s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:45,  1.68s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:42,  1.63s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:39,  1.58s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:37,  1.56s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:35,  1.60s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:33,  1.62s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:33,  1.67s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:30,  1.63s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:30,  1.68s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:27,  1.64s/it]\u001b[A\n",
            " 68% 34/50 [00:53<00:26,  1.63s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:23,  1.60s/it]\u001b[A\n",
            " 72% 36/50 [00:56<00:23,  1.67s/it]\u001b[A\n",
            " 74% 37/50 [00:58<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [00:59<00:20,  1.68s/it]\u001b[A\n",
            " 78% 39/50 [01:01<00:17,  1.61s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:16,  1.67s/it]\u001b[A\n",
            " 82% 41/50 [01:04<00:15,  1.72s/it]\u001b[A\n",
            " 84% 42/50 [01:06<00:14,  1.81s/it]\u001b[A\n",
            " 86% 43/50 [01:08<00:11,  1.66s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:10,  1.70s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:08,  1.64s/it]\u001b[A\n",
            " 92% 46/50 [01:13<00:06,  1.65s/it]\u001b[A\n",
            " 94% 47/50 [01:14<00:04,  1.64s/it]\u001b[A\n",
            " 96% 48/50 [01:16<00:03,  1.58s/it]\u001b[A\n",
            " 98% 49/50 [01:17<00:01,  1.60s/it]\u001b[A\n",
            "                                            \n",
            "\u001b[A{'eval_loss': 2.0912344455718994, 'eval_rouge1': 36.8901, 'eval_rouge2': 13.6944, 'eval_rougeL': 29.6322, 'eval_rougeLsum': 29.6928, 'eval_gen_len': 24.8, 'epoch': 0.49019607843137253}\n",
            " 16% 5000/30600 [1:01:57<4:02:49,  1.76it/s]\n",
            "100% 50/50 [01:20<00:00,  1.63s/it]\u001b[A\n",
            "{'loss': 2.246970703125, 'learning_rate': 2.4607843137254902e-05, 'epoch': 0.5392156862745098}\n",
            " 18% 5500/30600 [1:06:49<4:04:05,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 16:05:23,800 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:05:23,800 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:05:23,800 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.20it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:56,  1.20s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.34s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:06,  1.48s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:07,  1.53s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:05,  1.53s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:06,  1.57s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.58s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:02,  1.57s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:03,  1.63s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:02,  1.64s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:00,  1.63s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:58,  1.63s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:54,  1.55s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:48,  1.44s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:50,  1.52s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:54,  1.70s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:55,  1.81s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:52,  1.75s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:50,  1.76s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.62s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:46,  1.73s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:44,  1.72s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.62s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:36,  1.59s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:36,  1.67s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:36,  1.74s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:34,  1.71s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:34,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:33,  1.85s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:30,  1.79s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:27,  1.72s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:25,  1.69s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:24,  1.72s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:20,  1.71s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:18,  1.68s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:17,  1.72s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:15,  1.73s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.73s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.64s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:09,  1.63s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.61s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:04,  1.66s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.58s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.48s/it]\u001b[A\n",
            "                                            \n",
            "\u001b[A{'eval_loss': 2.073159694671631, 'eval_rouge1': 37.4599, 'eval_rouge2': 14.3243, 'eval_rougeL': 28.9527, 'eval_rougeLsum': 29.0057, 'eval_gen_len': 26.1, 'epoch': 0.5392156862745098}\n",
            " 18% 5500/30600 [1:08:13<4:04:05,  1.71it/s]\n",
            "100% 50/50 [01:22<00:00,  1.51s/it]\u001b[A\n",
            "{'loss': 2.258375, 'learning_rate': 2.411764705882353e-05, 'epoch': 0.5882352941176471}\n",
            " 20% 6000/30600 [1:13:04<3:56:10,  1.74it/s][INFO|trainer.py:1387] 2020-11-17 16:11:39,137 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:11:39,137 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:11:39,137 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.28it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:00,  1.32s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:00,  1.34s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.39s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:00,  1.45s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:57,  1.41s/it]\u001b[A\n",
            " 20% 10/50 [00:14<00:59,  1.48s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:57,  1.48s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:57,  1.51s/it]\u001b[A\n",
            " 26% 13/50 [00:18<00:56,  1.52s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:56,  1.57s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:53,  1.52s/it]\u001b[A\n",
            " 32% 16/50 [00:22<00:47,  1.41s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:45,  1.38s/it]\u001b[A\n",
            " 36% 18/50 [00:25<00:44,  1.40s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:45,  1.47s/it]\u001b[A\n",
            " 40% 20/50 [00:28<00:42,  1.43s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:42,  1.46s/it]\u001b[A\n",
            " 44% 22/50 [00:31<00:40,  1.44s/it]\u001b[A\n",
            " 46% 23/50 [00:32<00:38,  1.43s/it]\u001b[A\n",
            " 48% 24/50 [00:34<00:36,  1.39s/it]\u001b[A\n",
            " 50% 25/50 [00:35<00:34,  1.37s/it]\u001b[A\n",
            " 52% 26/50 [00:36<00:32,  1.37s/it]\u001b[A\n",
            " 54% 27/50 [00:38<00:31,  1.38s/it]\u001b[A\n",
            " 56% 28/50 [00:39<00:30,  1.38s/it]\u001b[A\n",
            " 58% 29/50 [00:41<00:31,  1.50s/it]\u001b[A\n",
            " 60% 30/50 [00:42<00:27,  1.40s/it]\u001b[A\n",
            " 62% 31/50 [00:44<00:28,  1.52s/it]\u001b[A\n",
            " 64% 32/50 [00:46<00:28,  1.57s/it]\u001b[A\n",
            " 66% 33/50 [00:47<00:26,  1.56s/it]\u001b[A\n",
            " 68% 34/50 [00:49<00:24,  1.51s/it]\u001b[A\n",
            " 70% 35/50 [00:50<00:22,  1.49s/it]\u001b[A\n",
            " 72% 36/50 [00:52<00:21,  1.55s/it]\u001b[A\n",
            " 74% 37/50 [00:53<00:19,  1.53s/it]\u001b[A\n",
            " 76% 38/50 [00:55<00:17,  1.47s/it]\u001b[A\n",
            " 78% 39/50 [00:56<00:16,  1.51s/it]\u001b[A\n",
            " 80% 40/50 [00:58<00:15,  1.55s/it]\u001b[A\n",
            " 82% 41/50 [00:59<00:13,  1.45s/it]\u001b[A\n",
            " 84% 42/50 [01:01<00:11,  1.50s/it]\u001b[A\n",
            " 86% 43/50 [01:02<00:10,  1.52s/it]\u001b[A\n",
            " 88% 44/50 [01:04<00:10,  1.69s/it]\u001b[A\n",
            " 90% 45/50 [01:06<00:08,  1.62s/it]\u001b[A\n",
            " 92% 46/50 [01:07<00:06,  1.59s/it]\u001b[A\n",
            " 94% 47/50 [01:09<00:04,  1.61s/it]\u001b[A\n",
            " 96% 48/50 [01:10<00:03,  1.54s/it]\u001b[A\n",
            " 98% 49/50 [01:12<00:01,  1.53s/it]\u001b[A\n",
            "100% 50/50 [01:13<00:00,  1.48s/it]\u001b[A\n",
            "{'eval_loss': 2.101222038269043, 'eval_rouge1': 36.4949, 'eval_rouge2': 13.8728, 'eval_rougeL': 29.1504, 'eval_rougeLsum': 29.2481, 'eval_gen_len': 24.0, 'epoch': 0.5882352941176471}\n",
            "\n",
            " 20% 6000/30600 [1:14:20<3:56:10,  1.74it/s]\n",
            "{'loss': 2.276171875, 'learning_rate': 2.3627450980392156e-05, 'epoch': 0.6372549019607843}\n",
            " 21% 6500/30600 [1:19:08<4:00:32,  1.67it/s][INFO|trainer.py:1387] 2020-11-17 16:17:43,009 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:17:43,009 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:17:43,009 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:35,  1.34it/s]\u001b[A\n",
            "  6% 3/50 [00:02<00:43,  1.08it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:52,  1.14s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:57,  1.27s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:59,  1.36s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:10<01:00,  1.43s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:59,  1.46s/it]\u001b[A\n",
            " 20% 10/50 [00:13<01:00,  1.51s/it]\u001b[A\n",
            " 22% 11/50 [00:15<01:04,  1.66s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:59,  1.58s/it]\u001b[A\n",
            " 26% 13/50 [00:18<01:00,  1.65s/it]\u001b[A\n",
            " 28% 14/50 [00:20<01:00,  1.68s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:59,  1.69s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:52,  1.55s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.51s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:48,  1.51s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:48,  1.57s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:50,  1.68s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:49,  1.71s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:45,  1.63s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:44,  1.66s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:41,  1.59s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:40,  1.63s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:38,  1.60s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:34,  1.50s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:34,  1.56s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:32,  1.55s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:30,  1.53s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:30,  1.63s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:30,  1.69s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:28,  1.68s/it]\u001b[A\n",
            " 68% 34/50 [00:52<00:26,  1.63s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:24,  1.62s/it]\u001b[A\n",
            " 72% 36/50 [00:55<00:22,  1.58s/it]\u001b[A\n",
            " 74% 37/50 [00:57<00:20,  1.57s/it]\u001b[A\n",
            " 76% 38/50 [00:58<00:18,  1.51s/it]\u001b[A\n",
            " 78% 39/50 [01:00<00:16,  1.53s/it]\u001b[A\n",
            " 80% 40/50 [01:02<00:16,  1.60s/it]\u001b[A\n",
            " 82% 41/50 [01:03<00:14,  1.59s/it]\u001b[A\n",
            " 84% 42/50 [01:05<00:12,  1.60s/it]\u001b[A\n",
            " 86% 43/50 [01:06<00:10,  1.56s/it]\u001b[A\n",
            " 88% 44/50 [01:08<00:09,  1.60s/it]\u001b[A\n",
            " 90% 45/50 [01:10<00:08,  1.62s/it]\u001b[A\n",
            " 92% 46/50 [01:11<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:13<00:04,  1.65s/it]\u001b[A\n",
            " 96% 48/50 [01:14<00:03,  1.62s/it]\u001b[A\n",
            " 98% 49/50 [01:16<00:01,  1.67s/it]\u001b[A\n",
            "100% 50/50 [01:18<00:00,  1.65s/it]\u001b[A\n",
            "{'eval_loss': 2.066157579421997, 'eval_rouge1': 35.7128, 'eval_rouge2': 12.0595, 'eval_rougeL': 27.4511, 'eval_rougeLsum': 27.4708, 'eval_gen_len': 25.8, 'epoch': 0.6372549019607843}\n",
            "\n",
            " 21% 6500/30600 [1:20:28<4:00:32,  1.67it/s]\n",
            "{'loss': 2.2565, 'learning_rate': 2.3137254901960785e-05, 'epoch': 0.6862745098039216}\n",
            " 23% 7000/30600 [1:25:14<3:50:56,  1.70it/s][INFO|trainer.py:1387] 2020-11-17 16:23:49,055 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:23:49,055 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:23:49,055 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:36,  1.31it/s]\u001b[A\n",
            "  6% 3/50 [00:02<00:43,  1.07it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:50,  1.10s/it]\u001b[A\n",
            " 10% 5/50 [00:05<00:52,  1.16s/it]\u001b[A\n",
            " 12% 6/50 [00:06<00:53,  1.21s/it]\u001b[A\n",
            " 14% 7/50 [00:08<00:54,  1.26s/it]\u001b[A\n",
            " 16% 8/50 [00:09<00:56,  1.35s/it]\u001b[A\n",
            " 18% 9/50 [00:11<00:57,  1.40s/it]\u001b[A\n",
            " 20% 10/50 [00:13<00:58,  1.46s/it]\u001b[A\n",
            " 22% 11/50 [00:14<00:57,  1.47s/it]\u001b[A\n",
            " 24% 12/50 [00:15<00:54,  1.44s/it]\u001b[A\n",
            " 26% 13/50 [00:17<00:54,  1.48s/it]\u001b[A\n",
            " 28% 14/50 [00:18<00:53,  1.48s/it]\u001b[A\n",
            " 30% 15/50 [00:20<00:53,  1.52s/it]\u001b[A\n",
            " 32% 16/50 [00:21<00:46,  1.38s/it]\u001b[A\n",
            " 34% 17/50 [00:23<00:46,  1.42s/it]\u001b[A\n",
            " 36% 18/50 [00:24<00:44,  1.39s/it]\u001b[A\n",
            " 38% 19/50 [00:25<00:42,  1.36s/it]\u001b[A\n",
            " 40% 20/50 [00:27<00:45,  1.52s/it]\u001b[A\n",
            " 42% 21/50 [00:29<00:45,  1.58s/it]\u001b[A\n",
            " 44% 22/50 [00:30<00:42,  1.52s/it]\u001b[A\n",
            " 46% 23/50 [00:32<00:39,  1.48s/it]\u001b[A\n",
            " 48% 24/50 [00:33<00:38,  1.46s/it]\u001b[A\n",
            " 50% 25/50 [00:35<00:37,  1.48s/it]\u001b[A\n",
            " 52% 26/50 [00:36<00:35,  1.48s/it]\u001b[A\n",
            " 54% 27/50 [00:37<00:33,  1.46s/it]\u001b[A\n",
            " 56% 28/50 [00:39<00:33,  1.51s/it]\u001b[A\n",
            " 58% 29/50 [00:41<00:33,  1.62s/it]\u001b[A\n",
            " 60% 30/50 [00:42<00:30,  1.54s/it]\u001b[A\n",
            " 62% 31/50 [00:44<00:29,  1.57s/it]\u001b[A\n",
            " 64% 32/50 [00:46<00:28,  1.59s/it]\u001b[A\n",
            " 66% 33/50 [00:47<00:28,  1.66s/it]\u001b[A\n",
            " 68% 34/50 [00:49<00:25,  1.58s/it]\u001b[A\n",
            " 70% 35/50 [00:50<00:22,  1.50s/it]\u001b[A\n",
            " 72% 36/50 [00:52<00:22,  1.60s/it]\u001b[A\n",
            " 74% 37/50 [00:53<00:20,  1.59s/it]\u001b[A\n",
            " 76% 38/50 [00:55<00:18,  1.58s/it]\u001b[A\n",
            " 78% 39/50 [00:57<00:17,  1.57s/it]\u001b[A\n",
            " 80% 40/50 [00:58<00:15,  1.51s/it]\u001b[A\n",
            " 82% 41/50 [00:59<00:13,  1.50s/it]\u001b[A\n",
            " 84% 42/50 [01:02<00:13,  1.67s/it]\u001b[A\n",
            " 86% 43/50 [01:03<00:10,  1.55s/it]\u001b[A\n",
            " 88% 44/50 [01:04<00:09,  1.57s/it]\u001b[A\n",
            " 90% 45/50 [01:06<00:07,  1.50s/it]\u001b[A\n",
            " 92% 46/50 [01:07<00:05,  1.50s/it]\u001b[A\n",
            " 94% 47/50 [01:09<00:04,  1.44s/it]\u001b[A\n",
            " 96% 48/50 [01:10<00:02,  1.40s/it]\u001b[A\n",
            " 98% 49/50 [01:11<00:01,  1.46s/it]\u001b[A\n",
            "100% 50/50 [01:13<00:00,  1.45s/it]\u001b[A\n",
            "{'eval_loss': 2.0928397178649902, 'eval_rouge1': 34.8673, 'eval_rouge2': 12.6307, 'eval_rougeL': 27.3406, 'eval_rougeLsum': 27.4417, 'eval_gen_len': 23.9, 'epoch': 0.6862745098039216}\n",
            "\n",
            " 23% 7000/30600 [1:26:29<3:50:56,  1.70it/s]\n",
            "{'loss': 2.218107421875, 'learning_rate': 2.2647058823529413e-05, 'epoch': 0.7352941176470589}\n",
            " 25% 7500/30600 [1:31:15<3:40:18,  1.75it/s][INFO|trainer.py:1387] 2020-11-17 16:29:50,046 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:29:50,047 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:29:50,047 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.24it/s]\u001b[A\n",
            "  6% 3/50 [00:02<00:45,  1.04it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:55,  1.20s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.37s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.41s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:59,  1.42s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:58,  1.42s/it]\u001b[A\n",
            " 20% 10/50 [00:13<00:58,  1.45s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:59,  1.54s/it]\u001b[A\n",
            " 24% 12/50 [00:16<00:56,  1.48s/it]\u001b[A\n",
            " 26% 13/50 [00:18<00:54,  1.47s/it]\u001b[A\n",
            " 28% 14/50 [00:19<00:53,  1.49s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:51,  1.48s/it]\u001b[A\n",
            " 32% 16/50 [00:22<00:47,  1.40s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.43s/it]\u001b[A\n",
            " 36% 18/50 [00:25<00:44,  1.40s/it]\u001b[A\n",
            " 38% 19/50 [00:26<00:43,  1.40s/it]\u001b[A\n",
            " 40% 20/50 [00:28<00:44,  1.47s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:44,  1.53s/it]\u001b[A\n",
            " 44% 22/50 [00:31<00:41,  1.47s/it]\u001b[A\n",
            " 46% 23/50 [00:32<00:40,  1.50s/it]\u001b[A\n",
            " 48% 24/50 [00:34<00:39,  1.52s/it]\u001b[A\n",
            " 50% 25/50 [00:35<00:36,  1.48s/it]\u001b[A\n",
            " 52% 26/50 [00:37<00:35,  1.48s/it]\u001b[A\n",
            " 54% 27/50 [00:38<00:31,  1.37s/it]\u001b[A\n",
            " 56% 28/50 [00:40<00:30,  1.41s/it]\u001b[A\n",
            " 58% 29/50 [00:42<00:33,  1.61s/it]\u001b[A\n",
            " 60% 30/50 [00:44<00:36,  1.80s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:34,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:47<00:31,  1.74s/it]\u001b[A\n",
            " 66% 33/50 [00:49<00:31,  1.83s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:27,  1.71s/it]\u001b[A\n",
            " 70% 35/50 [00:52<00:24,  1.63s/it]\u001b[A\n",
            " 72% 36/50 [00:54<00:22,  1.63s/it]\u001b[A\n",
            " 74% 37/50 [00:55<00:20,  1.59s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:19,  1.64s/it]\u001b[A\n",
            " 78% 39/50 [00:58<00:17,  1.58s/it]\u001b[A\n",
            " 80% 40/50 [01:00<00:15,  1.58s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:14,  1.61s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:13,  1.68s/it]\u001b[A\n",
            " 86% 43/50 [01:05<00:11,  1.68s/it]\u001b[A\n",
            " 88% 44/50 [01:07<00:09,  1.63s/it]\u001b[A\n",
            " 90% 45/50 [01:08<00:08,  1.63s/it]\u001b[A\n",
            " 92% 46/50 [01:10<00:06,  1.63s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:04,  1.60s/it]\u001b[A\n",
            " 96% 48/50 [01:13<00:03,  1.53s/it]\u001b[A\n",
            " 98% 49/50 [01:14<00:01,  1.47s/it]\u001b[A\n",
            "                                            \n",
            "\u001b[A{'eval_loss': 2.0723466873168945, 'eval_rouge1': 37.1728, 'eval_rouge2': 14.7507, 'eval_rougeL': 28.8539, 'eval_rougeLsum': 28.8269, 'eval_gen_len': 24.2, 'epoch': 0.7352941176470589}\n",
            " 25% 7500/30600 [1:32:33<3:40:18,  1.75it/s]\n",
            "100% 50/50 [01:16<00:00,  1.45s/it]\u001b[A\n",
            "{'loss': 2.1724765625, 'learning_rate': 2.2156862745098042e-05, 'epoch': 0.7843137254901961}\n",
            " 26% 8000/30600 [1:37:21<3:36:41,  1.74it/s][INFO|trainer.py:1387] 2020-11-17 16:35:56,506 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:35:56,506 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:35:56,506 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.26it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:55,  1.19s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.33s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.46s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:04,  1.49s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.54s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.56s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.64s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:07,  1.74s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:05,  1.72s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:03,  1.72s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:01,  1.71s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:57,  1.65s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:51,  1.52s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:53,  1.62s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:51,  1.60s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.56s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:48,  1.68s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:46,  1.65s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:47,  1.74s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:45,  1.73s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:43,  1.75s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:41,  1.74s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:38,  1.67s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:35,  1.62s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:35,  1.67s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:33,  1.66s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:32,  1.71s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:32,  1.80s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:29,  1.76s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:27,  1.70s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:24,  1.64s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:23,  1.67s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:23,  1.77s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:21,  1.77s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:18,  1.67s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:16,  1.68s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:15,  1.67s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.75s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.71s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.70s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.79s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.63s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.60s/it]\u001b[A\n",
            "{'eval_loss': 2.0454189777374268, 'eval_rouge1': 35.3579, 'eval_rouge2': 13.8031, 'eval_rougeL': 27.6122, 'eval_rougeLsum': 27.551, 'eval_gen_len': 27.1, 'epoch': 0.7843137254901961}\n",
            "\n",
            " 26% 8000/30600 [1:38:46<3:36:41,  1.74it/s]\n",
            "{'loss': 2.23821484375, 'learning_rate': 2.1666666666666667e-05, 'epoch': 0.8333333333333334}\n",
            " 28% 8500/30600 [1:43:38<3:37:29,  1.69it/s][INFO|trainer.py:1387] 2020-11-17 16:42:12,678 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:42:12,678 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:42:12,678 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.27it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:49,  1.05s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:57,  1.25s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.40s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.43s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.42s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.53s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:02,  1.53s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:00,  1.52s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:00,  1.55s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:58,  1.54s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:58,  1.57s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:55,  1.55s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:52,  1.49s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:47,  1.41s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.45s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:47,  1.48s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:48,  1.56s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:49,  1.66s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:48,  1.68s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:45,  1.63s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:41,  1.53s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:40,  1.56s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:41,  1.65s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:40,  1.68s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:38,  1.66s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:36,  1.65s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:34,  1.63s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:34,  1.71s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:32,  1.71s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:31,  1.76s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:30,  1.76s/it]\u001b[A\n",
            " 68% 34/50 [00:53<00:27,  1.73s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:24,  1.61s/it]\u001b[A\n",
            " 72% 36/50 [00:56<00:23,  1.66s/it]\u001b[A\n",
            " 74% 37/50 [00:58<00:21,  1.66s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:20,  1.74s/it]\u001b[A\n",
            " 78% 39/50 [01:01<00:18,  1.64s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:04<00:14,  1.57s/it]\u001b[A\n",
            " 84% 42/50 [01:06<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:08<00:11,  1.66s/it]\u001b[A\n",
            " 88% 44/50 [01:09<00:09,  1.64s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:07,  1.55s/it]\u001b[A\n",
            " 92% 46/50 [01:12<00:06,  1.55s/it]\u001b[A\n",
            " 94% 47/50 [01:14<00:04,  1.62s/it]\u001b[A\n",
            " 96% 48/50 [01:15<00:03,  1.56s/it]\u001b[A\n",
            " 98% 49/50 [01:17<00:01,  1.58s/it]\u001b[A\n",
            "100% 50/50 [01:19<00:00,  1.60s/it]\u001b[A\n",
            "{'eval_loss': 2.060051918029785, 'eval_rouge1': 37.6962, 'eval_rouge2': 15.0508, 'eval_rougeL': 29.6479, 'eval_rougeLsum': 29.7274, 'eval_gen_len': 25.0, 'epoch': 0.8333333333333334}\n",
            "\n",
            " 28% 8500/30600 [1:44:59<3:37:29,  1.69it/s]\n",
            "{'loss': 2.18363671875, 'learning_rate': 2.1176470588235296e-05, 'epoch': 0.8823529411764706}\n",
            " 29% 9000/30600 [1:49:52<3:34:21,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 16:48:26,759 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:48:26,759 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:48:26,759 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:34,  1.40it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:46,  1.01it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:55,  1.21s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:02,  1.38s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.40s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.54s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.58s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:05,  1.64s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:05,  1.67s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.70s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:01,  1.65s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:58,  1.61s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:54,  1.56s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:49,  1.45s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.50s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:50,  1.58s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:49,  1.61s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:50,  1.67s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:49,  1.71s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.61s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:45,  1.70s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:43,  1.68s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:40,  1.61s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:38,  1.61s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.60s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.66s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:34,  1.63s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:34,  1.71s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:33,  1.75s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:32,  1.80s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:30,  1.78s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:27,  1.73s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:25,  1.71s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:24,  1.74s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.76s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:20,  1.73s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.63s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:15,  1.56s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:14,  1.59s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:12,  1.60s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:11,  1.64s/it]\u001b[A\n",
            " 88% 44/50 [01:11<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:08,  1.64s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.71s/it]\u001b[A\n",
            " 94% 47/50 [01:16<00:05,  1.73s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.63s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.50s/it]\u001b[A\n",
            "{'eval_loss': 2.076655387878418, 'eval_rouge1': 36.0325, 'eval_rouge2': 13.0023, 'eval_rougeL': 28.079, 'eval_rougeLsum': 27.9554, 'eval_gen_len': 26.3, 'epoch': 0.8823529411764706}\n",
            "\n",
            " 29% 9000/30600 [1:51:15<3:34:21,  1.68it/s]\n",
            "{'loss': 2.18879296875, 'learning_rate': 2.068627450980392e-05, 'epoch': 0.9313725490196079}\n",
            " 31% 9500/30600 [1:56:05<3:25:55,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 16:54:40,469 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 16:54:40,469 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 16:54:40,469 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:36,  1.32it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.22s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:02,  1.39s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.41s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.43s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:00,  1.43s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:01,  1.49s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:00,  1.51s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:58,  1.49s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:01,  1.61s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:00,  1.62s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:55,  1.55s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:51,  1.49s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:47,  1.40s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:48,  1.46s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:45,  1.41s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:45,  1.47s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:43,  1.46s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:44,  1.54s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:43,  1.54s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:43,  1.62s/it]\u001b[A\n",
            " 48% 24/50 [00:35<00:41,  1.59s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:37,  1.52s/it]\u001b[A\n",
            " 52% 26/50 [00:38<00:34,  1.46s/it]\u001b[A\n",
            " 54% 27/50 [00:39<00:32,  1.43s/it]\u001b[A\n",
            " 56% 28/50 [00:41<00:34,  1.57s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:33,  1.60s/it]\u001b[A\n",
            " 60% 30/50 [00:44<00:32,  1.63s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:32,  1.69s/it]\u001b[A\n",
            " 64% 32/50 [00:48<00:30,  1.68s/it]\u001b[A\n",
            " 66% 33/50 [00:50<00:29,  1.73s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:26,  1.63s/it]\u001b[A\n",
            " 70% 35/50 [00:52<00:22,  1.52s/it]\u001b[A\n",
            " 72% 36/50 [00:54<00:21,  1.56s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:20,  1.56s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:18,  1.56s/it]\u001b[A\n",
            " 78% 39/50 [00:59<00:17,  1.55s/it]\u001b[A\n",
            " 80% 40/50 [01:00<00:15,  1.57s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:14,  1.56s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:12,  1.62s/it]\u001b[A\n",
            " 86% 43/50 [01:05<00:11,  1.63s/it]\u001b[A\n",
            " 88% 44/50 [01:07<00:09,  1.64s/it]\u001b[A\n",
            " 90% 45/50 [01:08<00:07,  1.55s/it]\u001b[A\n",
            " 92% 46/50 [01:10<00:06,  1.54s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:04,  1.61s/it]\u001b[A\n",
            " 96% 48/50 [01:13<00:03,  1.65s/it]\u001b[A\n",
            " 98% 49/50 [01:15<00:01,  1.66s/it]\u001b[A\n",
            "100% 50/50 [01:17<00:00,  1.75s/it]\u001b[A\n",
            "{'eval_loss': 2.0293967723846436, 'eval_rouge1': 36.4727, 'eval_rouge2': 13.7247, 'eval_rougeL': 27.6967, 'eval_rougeLsum': 27.7447, 'eval_gen_len': 24.5, 'epoch': 0.9313725490196079}\n",
            "\n",
            " 31% 9500/30600 [1:57:25<3:25:55,  1.71it/s]\n",
            "{'loss': 2.21886328125, 'learning_rate': 2.019607843137255e-05, 'epoch': 0.9803921568627451}\n",
            " 33% 10000/30600 [2:02:16<3:28:28,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 17:00:51,435 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:00:51,435 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:00:51,435 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.28it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.16s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.34s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:07,  1.50s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.46s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:02,  1.46s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.55s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.59s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:04,  1.61s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:05,  1.68s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:05,  1.74s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:02,  1.70s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:57,  1.59s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:54,  1.55s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:49,  1.46s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:49,  1.50s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:47,  1.49s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:47,  1.52s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:46,  1.57s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:50,  1.76s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.61s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:44,  1.66s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:43,  1.66s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:38,  1.54s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:37,  1.56s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:34,  1.50s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:34,  1.58s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:34,  1.65s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:32,  1.61s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:30,  1.59s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:30,  1.72s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:29,  1.72s/it]\u001b[A\n",
            " 68% 34/50 [00:53<00:26,  1.66s/it]\u001b[A\n",
            " 70% 35/50 [00:55<00:25,  1.70s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:23,  1.70s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.71s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:19,  1.62s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.60s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:16,  1.65s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:15,  1.68s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:14,  1.79s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:10,  1.74s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:08,  1.65s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:16<00:05,  1.75s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.65s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.54s/it]\u001b[A\n",
            "{'eval_loss': 2.030104160308838, 'eval_rouge1': 37.2683, 'eval_rouge2': 14.1836, 'eval_rougeL': 29.2232, 'eval_rougeLsum': 29.2844, 'eval_gen_len': 25.4, 'epoch': 0.9803921568627451}\n",
            "\n",
            " 33% 10000/30600 [2:03:39<3:28:28,  1.65it/s]\n",
            "                                   \u001b[A[INFO|trainer.py:1222] 2020-11-17 17:02:14,199 >> Saving model checkpoint to prophetnet_large_uncased_xsum/checkpoint-10000\n",
            "[INFO|configuration_utils.py:282] 2020-11-17 17:02:14,202 >> Configuration saved in prophetnet_large_uncased_xsum/checkpoint-10000/config.json\n",
            "[INFO|modeling_utils.py:740] 2020-11-17 17:02:21,740 >> Model weights saved in prophetnet_large_uncased_xsum/checkpoint-10000/pytorch_model.bin\n",
            "{'loss': 1.81625390625, 'learning_rate': 1.9705882352941178e-05, 'epoch': 1.0294117647058822}\n",
            " 34% 10500/30600 [2:09:55<3:50:27,  1.45it/s][INFO|trainer.py:1387] 2020-11-17 17:08:30,219 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:08:30,219 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:08:30,219 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:36,  1.33it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:47,  1.01s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:55,  1.21s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:56,  1.26s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:57,  1.31s/it]\u001b[A\n",
            " 14% 7/50 [00:08<00:56,  1.31s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:58,  1.40s/it]\u001b[A\n",
            " 18% 9/50 [00:11<00:58,  1.42s/it]\u001b[A\n",
            " 20% 10/50 [00:13<00:58,  1.46s/it]\u001b[A\n",
            " 22% 11/50 [00:15<01:01,  1.58s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:02,  1.64s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:02,  1.70s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:59,  1.65s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:53,  1.53s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:49,  1.44s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:47,  1.43s/it]\u001b[A\n",
            " 36% 18/50 [00:25<00:44,  1.39s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:45,  1.48s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:49,  1.65s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:43,  1.56s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:41,  1.55s/it]\u001b[A\n",
            " 48% 24/50 [00:35<00:40,  1.56s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:42,  1.70s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:39,  1.66s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:36,  1.57s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:32,  1.55s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:32,  1.62s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:30,  1.63s/it]\u001b[A\n",
            " 64% 32/50 [00:48<00:28,  1.60s/it]\u001b[A\n",
            " 66% 33/50 [00:50<00:26,  1.57s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:24,  1.53s/it]\u001b[A\n",
            " 70% 35/50 [00:52<00:22,  1.47s/it]\u001b[A\n",
            " 72% 36/50 [00:54<00:21,  1.54s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:19,  1.51s/it]\u001b[A\n",
            " 76% 38/50 [00:57<00:18,  1.55s/it]\u001b[A\n",
            " 78% 39/50 [00:59<00:16,  1.50s/it]\u001b[A\n",
            " 80% 40/50 [01:01<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:14,  1.58s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:13,  1.66s/it]\u001b[A\n",
            " 86% 43/50 [01:05<00:10,  1.55s/it]\u001b[A\n",
            " 88% 44/50 [01:07<00:09,  1.60s/it]\u001b[A\n",
            " 90% 45/50 [01:08<00:07,  1.57s/it]\u001b[A\n",
            " 92% 46/50 [01:10<00:06,  1.60s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:04,  1.62s/it]\u001b[A\n",
            " 96% 48/50 [01:13<00:03,  1.55s/it]\u001b[A\n",
            " 98% 49/50 [01:14<00:01,  1.46s/it]\u001b[A\n",
            "100% 50/50 [01:16<00:00,  1.57s/it]\u001b[A\n",
            "{'eval_loss': 2.0765509605407715, 'eval_rouge1': 36.7227, 'eval_rouge2': 14.125, 'eval_rougeL': 29.0954, 'eval_rougeLsum': 29.1834, 'eval_gen_len': 23.3, 'epoch': 1.0294117647058822}\n",
            "\n",
            " 34% 10500/30600 [2:11:14<3:50:27,  1.45it/s]\n",
            "{'loss': 1.61987109375, 'learning_rate': 1.9215686274509803e-05, 'epoch': 1.0784313725490196}\n",
            " 36% 11000/30600 [2:16:06<3:14:28,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 17:14:40,845 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:14:40,845 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:14:40,845 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.28it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.22s/it]\u001b[A\n",
            " 10% 5/50 [00:06<00:59,  1.32s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:57,  1.31s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.42s/it]\u001b[A\n",
            " 16% 8/50 [00:10<01:00,  1.43s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:02,  1.52s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:02,  1.56s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:59,  1.52s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:04,  1.68s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:00,  1.63s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:57,  1.61s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:56,  1.61s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:51,  1.51s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:50,  1.54s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:47,  1.50s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:50,  1.64s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:48,  1.67s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:46,  1.65s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:48,  1.80s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:46,  1.78s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:44,  1.76s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:41,  1.72s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:39,  1.72s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:37,  1.71s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:35,  1.70s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:34,  1.74s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:31,  1.68s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:30,  1.69s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:30,  1.77s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:28,  1.76s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:25,  1.67s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:23,  1.71s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.73s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:19,  1.66s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.61s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:16,  1.65s/it]\u001b[A\n",
            " 82% 41/50 [01:06<00:15,  1.69s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:14,  1.79s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:12,  1.80s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:11,  1.83s/it]\u001b[A\n",
            " 90% 45/50 [01:13<00:09,  1.86s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:07,  1.79s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.72s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.61s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.57s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.59s/it]\u001b[A\n",
            "{'eval_loss': 2.074441909790039, 'eval_rouge1': 36.9519, 'eval_rouge2': 13.0667, 'eval_rougeL': 28.7135, 'eval_rougeLsum': 28.8154, 'eval_gen_len': 26.5, 'epoch': 1.0784313725490196}\n",
            "\n",
            " 36% 11000/30600 [2:17:30<3:14:28,  1.68it/s]\n",
            "{'loss': 1.60398828125, 'learning_rate': 1.8725490196078432e-05, 'epoch': 1.1274509803921569}\n",
            " 38% 11500/30600 [2:22:20<3:08:11,  1.69it/s][INFO|trainer.py:1387] 2020-11-17 17:20:54,812 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:20:54,812 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:20:54,812 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.21it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:49,  1.05s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.23s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:00,  1.35s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.43s/it]\u001b[A\n",
            " 16% 8/50 [00:10<00:58,  1.40s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:58,  1.43s/it]\u001b[A\n",
            " 20% 10/50 [00:13<00:57,  1.43s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:54,  1.40s/it]\u001b[A\n",
            " 24% 12/50 [00:16<00:57,  1.50s/it]\u001b[A\n",
            " 26% 13/50 [00:18<00:56,  1.52s/it]\u001b[A\n",
            " 28% 14/50 [00:19<00:52,  1.45s/it]\u001b[A\n",
            " 30% 15/50 [00:21<00:54,  1.54s/it]\u001b[A\n",
            " 32% 16/50 [00:22<00:48,  1.44s/it]\u001b[A\n",
            " 34% 17/50 [00:24<00:49,  1.49s/it]\u001b[A\n",
            " 36% 18/50 [00:25<00:46,  1.47s/it]\u001b[A\n",
            " 38% 19/50 [00:27<00:44,  1.45s/it]\u001b[A\n",
            " 40% 20/50 [00:28<00:45,  1.53s/it]\u001b[A\n",
            " 42% 21/50 [00:30<00:47,  1.63s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:43,  1.56s/it]\u001b[A\n",
            " 46% 23/50 [00:33<00:44,  1.64s/it]\u001b[A\n",
            " 48% 24/50 [00:35<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:36<00:38,  1.56s/it]\u001b[A\n",
            " 52% 26/50 [00:38<00:38,  1.61s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:36,  1.58s/it]\u001b[A\n",
            " 56% 28/50 [00:41<00:36,  1.64s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:33,  1.57s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:32,  1.63s/it]\u001b[A\n",
            " 62% 31/50 [00:46<00:29,  1.55s/it]\u001b[A\n",
            " 64% 32/50 [00:48<00:28,  1.61s/it]\u001b[A\n",
            " 66% 33/50 [00:49<00:28,  1.66s/it]\u001b[A\n",
            " 68% 34/50 [00:51<00:26,  1.67s/it]\u001b[A\n",
            " 70% 35/50 [00:53<00:24,  1.63s/it]\u001b[A\n",
            " 72% 36/50 [00:55<00:23,  1.69s/it]\u001b[A\n",
            " 74% 37/50 [00:56<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [00:58<00:19,  1.60s/it]\u001b[A\n",
            " 78% 39/50 [00:59<00:17,  1.57s/it]\u001b[A\n",
            " 80% 40/50 [01:01<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:02<00:14,  1.64s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:06<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:08<00:09,  1.66s/it]\u001b[A\n",
            " 90% 45/50 [01:09<00:07,  1.59s/it]\u001b[A\n",
            " 92% 46/50 [01:11<00:06,  1.62s/it]\u001b[A\n",
            " 94% 47/50 [01:12<00:04,  1.61s/it]\u001b[A\n",
            " 96% 48/50 [01:14<00:03,  1.53s/it]\u001b[A\n",
            " 98% 49/50 [01:15<00:01,  1.49s/it]\u001b[A\n",
            "100% 50/50 [01:17<00:00,  1.50s/it]\u001b[A\n",
            "{'eval_loss': 2.072082281112671, 'eval_rouge1': 35.5595, 'eval_rouge2': 12.304, 'eval_rougeL': 27.5911, 'eval_rougeLsum': 27.5441, 'eval_gen_len': 24.5, 'epoch': 1.1274509803921569}\n",
            "\n",
            " 38% 11500/30600 [2:23:39<3:08:11,  1.69it/s]\n",
            "{'loss': 1.634765625, 'learning_rate': 1.8235294117647057e-05, 'epoch': 1.1764705882352942}\n",
            " 39% 12000/30600 [2:28:30<3:04:22,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 17:27:05,478 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:27:05,478 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:27:05,478 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.20it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.07s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:59,  1.28s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:00,  1.34s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:00,  1.38s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.46s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:02,  1.53s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:01,  1.55s/it]\u001b[A\n",
            " 22% 11/50 [00:15<00:59,  1.53s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:00,  1.58s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:57,  1.56s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:58,  1.63s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:55,  1.58s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:50,  1.48s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:50,  1.54s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:47,  1.49s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:47,  1.52s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:47,  1.59s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:45,  1.57s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:43,  1.60s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:42,  1.64s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:39,  1.56s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:37,  1.56s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:34,  1.50s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:43<00:32,  1.56s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:33,  1.66s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:31,  1.65s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:30,  1.67s/it]\u001b[A\n",
            " 66% 33/50 [00:50<00:28,  1.69s/it]\u001b[A\n",
            " 68% 34/50 [00:52<00:27,  1.71s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:25,  1.71s/it]\u001b[A\n",
            " 72% 36/50 [00:56<00:23,  1.67s/it]\u001b[A\n",
            " 74% 37/50 [00:57<00:21,  1.69s/it]\u001b[A\n",
            " 76% 38/50 [00:59<00:20,  1.68s/it]\u001b[A\n",
            " 78% 39/50 [01:00<00:18,  1.65s/it]\u001b[A\n",
            " 80% 40/50 [01:02<00:16,  1.69s/it]\u001b[A\n",
            " 82% 41/50 [01:04<00:15,  1.68s/it]\u001b[A\n",
            " 84% 42/50 [01:06<00:13,  1.73s/it]\u001b[A\n",
            " 86% 43/50 [01:07<00:11,  1.65s/it]\u001b[A\n",
            " 88% 44/50 [01:09<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:08,  1.73s/it]\u001b[A\n",
            " 92% 46/50 [01:12<00:06,  1.69s/it]\u001b[A\n",
            " 94% 47/50 [01:14<00:04,  1.66s/it]\u001b[A\n",
            " 96% 48/50 [01:16<00:03,  1.61s/it]\u001b[A\n",
            " 98% 49/50 [01:17<00:01,  1.59s/it]\u001b[A\n",
            "100% 50/50 [01:19<00:00,  1.63s/it]\u001b[A\n",
            "{'eval_loss': 2.0588650703430176, 'eval_rouge1': 36.1226, 'eval_rouge2': 12.979, 'eval_rougeL': 27.8791, 'eval_rougeLsum': 27.9319, 'eval_gen_len': 25.0, 'epoch': 1.1764705882352942}\n",
            "\n",
            " 39% 12000/30600 [2:29:52<3:04:22,  1.68it/s]\n",
            "{'loss': 1.61061328125, 'learning_rate': 1.7745098039215686e-05, 'epoch': 1.2254901960784315}\n",
            " 41% 12500/30600 [2:34:46<2:55:05,  1.72it/s][INFO|trainer.py:1387] 2020-11-17 17:33:21,463 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:33:21,463 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:33:21,463 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:37,  1.29it/s]\u001b[A\n",
            "  6% 3/50 [00:04<01:00,  1.28s/it]\u001b[A\n",
            "  8% 4/50 [00:06<01:09,  1.51s/it]\u001b[A\n",
            " 10% 5/50 [00:08<01:13,  1.64s/it]\u001b[A\n",
            " 12% 6/50 [00:09<01:07,  1.54s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:06,  1.54s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:06,  1.59s/it]\u001b[A\n",
            " 18% 9/50 [00:14<01:14,  1.81s/it]\u001b[A\n",
            " 20% 10/50 [00:17<01:21,  2.05s/it]\u001b[A\n",
            " 22% 11/50 [00:19<01:17,  1.98s/it]\u001b[A\n",
            " 24% 12/50 [00:21<01:13,  1.93s/it]\u001b[A\n",
            " 26% 13/50 [00:22<01:08,  1.86s/it]\u001b[A\n",
            " 28% 14/50 [00:24<01:03,  1.76s/it]\u001b[A\n",
            " 30% 15/50 [00:25<00:59,  1.70s/it]\u001b[A\n",
            " 32% 16/50 [00:27<00:53,  1.57s/it]\u001b[A\n",
            " 34% 17/50 [00:28<00:52,  1.58s/it]\u001b[A\n",
            " 36% 18/50 [00:30<00:48,  1.52s/it]\u001b[A\n",
            " 38% 19/50 [00:32<00:51,  1.66s/it]\u001b[A\n",
            " 40% 20/50 [00:34<00:52,  1.74s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:50,  1.72s/it]\u001b[A\n",
            " 44% 22/50 [00:37<00:47,  1.69s/it]\u001b[A\n",
            " 46% 23/50 [00:39<00:46,  1.73s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:44,  1.70s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:40,  1.60s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:36,  1.53s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:34,  1.52s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:48<00:36,  1.75s/it]\u001b[A\n",
            " 60% 30/50 [00:50<00:34,  1.72s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:31,  1.64s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:30,  1.67s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:29,  1.71s/it]\u001b[A\n",
            " 68% 34/50 [00:57<00:26,  1.66s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:23,  1.56s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:22,  1.61s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:21,  1.68s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:20,  1.67s/it]\u001b[A\n",
            " 78% 39/50 [01:05<00:17,  1.62s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:16,  1.60s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.62s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.73s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.70s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.79s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:07,  1.79s/it]\u001b[A\n",
            " 94% 47/50 [01:19<00:05,  1.89s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.78s/it]\u001b[A\n",
            " 98% 49/50 [01:22<00:01,  1.69s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.0735743045806885, 'eval_rouge1': 36.5534, 'eval_rouge2': 13.8958, 'eval_rougeL': 28.4646, 'eval_rougeLsum': 28.4715, 'eval_gen_len': 26.5, 'epoch': 1.2254901960784315}\n",
            "\n",
            " 41% 12500/30600 [2:36:13<2:55:05,  1.72it/s]\n",
            "{'loss': 1.5973515625, 'learning_rate': 1.7254901960784314e-05, 'epoch': 1.2745098039215685}\n",
            " 42% 13000/30600 [2:41:05<2:54:54,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 17:39:40,537 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:39:40,537 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:39:40,537 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:41,  1.16it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:58,  1.27s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.37s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:00,  1.38s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:04,  1.50s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:05,  1.56s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:06,  1.62s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:06,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:03,  1.64s/it]\u001b[A\n",
            " 24% 12/50 [00:20<01:31,  2.41s/it]\u001b[A\n",
            " 26% 13/50 [00:22<01:25,  2.30s/it]\u001b[A\n",
            " 28% 14/50 [00:24<01:15,  2.11s/it]\u001b[A\n",
            " 30% 15/50 [00:26<01:09,  1.97s/it]\u001b[A\n",
            " 32% 16/50 [00:27<00:59,  1.74s/it]\u001b[A\n",
            " 34% 17/50 [00:29<00:57,  1.75s/it]\u001b[A\n",
            " 36% 18/50 [00:30<00:53,  1.69s/it]\u001b[A\n",
            " 38% 19/50 [00:32<00:51,  1.67s/it]\u001b[A\n",
            " 40% 20/50 [00:34<00:51,  1.72s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:49,  1.72s/it]\u001b[A\n",
            " 44% 22/50 [00:37<00:45,  1.64s/it]\u001b[A\n",
            " 46% 23/50 [00:39<00:46,  1.72s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:44,  1.71s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:40,  1.63s/it]\u001b[A\n",
            " 52% 26/50 [00:44<00:42,  1.77s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:38,  1.66s/it]\u001b[A\n",
            " 56% 28/50 [00:47<00:37,  1.71s/it]\u001b[A\n",
            " 58% 29/50 [00:49<00:37,  1.78s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:35,  1.80s/it]\u001b[A\n",
            " 62% 31/50 [00:53<00:33,  1.76s/it]\u001b[A\n",
            " 64% 32/50 [00:55<00:32,  1.79s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:31,  1.83s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:27,  1.74s/it]\u001b[A\n",
            " 70% 35/50 [01:00<00:26,  1.79s/it]\u001b[A\n",
            " 72% 36/50 [01:02<00:25,  1.82s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:23,  1.78s/it]\u001b[A\n",
            " 76% 38/50 [01:05<00:22,  1.85s/it]\u001b[A\n",
            " 78% 39/50 [01:07<00:18,  1.72s/it]\u001b[A\n",
            " 80% 40/50 [01:09<00:17,  1.80s/it]\u001b[A\n",
            " 82% 41/50 [01:11<00:16,  1.84s/it]\u001b[A\n",
            " 84% 42/50 [01:13<00:16,  2.01s/it]\u001b[A\n",
            " 86% 43/50 [01:15<00:12,  1.82s/it]\u001b[A\n",
            " 88% 44/50 [01:16<00:10,  1.80s/it]\u001b[A\n",
            " 90% 45/50 [01:18<00:08,  1.76s/it]\u001b[A\n",
            " 92% 46/50 [01:20<00:07,  1.80s/it]\u001b[A\n",
            " 94% 47/50 [01:22<00:05,  1.81s/it]\u001b[A\n",
            " 96% 48/50 [01:23<00:03,  1.70s/it]\u001b[A\n",
            " 98% 49/50 [01:25<00:01,  1.75s/it]\u001b[A\n",
            "100% 50/50 [01:27<00:00,  1.81s/it]\u001b[A\n",
            "{'eval_loss': 2.0811116695404053, 'eval_rouge1': 35.8204, 'eval_rouge2': 12.8193, 'eval_rougeL': 28.1653, 'eval_rougeLsum': 28.067, 'eval_gen_len': 26.8, 'epoch': 1.2745098039215685}\n",
            "\n",
            " 42% 13000/30600 [2:42:35<2:54:54,  1.68it/s]\n",
            "{'loss': 1.62373046875, 'learning_rate': 1.6764705882352943e-05, 'epoch': 1.3235294117647058}\n",
            " 44% 13500/30600 [2:47:30<2:42:08,  1.76it/s][INFO|trainer.py:1387] 2020-11-17 17:46:05,326 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:46:05,326 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:46:05,326 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:35,  1.34it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:47,  1.02s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:57,  1.24s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.36s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.40s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.47s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:01,  1.51s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:03,  1.58s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.58s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:04,  1.69s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:03,  1.70s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:59,  1.64s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:56,  1.63s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:50,  1.50s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:51,  1.56s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:48,  1.50s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:51,  1.66s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:52,  1.75s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:52,  1.79s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:46,  1.66s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:44,  1.65s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:52,  2.00s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:47,  1.90s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:41,  1.72s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:36,  1.61s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.66s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:40,  1.93s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:37,  1.85s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:34,  1.81s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:32,  1.79s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:31,  1.84s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.75s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:23,  1.60s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:25,  1.79s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:23,  1.78s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:21,  1.77s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:18,  1.68s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:17,  1.71s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:16,  1.81s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:15,  1.89s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:12,  1.75s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.75s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.71s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.67s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:04,  1.64s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.55s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.54s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.57s/it]\u001b[A\n",
            "{'eval_loss': 2.0391488075256348, 'eval_rouge1': 36.8903, 'eval_rouge2': 15.5071, 'eval_rougeL': 30.2312, 'eval_rougeLsum': 30.15, 'eval_gen_len': 25.2, 'epoch': 1.3235294117647058}\n",
            "\n",
            " 44% 13500/30600 [2:48:56<2:42:08,  1.76it/s]\n",
            "{'loss': 1.64728515625, 'learning_rate': 1.627450980392157e-05, 'epoch': 1.3725490196078431}\n",
            " 46% 14000/30600 [2:53:50<2:45:04,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 17:52:24,869 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:52:24,869 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:52:24,870 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:38,  1.25it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.17s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:11,  1.59s/it]\u001b[A\n",
            " 12% 6/50 [00:09<01:08,  1.57s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:03,  1.48s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:04,  1.54s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:01,  1.50s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:01,  1.54s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:02,  1.60s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.69s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:03,  1.72s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:00,  1.67s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:01,  1.75s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:53,  1.56s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:54,  1.64s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.59s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:50,  1.64s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.66s/it]\u001b[A\n",
            " 42% 21/50 [00:34<00:56,  1.94s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:49,  1.78s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:48,  1.79s/it]\u001b[A\n",
            " 48% 24/50 [00:39<00:43,  1.69s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:40,  1.60s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:35,  1.55s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:34,  1.58s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:37,  1.77s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:34,  1.72s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:32,  1.71s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:32,  1.80s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:31,  1.83s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.76s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:25,  1.72s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:24,  1.78s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:22,  1.73s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:20,  1.74s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:17,  1.62s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:16,  1.66s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:14,  1.62s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.65s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.66s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.60s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.67s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:04,  1.59s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.52s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.54s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.57s/it]\u001b[A\n",
            "{'eval_loss': 2.0405876636505127, 'eval_rouge1': 37.883, 'eval_rouge2': 15.3168, 'eval_rougeL': 30.1684, 'eval_rougeLsum': 30.0894, 'eval_gen_len': 25.5, 'epoch': 1.3725490196078431}\n",
            "\n",
            " 46% 14000/30600 [2:55:14<2:45:04,  1.68it/s]\n",
            "{'loss': 1.59481640625, 'learning_rate': 1.5784313725490197e-05, 'epoch': 1.4215686274509804}\n",
            " 47% 14500/30600 [3:00:08<2:38:08,  1.70it/s][INFO|trainer.py:1387] 2020-11-17 17:58:43,516 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 17:58:43,516 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 17:58:43,516 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.11it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.10s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:00,  1.31s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:07,  1.49s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:05,  1.49s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:04,  1.49s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.53s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.60s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:07,  1.69s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:02,  1.61s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:05,  1.71s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.78s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:01,  1.71s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:01,  1.75s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:54,  1.60s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:52,  1.58s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.58s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:50,  1.61s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:49,  1.69s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:43,  1.57s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:42,  1.56s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:39,  1.54s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.62s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:33,  1.60s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:32,  1.61s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:30,  1.60s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:29,  1.66s/it]\u001b[A\n",
            " 66% 33/50 [00:53<00:30,  1.80s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:27,  1.70s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:24,  1.60s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:23,  1.70s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:21,  1.62s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:20,  1.70s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:18,  1.64s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:15,  1.59s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:14,  1.57s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:13,  1.69s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:11,  1.60s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:09,  1.63s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:07,  1.60s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.61s/it]\u001b[A\n",
            " 94% 47/50 [01:15<00:04,  1.63s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.63s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.63s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.0521907806396484, 'eval_rouge1': 37.0514, 'eval_rouge2': 14.4372, 'eval_rougeL': 28.7065, 'eval_rougeLsum': 28.8457, 'eval_gen_len': 25.8, 'epoch': 1.4215686274509804}\n",
            "\n",
            " 47% 14500/30600 [3:01:32<2:38:08,  1.70it/s]\n",
            "{'loss': 1.60875390625, 'learning_rate': 1.5294117647058822e-05, 'epoch': 1.4705882352941178}\n",
            " 49% 15000/30600 [3:06:25<2:34:20,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 18:05:00,116 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:05:00,116 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:05:00,116 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:35,  1.37it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:46,  1.01it/s]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.23s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.43s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:05,  1.51s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:05,  1.56s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:03,  1.55s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.71s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:07,  1.73s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:02,  1.64s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:04,  1.73s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.69s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:04,  1.83s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:54,  1.59s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:51,  1.57s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:47,  1.49s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.57s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:47,  1.65s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:43,  1.56s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:47,  1.76s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:45,  1.75s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.58s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:34,  1.57s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:36,  1.75s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:34,  1.75s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:31,  1.66s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:30,  1.71s/it]\u001b[A\n",
            " 66% 33/50 [00:53<00:29,  1.76s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:26,  1.65s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:24,  1.62s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:23,  1.68s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:20,  1.69s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:18,  1.65s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:16,  1.60s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:13,  1.50s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:13,  1.63s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:10,  1.53s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:09,  1.60s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:07,  1.57s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.63s/it]\u001b[A\n",
            " 94% 47/50 [01:15<00:04,  1.62s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.57s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.79s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.74s/it]\u001b[A\n",
            "{'eval_loss': 2.038275718688965, 'eval_rouge1': 36.1121, 'eval_rouge2': 13.5518, 'eval_rougeL': 28.5097, 'eval_rougeLsum': 28.7222, 'eval_gen_len': 25.3, 'epoch': 1.4705882352941178}\n",
            "\n",
            " 49% 15000/30600 [3:07:49<2:34:20,  1.68it/s]\n",
            "{'loss': 1.63239453125, 'learning_rate': 1.4803921568627453e-05, 'epoch': 1.5196078431372548}\n",
            " 51% 15500/30600 [3:12:41<2:32:26,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 18:11:16,085 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:11:16,086 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:11:16,086 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:41,  1.15it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:05,  1.53s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.53s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.57s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:04,  1.61s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:02,  1.60s/it]\u001b[A\n",
            " 24% 12/50 [00:18<00:59,  1.57s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:00,  1.65s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:58,  1.64s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:57,  1.63s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:51,  1.52s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.50s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:48,  1.51s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:47,  1.53s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:50,  1.73s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:42,  1.57s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:42,  1.65s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:40,  1.70s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:37,  1.65s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.66s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:36,  1.76s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:33,  1.68s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:31,  1.65s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:30,  1.71s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:29,  1.72s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:27,  1.74s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:26,  1.76s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:25,  1.83s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.74s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:20,  1.70s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.61s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:16,  1.64s/it]\u001b[A\n",
            " 82% 41/50 [01:06<00:16,  1.81s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:14,  1.84s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:12,  1.72s/it]\u001b[A\n",
            " 88% 44/50 [01:11<00:10,  1.72s/it]\u001b[A\n",
            " 90% 45/50 [01:13<00:08,  1.69s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.72s/it]\u001b[A\n",
            " 94% 47/50 [01:16<00:05,  1.69s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.61s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.68s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.61s/it]\u001b[A\n",
            "{'eval_loss': 2.0522823333740234, 'eval_rouge1': 37.2445, 'eval_rouge2': 14.6289, 'eval_rougeL': 29.5932, 'eval_rougeLsum': 29.6322, 'eval_gen_len': 26.0, 'epoch': 1.5196078431372548}\n",
            "\n",
            " 51% 15500/30600 [3:14:05<2:32:26,  1.65it/s]\n",
            "{'loss': 1.58833203125, 'learning_rate': 1.431372549019608e-05, 'epoch': 1.5686274509803921}\n",
            " 52% 16000/30600 [3:18:58<2:20:16,  1.73it/s][INFO|trainer.py:1387] 2020-11-17 18:17:33,590 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:17:33,590 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:17:33,591 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:41,  1.15it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:55,  1.18s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:00,  1.31s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.41s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.40s/it]\u001b[A\n",
            " 16% 8/50 [00:11<00:59,  1.41s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.47s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:02,  1.57s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.57s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:01,  1.62s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:00,  1.63s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:59,  1.65s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:55,  1.60s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:49,  1.47s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.49s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:48,  1.50s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:46,  1.49s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:45,  1.53s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:45,  1.57s/it]\u001b[A\n",
            " 44% 22/50 [00:32<00:41,  1.49s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:42,  1.58s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:41,  1.60s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:40,  1.61s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:37,  1.57s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:35,  1.54s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:33,  1.53s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:34,  1.62s/it]\u001b[A\n",
            " 60% 30/50 [00:45<00:31,  1.57s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:31,  1.64s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:30,  1.72s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:29,  1.72s/it]\u001b[A\n",
            " 68% 34/50 [00:52<00:26,  1.68s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:25,  1.68s/it]\u001b[A\n",
            " 72% 36/50 [00:56<00:24,  1.77s/it]\u001b[A\n",
            " 74% 37/50 [00:57<00:22,  1.70s/it]\u001b[A\n",
            " 76% 38/50 [00:59<00:20,  1.74s/it]\u001b[A\n",
            " 78% 39/50 [01:01<00:18,  1.66s/it]\u001b[A\n",
            " 80% 40/50 [01:02<00:16,  1.61s/it]\u001b[A\n",
            " 82% 41/50 [01:04<00:14,  1.56s/it]\u001b[A\n",
            " 84% 42/50 [01:06<00:13,  1.67s/it]\u001b[A\n",
            " 86% 43/50 [01:07<00:10,  1.55s/it]\u001b[A\n",
            " 88% 44/50 [01:09<00:09,  1.60s/it]\u001b[A\n",
            " 90% 45/50 [01:10<00:07,  1.58s/it]\u001b[A\n",
            " 92% 46/50 [01:12<00:06,  1.62s/it]\u001b[A\n",
            " 94% 47/50 [01:14<00:05,  1.68s/it]\u001b[A\n",
            " 96% 48/50 [01:15<00:03,  1.60s/it]\u001b[A\n",
            " 98% 49/50 [01:17<00:01,  1.74s/it]\u001b[A\n",
            "100% 50/50 [01:19<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.0360426902770996, 'eval_rouge1': 37.8722, 'eval_rouge2': 14.5289, 'eval_rougeL': 29.6948, 'eval_rougeLsum': 29.696, 'eval_gen_len': 24.8, 'epoch': 1.5686274509803921}\n",
            "\n",
            " 52% 16000/30600 [3:20:20<2:20:16,  1.73it/s]\n",
            "{'loss': 1.5743671875, 'learning_rate': 1.3823529411764705e-05, 'epoch': 1.6176470588235294}\n",
            " 54% 16500/30600 [3:25:17<2:11:52,  1.78it/s][INFO|trainer.py:1387] 2020-11-17 18:23:52,173 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:23:52,173 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:23:52,173 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.23it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.11s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:57,  1.25s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.41s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.46s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.45s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.48s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:02,  1.57s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:00,  1.56s/it]\u001b[A\n",
            " 24% 12/50 [00:17<00:59,  1.56s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:58,  1.58s/it]\u001b[A\n",
            " 28% 14/50 [00:20<00:57,  1.59s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:53,  1.53s/it]\u001b[A\n",
            " 32% 16/50 [00:23<00:48,  1.44s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:49,  1.50s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:48,  1.52s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:50,  1.62s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:48,  1.62s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:48,  1.68s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:41,  1.54s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:40,  1.57s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:38,  1.54s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:37,  1.62s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:36,  1.64s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:34,  1.65s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:32,  1.61s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:31,  1.63s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:31,  1.73s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:30,  1.78s/it]\u001b[A\n",
            " 68% 34/50 [00:53<00:29,  1.82s/it]\u001b[A\n",
            " 70% 35/50 [00:55<00:26,  1.78s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:25,  1.84s/it]\u001b[A\n",
            " 74% 37/50 [00:58<00:23,  1.79s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:20,  1.67s/it]\u001b[A\n",
            " 78% 39/50 [01:01<00:17,  1.60s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:16,  1.68s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:14,  1.67s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:14,  1.77s/it]\u001b[A\n",
            " 86% 43/50 [01:08<00:11,  1.61s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:08,  1.65s/it]\u001b[A\n",
            " 92% 46/50 [01:13<00:06,  1.68s/it]\u001b[A\n",
            " 94% 47/50 [01:15<00:05,  1.68s/it]\u001b[A\n",
            " 96% 48/50 [01:16<00:03,  1.61s/it]\u001b[A\n",
            " 98% 49/50 [01:18<00:01,  1.64s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.62s/it]\u001b[A\n",
            "{'eval_loss': 2.0181193351745605, 'eval_rouge1': 38.7778, 'eval_rouge2': 14.8907, 'eval_rougeL': 29.9757, 'eval_rougeLsum': 29.9318, 'eval_gen_len': 25.0, 'epoch': 1.6176470588235294}\n",
            "\n",
            " 54% 16500/30600 [3:26:39<2:11:52,  1.78it/s]\n",
            "{'loss': 1.593515625, 'learning_rate': 1.3333333333333333e-05, 'epoch': 1.6666666666666665}\n",
            " 56% 17000/30600 [3:31:33<2:23:58,  1.57it/s][INFO|trainer.py:1387] 2020-11-17 18:30:08,527 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:30:08,527 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:30:08,527 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.10it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:58,  1.24s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:03,  1.39s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:06,  1.47s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.46s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:01,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.48s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:03,  1.55s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:03,  1.59s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.59s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:07,  1.77s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:04,  1.74s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.67s/it]\u001b[A\n",
            " 30% 15/50 [00:23<01:00,  1.72s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:54,  1.59s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:52,  1.60s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:53,  1.69s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:52,  1.69s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:54,  1.81s/it]\u001b[A\n",
            " 42% 21/50 [00:34<00:52,  1.82s/it]\u001b[A\n",
            " 44% 22/50 [00:37<01:03,  2.28s/it]\u001b[A\n",
            " 46% 23/50 [00:39<00:56,  2.08s/it]\u001b[A\n",
            " 48% 24/50 [00:41<00:53,  2.04s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:49,  1.97s/it]\u001b[A\n",
            " 52% 26/50 [00:44<00:44,  1.87s/it]\u001b[A\n",
            " 54% 27/50 [00:46<00:40,  1.77s/it]\u001b[A\n",
            " 56% 28/50 [00:47<00:37,  1.68s/it]\u001b[A\n",
            " 58% 29/50 [00:49<00:38,  1.85s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:35,  1.79s/it]\u001b[A\n",
            " 62% 31/50 [00:53<00:32,  1.70s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:31,  1.73s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:29,  1.72s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:26,  1.65s/it]\u001b[A\n",
            " 70% 35/50 [00:59<00:24,  1.61s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:24,  1.77s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:22,  1.77s/it]\u001b[A\n",
            " 76% 38/50 [01:04<00:20,  1.71s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:17,  1.58s/it]\u001b[A\n",
            " 80% 40/50 [01:07<00:16,  1.61s/it]\u001b[A\n",
            " 82% 41/50 [01:09<00:13,  1.46s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.63s/it]\u001b[A\n",
            " 86% 43/50 [01:12<00:11,  1.57s/it]\u001b[A\n",
            " 88% 44/50 [01:14<00:09,  1.61s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:08,  1.66s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:06,  1.65s/it]\u001b[A\n",
            " 94% 47/50 [01:19<00:05,  1.79s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.68s/it]\u001b[A\n",
            " 98% 49/50 [01:22<00:01,  1.68s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.65s/it]\u001b[A\n",
            "{'eval_loss': 2.0435397624969482, 'eval_rouge1': 36.995, 'eval_rouge2': 14.5014, 'eval_rougeL': 29.0721, 'eval_rougeLsum': 29.3051, 'eval_gen_len': 25.2, 'epoch': 1.6666666666666665}\n",
            "\n",
            " 56% 17000/30600 [3:33:00<2:23:58,  1.57it/s]\n",
            "{'loss': 1.6182890625, 'learning_rate': 1.284313725490196e-05, 'epoch': 1.715686274509804}\n",
            " 57% 17500/30600 [3:37:59<1:52:23,  1.94it/s][INFO|trainer.py:1387] 2020-11-17 18:36:34,504 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:36:34,504 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:36:34,505 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.20it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.17s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:08,  1.53s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:06,  1.52s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:09,  1.61s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:08,  1.64s/it]\u001b[A\n",
            " 18% 9/50 [00:14<01:10,  1.72s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.71s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:06,  1.72s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:03,  1.67s/it]\u001b[A\n",
            " 26% 13/50 [00:21<01:06,  1.81s/it]\u001b[A\n",
            " 28% 14/50 [00:23<01:05,  1.83s/it]\u001b[A\n",
            " 30% 15/50 [00:25<01:05,  1.87s/it]\u001b[A\n",
            " 32% 16/50 [00:26<00:57,  1.70s/it]\u001b[A\n",
            " 34% 17/50 [00:28<00:55,  1.67s/it]\u001b[A\n",
            " 36% 18/50 [00:29<00:52,  1.65s/it]\u001b[A\n",
            " 38% 19/50 [00:31<00:55,  1.80s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:53,  1.78s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:51,  1.79s/it]\u001b[A\n",
            " 44% 22/50 [00:37<00:49,  1.75s/it]\u001b[A\n",
            " 46% 23/50 [00:39<00:48,  1.80s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:45,  1.77s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:46,  1.84s/it]\u001b[A\n",
            " 52% 26/50 [00:44<00:42,  1.76s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:38,  1.69s/it]\u001b[A\n",
            " 56% 28/50 [00:47<00:37,  1.70s/it]\u001b[A\n",
            " 58% 29/50 [00:49<00:38,  1.81s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:35,  1.77s/it]\u001b[A\n",
            " 62% 31/50 [00:52<00:32,  1.73s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:31,  1.77s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:29,  1.75s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:27,  1.72s/it]\u001b[A\n",
            " 70% 35/50 [00:59<00:24,  1.63s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:23,  1.71s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:22,  1.75s/it]\u001b[A\n",
            " 76% 38/50 [01:05<00:20,  1.75s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:18,  1.65s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:09<00:14,  1.59s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.66s/it]\u001b[A\n",
            " 86% 43/50 [01:12<00:11,  1.58s/it]\u001b[A\n",
            " 88% 44/50 [01:14<00:09,  1.64s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:08,  1.64s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:19<00:04,  1.66s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.69s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.72s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.73s/it]\u001b[A\n",
            "{'eval_loss': 2.0372040271759033, 'eval_rouge1': 38.2601, 'eval_rouge2': 15.6273, 'eval_rougeL': 29.7891, 'eval_rougeLsum': 29.8039, 'eval_gen_len': 26.6, 'epoch': 1.715686274509804}\n",
            "\n",
            " 57% 17500/30600 [3:39:27<1:52:23,  1.94it/s]\n",
            "{'loss': 1.5914375, 'learning_rate': 1.2352941176470587e-05, 'epoch': 1.7647058823529411}\n",
            " 59% 18000/30600 [3:44:26<2:04:10,  1.69it/s][INFO|trainer.py:1387] 2020-11-17 18:43:01,402 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:43:01,402 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:43:01,402 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.23it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.15s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:03,  1.38s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:05,  1.45s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:05,  1.48s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:03,  1.48s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:04,  1.54s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:06,  1.61s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:04,  1.65s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:01,  1.62s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:03,  1.71s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:00,  1.69s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:54,  1.55s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:49,  1.45s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:49,  1.51s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:46,  1.46s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:46,  1.51s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:45,  1.51s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:49,  1.71s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.62s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:43,  1.60s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:44,  1.70s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:44,  1.79s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:40,  1.70s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.57s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:32,  1.57s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:33,  1.70s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:30,  1.60s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:30,  1.69s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:28,  1.67s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:26,  1.64s/it]\u001b[A\n",
            " 70% 35/50 [00:55<00:25,  1.69s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:24,  1.78s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.71s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:20,  1.67s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.61s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:15,  1.58s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:13,  1.48s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:12,  1.58s/it]\u001b[A\n",
            " 86% 43/50 [01:08<00:10,  1.51s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:09,  1.57s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:07,  1.54s/it]\u001b[A\n",
            " 92% 46/50 [01:13<00:06,  1.57s/it]\u001b[A\n",
            " 94% 47/50 [01:14<00:04,  1.64s/it]\u001b[A\n",
            " 96% 48/50 [01:16<00:03,  1.59s/it]\u001b[A\n",
            " 98% 49/50 [01:18<00:01,  1.60s/it]\u001b[A\n",
            "100% 50/50 [01:19<00:00,  1.51s/it]\u001b[A\n",
            "{'eval_loss': 2.0283091068267822, 'eval_rouge1': 38.4068, 'eval_rouge2': 16.0489, 'eval_rougeL': 31.0575, 'eval_rougeLsum': 31.1018, 'eval_gen_len': 24.5, 'epoch': 1.7647058823529411}\n",
            "\n",
            " 59% 18000/30600 [3:45:48<2:04:10,  1.69it/s]\n",
            "{'loss': 1.599015625, 'learning_rate': 1.1862745098039216e-05, 'epoch': 1.8137254901960784}\n",
            " 60% 18500/30600 [3:50:45<1:51:16,  1.81it/s][INFO|trainer.py:1387] 2020-11-17 18:49:19,697 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:49:19,698 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:49:19,698 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.20it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:00,  1.31s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:09,  1.54s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:05,  1.49s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:09,  1.61s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:08,  1.62s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:08,  1.66s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.72s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:09,  1.77s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:08,  1.80s/it]\u001b[A\n",
            " 26% 13/50 [00:21<01:05,  1.77s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:01,  1.72s/it]\u001b[A\n",
            " 30% 15/50 [00:24<00:58,  1.67s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:51,  1.52s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:52,  1.59s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:49,  1.54s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:49,  1.58s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:47,  1.57s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:45,  1.63s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:45,  1.69s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:44,  1.70s/it]\u001b[A\n",
            " 50% 25/50 [00:41<00:46,  1.86s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:41,  1.73s/it]\u001b[A\n",
            " 54% 27/50 [00:44<00:38,  1.69s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:38,  1.76s/it]\u001b[A\n",
            " 58% 29/50 [00:48<00:38,  1.82s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:36,  1.80s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:34,  1.81s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:33,  1.88s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:30,  1.81s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:27,  1.70s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:23,  1.59s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:25,  1.79s/it]\u001b[A\n",
            " 74% 37/50 [01:02<00:22,  1.73s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:20,  1.72s/it]\u001b[A\n",
            " 78% 39/50 [01:05<00:17,  1.63s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:16,  1.62s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.58s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.64s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.60s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.70s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.70s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:06,  1.74s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.77s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.69s/it]\u001b[A\n",
            " 98% 49/50 [01:22<00:01,  1.88s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.78s/it]\u001b[A\n",
            "{'eval_loss': 2.0344629287719727, 'eval_rouge1': 37.4454, 'eval_rouge2': 14.589, 'eval_rougeL': 29.4759, 'eval_rougeLsum': 29.439, 'eval_gen_len': 25.7, 'epoch': 1.8137254901960784}\n",
            "\n",
            " 60% 18500/30600 [3:52:11<1:51:16,  1.81it/s]\n",
            "{'loss': 1.587875, 'learning_rate': 1.1372549019607843e-05, 'epoch': 1.8627450980392157}\n",
            " 62% 19000/30600 [3:57:06<1:51:05,  1.74it/s][INFO|trainer.py:1387] 2020-11-17 18:55:40,798 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 18:55:40,798 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 18:55:40,798 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.12it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.16s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.35s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:05,  1.45s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:07,  1.54s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:07,  1.57s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:07,  1.60s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:08,  1.67s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.71s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:10,  1.81s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:10,  1.87s/it]\u001b[A\n",
            " 26% 13/50 [00:21<01:10,  1.90s/it]\u001b[A\n",
            " 28% 14/50 [00:23<01:07,  1.87s/it]\u001b[A\n",
            " 30% 15/50 [00:25<01:01,  1.77s/it]\u001b[A\n",
            " 32% 16/50 [00:26<00:54,  1.59s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:52,  1.60s/it]\u001b[A\n",
            " 36% 18/50 [00:29<00:50,  1.57s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:48,  1.55s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:52,  1.77s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:52,  1.79s/it]\u001b[A\n",
            " 44% 22/50 [00:36<00:47,  1.71s/it]\u001b[A\n",
            " 46% 23/50 [00:38<00:48,  1.81s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:47,  1.81s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:46,  1.85s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:42,  1.78s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:38,  1.68s/it]\u001b[A\n",
            " 56% 28/50 [00:47<00:38,  1.76s/it]\u001b[A\n",
            " 58% 29/50 [00:49<00:39,  1.86s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:37,  1.87s/it]\u001b[A\n",
            " 62% 31/50 [00:53<00:36,  1.90s/it]\u001b[A\n",
            " 64% 32/50 [00:55<00:34,  1.91s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:30,  1.81s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:27,  1.72s/it]\u001b[A\n",
            " 70% 35/50 [00:59<00:24,  1.67s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:24,  1.72s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:21,  1.68s/it]\u001b[A\n",
            " 76% 38/50 [01:04<00:20,  1.68s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:17,  1.61s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:16,  1.67s/it]\u001b[A\n",
            " 82% 41/50 [01:10<00:15,  1.70s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.74s/it]\u001b[A\n",
            " 86% 43/50 [01:13<00:11,  1.70s/it]\u001b[A\n",
            " 88% 44/50 [01:15<00:10,  1.73s/it]\u001b[A\n",
            " 90% 45/50 [01:17<00:08,  1.74s/it]\u001b[A\n",
            " 92% 46/50 [01:18<00:06,  1.71s/it]\u001b[A\n",
            " 94% 47/50 [01:20<00:05,  1.75s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.67s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.69s/it]\u001b[A\n",
            "100% 50/50 [01:25<00:00,  1.66s/it]\u001b[A\n",
            "{'eval_loss': 2.0370054244995117, 'eval_rouge1': 37.1852, 'eval_rouge2': 14.5318, 'eval_rougeL': 29.3558, 'eval_rougeLsum': 29.4894, 'eval_gen_len': 26.0, 'epoch': 1.8627450980392157}\n",
            "\n",
            " 62% 19000/30600 [3:58:33<1:51:05,  1.74it/s]\n",
            "{'loss': 1.5997734375, 'learning_rate': 1.0882352941176471e-05, 'epoch': 1.9117647058823528}\n",
            " 64% 19500/30600 [4:03:29<1:51:22,  1.66it/s][INFO|trainer.py:1387] 2020-11-17 19:02:04,382 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:02:04,382 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:02:04,382 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.19it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:52,  1.12s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:58,  1.27s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.39s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.50s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:03,  1.54s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:02,  1.57s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.58s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:00,  1.59s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:59,  1.62s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.68s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:56,  1.62s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:51,  1.51s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:52,  1.58s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:51,  1.62s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.57s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:47,  1.60s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:48,  1.69s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.64s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:46,  1.74s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:44,  1.71s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:42,  1.72s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:40,  1.70s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:38,  1.67s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:37,  1.72s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:38,  1.82s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:34,  1.70s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:32,  1.69s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:31,  1.72s/it]\u001b[A\n",
            " 66% 33/50 [00:53<00:28,  1.70s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:26,  1.63s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:23,  1.58s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:23,  1.67s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:21,  1.64s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:19,  1.61s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:16,  1.51s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:15,  1.59s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:14,  1.59s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:13,  1.67s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:12,  1.72s/it]\u001b[A\n",
            " 88% 44/50 [01:11<00:10,  1.73s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:08,  1.74s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.71s/it]\u001b[A\n",
            " 94% 47/50 [01:16<00:05,  1.71s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.65s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.55s/it]\u001b[A\n",
            "{'eval_loss': 2.0271506309509277, 'eval_rouge1': 38.2583, 'eval_rouge2': 15.6179, 'eval_rougeL': 30.1819, 'eval_rougeLsum': 30.1959, 'eval_gen_len': 25.0, 'epoch': 1.9117647058823528}\n",
            "\n",
            " 64% 19500/30600 [4:04:52<1:51:22,  1.66it/s]\n",
            "{'loss': 1.5896171875, 'learning_rate': 1.0392156862745098e-05, 'epoch': 1.9607843137254903}\n",
            " 65% 20000/30600 [4:09:46<1:46:09,  1.66it/s][INFO|trainer.py:1387] 2020-11-17 19:08:21,541 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:08:21,542 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:08:21,542 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.11it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:52,  1.13s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:03,  1.37s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:07,  1.51s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:07,  1.54s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:08,  1.60s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:06,  1.58s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:06,  1.63s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:04,  1.60s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:03,  1.62s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.71s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.77s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:01,  1.70s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:00,  1.73s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:54,  1.62s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:55,  1.69s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:51,  1.61s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:52,  1.70s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:48,  1.63s/it]\u001b[A\n",
            " 42% 21/50 [00:34<00:49,  1.70s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:45,  1.61s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:44,  1.66s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:42,  1.63s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:39,  1.59s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:36,  1.59s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:36,  1.64s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:35,  1.68s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:33,  1.70s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:34,  1.82s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:32,  1.81s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:31,  1.85s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:27,  1.74s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:24,  1.64s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:23,  1.65s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.64s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:19,  1.62s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:16,  1.53s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:15,  1.55s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:14,  1.60s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:13,  1.70s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:13<00:08,  1.69s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.66s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:04,  1.61s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.58s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.61s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.52s/it]\u001b[A\n",
            "{'eval_loss': 2.0170342922210693, 'eval_rouge1': 38.2149, 'eval_rouge2': 15.4263, 'eval_rougeL': 29.7021, 'eval_rougeLsum': 29.6982, 'eval_gen_len': 24.9, 'epoch': 1.9607843137254903}\n",
            "\n",
            " 65% 20000/30600 [4:11:10<1:46:09,  1.66it/s]\n",
            "                                   \u001b[A[INFO|trainer.py:1222] 2020-11-17 19:09:45,505 >> Saving model checkpoint to prophetnet_large_uncased_xsum/checkpoint-20000\n",
            "[INFO|configuration_utils.py:282] 2020-11-17 19:09:45,507 >> Configuration saved in prophetnet_large_uncased_xsum/checkpoint-20000/config.json\n",
            "[INFO|modeling_utils.py:740] 2020-11-17 19:09:53,309 >> Model weights saved in prophetnet_large_uncased_xsum/checkpoint-20000/pytorch_model.bin\n",
            "{'loss': 1.4963984375, 'learning_rate': 9.901960784313725e-06, 'epoch': 2.0098039215686274}\n",
            " 67% 20500/30600 [4:17:33<1:38:14,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 19:16:07,768 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:16:07,768 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:16:07,768 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:44,  1.09it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:51,  1.10s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:58,  1.27s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:04,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:05,  1.49s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:03,  1.47s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.50s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.60s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.63s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:02,  1.60s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:03,  1.66s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:01,  1.67s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.68s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:59,  1.69s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:53,  1.56s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:53,  1.61s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:49,  1.55s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:47,  1.53s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:49,  1.64s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:52,  1.82s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:47,  1.71s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:45,  1.69s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:43,  1.69s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:39,  1.56s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:37,  1.55s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:34,  1.52s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:35,  1.60s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:37,  1.78s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:35,  1.79s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:32,  1.72s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:32,  1.81s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:32,  1.91s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:28,  1.81s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:25,  1.72s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:24,  1.75s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:22,  1.76s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:21,  1.78s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:18,  1.65s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:16,  1.62s/it]\u001b[A\n",
            " 82% 41/50 [01:06<00:14,  1.60s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.68s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.72s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.75s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.69s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.73s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.62s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.64s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.60s/it]\u001b[A\n",
            "{'eval_loss': 2.0448899269104004, 'eval_rouge1': 38.8662, 'eval_rouge2': 15.3382, 'eval_rougeL': 30.4239, 'eval_rougeLsum': 30.4631, 'eval_gen_len': 25.4, 'epoch': 2.0098039215686274}\n",
            "\n",
            " 67% 20500/30600 [4:18:57<1:38:14,  1.71it/s]\n",
            "{'loss': 1.1151953125, 'learning_rate': 9.411764705882354e-06, 'epoch': 2.0588235294117645}\n",
            " 69% 21000/30600 [4:23:56<1:37:04,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 19:22:31,498 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:22:31,498 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:22:31,498 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.21it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:49,  1.04s/it]\u001b[A\n",
            "  8% 4/50 [00:04<00:56,  1.23s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:01,  1.36s/it]\u001b[A\n",
            " 12% 6/50 [00:07<00:59,  1.36s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.39s/it]\u001b[A\n",
            " 16% 8/50 [00:10<01:01,  1.46s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.47s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:04,  1.61s/it]\u001b[A\n",
            " 22% 11/50 [00:15<01:01,  1.57s/it]\u001b[A\n",
            " 24% 12/50 [00:17<01:03,  1.66s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:02,  1.69s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.67s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:58,  1.66s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:52,  1.54s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:53,  1.61s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:49,  1.56s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:50,  1.63s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:48,  1.63s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:48,  1.69s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:41,  1.55s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:41,  1.58s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:38,  1.54s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:36,  1.53s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:34,  1.49s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:34,  1.56s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:35,  1.71s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:35,  1.77s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:32,  1.72s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:32,  1.79s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:32,  1.93s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:28,  1.80s/it]\u001b[A\n",
            " 70% 35/50 [00:55<00:25,  1.70s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:24,  1.75s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.73s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:20,  1.67s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.59s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:21,  2.15s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:18,  2.09s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:16,  2.02s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:13,  1.90s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.77s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:09,  1.81s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.75s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.77s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.76s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.73s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.65s/it]\u001b[A\n",
            "{'eval_loss': 2.061039686203003, 'eval_rouge1': 37.4909, 'eval_rouge2': 14.6998, 'eval_rougeL': 29.5198, 'eval_rougeLsum': 29.6032, 'eval_gen_len': 25.0, 'epoch': 2.0588235294117645}\n",
            "\n",
            " 69% 21000/30600 [4:25:22<1:37:04,  1.65it/s]\n",
            "{'loss': 1.103859375, 'learning_rate': 8.92156862745098e-06, 'epoch': 2.107843137254902}\n",
            " 70% 21500/30600 [4:30:20<1:25:14,  1.78it/s][INFO|trainer.py:1387] 2020-11-17 19:28:55,333 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:28:55,334 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:28:55,334 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:41,  1.16it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:50,  1.08s/it]\u001b[A\n",
            "  8% 4/50 [00:05<00:58,  1.28s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:02,  1.39s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.45s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.50s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.56s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:06,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:04,  1.65s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.69s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:04,  1.75s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:04,  1.79s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:59,  1.69s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:53,  1.57s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:55,  1.67s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:52,  1.63s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.58s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:48,  1.66s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:43,  1.57s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:43,  1.60s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:42,  1.63s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:43,  1.73s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:42,  1.78s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:38,  1.67s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.68s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:38,  1.81s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:36,  1.85s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:33,  1.78s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:31,  1.75s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:32,  1.93s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.79s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:25,  1.67s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:23,  1.68s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.67s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:19,  1.60s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:16,  1.52s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:15,  1.55s/it]\u001b[A\n",
            " 82% 41/50 [01:06<00:14,  1.57s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:12,  1.73s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.76s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:12,  2.56s/it]\u001b[A\n",
            " 92% 46/50 [01:18<00:09,  2.27s/it]\u001b[A\n",
            " 94% 47/50 [01:20<00:06,  2.12s/it]\u001b[A\n",
            " 96% 48/50 [01:22<00:04,  2.05s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.96s/it]\u001b[A\n",
            "100% 50/50 [01:25<00:00,  1.84s/it]\u001b[A\n",
            "{'eval_loss': 2.0771539211273193, 'eval_rouge1': 39.026, 'eval_rouge2': 15.7849, 'eval_rougeL': 30.6598, 'eval_rougeLsum': 30.5898, 'eval_gen_len': 25.3, 'epoch': 2.107843137254902}\n",
            "\n",
            " 70% 21500/30600 [4:31:48<1:25:14,  1.78it/s]\n",
            "{'loss': 1.118546875, 'learning_rate': 8.431372549019608e-06, 'epoch': 2.156862745098039}\n",
            " 72% 22000/30600 [4:36:47<1:20:21,  1.78it/s][INFO|trainer.py:1387] 2020-11-17 19:35:22,316 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:35:22,316 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:35:22,316 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:40,  1.19it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.15s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:00,  1.31s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:04,  1.42s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.47s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:06,  1.54s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:05,  1.56s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:06,  1.61s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:07,  1.68s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.59s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.70s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:00,  1.65s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:57,  1.60s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:54,  1.56s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:50,  1.47s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:50,  1.54s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:48,  1.51s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.58s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:47,  1.65s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.62s/it]\u001b[A\n",
            " 46% 23/50 [00:38<01:03,  2.34s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:58,  2.23s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:52,  2.08s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:47,  1.97s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:42,  1.83s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:38,  1.74s/it]\u001b[A\n",
            " 58% 29/50 [00:48<00:36,  1.72s/it]\u001b[A\n",
            " 60% 30/50 [00:50<00:35,  1.79s/it]\u001b[A\n",
            " 62% 31/50 [00:52<00:34,  1.83s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:33,  1.88s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:33,  1.99s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:29,  1.84s/it]\u001b[A\n",
            " 70% 35/50 [00:59<00:25,  1.72s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:24,  1.73s/it]\u001b[A\n",
            " 74% 37/50 [01:02<00:21,  1.63s/it]\u001b[A\n",
            " 76% 38/50 [01:04<00:21,  1.76s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:18,  1.64s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:19,  1.91s/it]\u001b[A\n",
            " 82% 41/50 [01:11<00:20,  2.29s/it]\u001b[A\n",
            " 84% 42/50 [01:13<00:17,  2.14s/it]\u001b[A\n",
            " 86% 43/50 [01:15<00:13,  1.97s/it]\u001b[A\n",
            " 88% 44/50 [01:16<00:11,  1.86s/it]\u001b[A\n",
            " 90% 45/50 [01:18<00:08,  1.73s/it]\u001b[A\n",
            " 92% 46/50 [01:19<00:06,  1.70s/it]\u001b[A\n",
            " 94% 47/50 [01:21<00:05,  1.68s/it]\u001b[A\n",
            " 96% 48/50 [01:23<00:03,  1.71s/it]\u001b[A\n",
            " 98% 49/50 [01:25<00:01,  1.73s/it]\u001b[A\n",
            "100% 50/50 [01:26<00:00,  1.66s/it]\u001b[A\n",
            "{'eval_loss': 2.0674939155578613, 'eval_rouge1': 36.3269, 'eval_rouge2': 14.6845, 'eval_rougeL': 29.0308, 'eval_rougeLsum': 28.9445, 'eval_gen_len': 25.3, 'epoch': 2.156862745098039}\n",
            "\n",
            " 72% 22000/30600 [4:38:16<1:20:21,  1.78it/s]\n",
            "{'loss': 1.1283046875, 'learning_rate': 7.941176470588236e-06, 'epoch': 2.2058823529411766}\n",
            " 74% 22500/30600 [4:43:10<1:18:52,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 19:41:45,092 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:41:45,092 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:41:45,092 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:46,  1.04it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:58,  1.23s/it]\u001b[A\n",
            "  8% 4/50 [00:06<01:10,  1.53s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:11,  1.58s/it]\u001b[A\n",
            " 12% 6/50 [00:11<01:32,  2.10s/it]\u001b[A\n",
            " 14% 7/50 [00:13<01:29,  2.09s/it]\u001b[A\n",
            " 16% 8/50 [00:14<01:20,  1.92s/it]\u001b[A\n",
            " 18% 9/50 [00:16<01:16,  1.87s/it]\u001b[A\n",
            " 20% 10/50 [00:18<01:12,  1.82s/it]\u001b[A\n",
            " 22% 11/50 [00:19<01:10,  1.80s/it]\u001b[A\n",
            " 24% 12/50 [00:21<01:11,  1.89s/it]\u001b[A\n",
            " 26% 13/50 [00:23<01:06,  1.80s/it]\u001b[A\n",
            " 28% 14/50 [00:25<01:02,  1.74s/it]\u001b[A\n",
            " 30% 15/50 [00:28<01:13,  2.11s/it]\u001b[A\n",
            " 32% 16/50 [00:29<01:05,  1.92s/it]\u001b[A\n",
            " 34% 17/50 [00:31<01:03,  1.92s/it]\u001b[A\n",
            " 36% 18/50 [00:32<00:56,  1.78s/it]\u001b[A\n",
            " 38% 19/50 [00:34<00:57,  1.86s/it]\u001b[A\n",
            " 40% 20/50 [00:36<00:53,  1.78s/it]\u001b[A\n",
            " 42% 21/50 [00:38<00:52,  1.79s/it]\u001b[A\n",
            " 44% 22/50 [00:40<00:48,  1.74s/it]\u001b[A\n",
            " 46% 23/50 [00:41<00:48,  1.79s/it]\u001b[A\n",
            " 48% 24/50 [00:43<00:45,  1.76s/it]\u001b[A\n",
            " 50% 25/50 [00:45<00:47,  1.91s/it]\u001b[A\n",
            " 52% 26/50 [00:47<00:44,  1.83s/it]\u001b[A\n",
            " 54% 27/50 [00:49<00:39,  1.73s/it]\u001b[A\n",
            " 56% 28/50 [00:50<00:37,  1.69s/it]\u001b[A\n",
            " 58% 29/50 [00:52<00:36,  1.72s/it]\u001b[A\n",
            " 60% 30/50 [00:54<00:35,  1.78s/it]\u001b[A\n",
            " 62% 31/50 [00:56<00:34,  1.83s/it]\u001b[A\n",
            " 64% 32/50 [00:58<00:35,  1.95s/it]\u001b[A\n",
            " 66% 33/50 [01:00<00:32,  1.94s/it]\u001b[A\n",
            " 68% 34/50 [01:01<00:29,  1.83s/it]\u001b[A\n",
            " 70% 35/50 [01:03<00:27,  1.81s/it]\u001b[A\n",
            " 72% 36/50 [01:05<00:25,  1.81s/it]\u001b[A\n",
            " 74% 37/50 [01:07<00:22,  1.75s/it]\u001b[A\n",
            " 76% 38/50 [01:09<00:21,  1.79s/it]\u001b[A\n",
            " 78% 39/50 [01:10<00:18,  1.67s/it]\u001b[A\n",
            " 80% 40/50 [01:11<00:16,  1.62s/it]\u001b[A\n",
            " 82% 41/50 [01:13<00:14,  1.58s/it]\u001b[A\n",
            " 84% 42/50 [01:15<00:13,  1.68s/it]\u001b[A\n",
            " 86% 43/50 [01:17<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:18<00:09,  1.66s/it]\u001b[A\n",
            " 90% 45/50 [01:20<00:08,  1.73s/it]\u001b[A\n",
            " 92% 46/50 [01:22<00:07,  1.83s/it]\u001b[A\n",
            " 94% 47/50 [01:24<00:05,  1.76s/it]\u001b[A\n",
            " 96% 48/50 [01:25<00:03,  1.75s/it]\u001b[A\n",
            " 98% 49/50 [01:27<00:01,  1.75s/it]\u001b[A\n",
            "100% 50/50 [01:29<00:00,  1.80s/it]\u001b[A\n",
            "{'eval_loss': 2.0732085704803467, 'eval_rouge1': 38.7563, 'eval_rouge2': 15.5955, 'eval_rougeL': 30.3753, 'eval_rougeLsum': 30.4563, 'eval_gen_len': 26.9, 'epoch': 2.2058823529411766}\n",
            "\n",
            " 74% 22500/30600 [4:44:42<1:18:52,  1.71it/s]\n",
            "{'loss': 1.101796875, 'learning_rate': 7.450980392156863e-06, 'epoch': 2.2549019607843137}\n",
            " 75% 23000/30600 [4:49:43<1:14:00,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 19:48:17,817 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:48:17,817 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:48:17,817 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.11it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:56,  1.20s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:04,  1.41s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:06,  1.47s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:02,  1.46s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.50s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.60s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.64s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:02,  1.61s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.70s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.77s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:03,  1.76s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:02,  1.78s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:55,  1.63s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:54,  1.65s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.57s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:53,  1.72s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:51,  1.73s/it]\u001b[A\n",
            " 42% 21/50 [00:34<00:52,  1.82s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:48,  1.74s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:45,  1.69s/it]\u001b[A\n",
            " 48% 24/50 [00:39<00:44,  1.71s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:42,  1.71s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:40,  1.68s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:36,  1.59s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:36,  1.74s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:36,  1.81s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:33,  1.78s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:32,  1.81s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:31,  1.84s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.75s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:24,  1.66s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:23,  1.71s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:22,  1.71s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:19,  1.62s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:16,  1.53s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:15,  1.53s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:13,  1.55s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.63s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.69s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.69s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.75s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.68s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.70s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.62s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.62s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.55s/it]\u001b[A\n",
            "{'eval_loss': 2.0786478519439697, 'eval_rouge1': 38.9435, 'eval_rouge2': 15.319, 'eval_rougeL': 30.3095, 'eval_rougeLsum': 30.2851, 'eval_gen_len': 25.7, 'epoch': 2.2549019607843137}\n",
            "\n",
            " 75% 23000/30600 [4:51:07<1:14:00,  1.71it/s]\n",
            "{'loss': 1.1330390625, 'learning_rate': 6.960784313725491e-06, 'epoch': 2.303921568627451}\n",
            " 77% 23500/30600 [4:56:07<1:08:08,  1.74it/s][INFO|trainer.py:1387] 2020-11-17 19:54:42,572 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 19:54:42,572 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 19:54:42,572 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:42,  1.12it/s]\u001b[A\n",
            "  6% 3/50 [00:04<01:01,  1.30s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:05,  1.43s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:10,  1.56s/it]\u001b[A\n",
            " 12% 6/50 [00:09<01:07,  1.52s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:09,  1.62s/it]\u001b[A\n",
            " 16% 8/50 [00:12<01:06,  1.59s/it]\u001b[A\n",
            " 18% 9/50 [00:14<01:08,  1.67s/it]\u001b[A\n",
            " 20% 10/50 [00:16<01:08,  1.72s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:07,  1.74s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:09,  1.82s/it]\u001b[A\n",
            " 26% 13/50 [00:21<01:07,  1.82s/it]\u001b[A\n",
            " 28% 14/50 [00:23<01:03,  1.77s/it]\u001b[A\n",
            " 30% 15/50 [00:25<01:04,  1.84s/it]\u001b[A\n",
            " 32% 16/50 [00:26<00:56,  1.66s/it]\u001b[A\n",
            " 34% 17/50 [00:28<00:57,  1.74s/it]\u001b[A\n",
            " 36% 18/50 [00:30<00:53,  1.68s/it]\u001b[A\n",
            " 38% 19/50 [00:31<00:52,  1.70s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:49,  1.67s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:48,  1.67s/it]\u001b[A\n",
            " 44% 22/50 [00:36<00:45,  1.64s/it]\u001b[A\n",
            " 46% 23/50 [00:38<00:45,  1.70s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:45,  1.74s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:44,  1.76s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:41,  1.71s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:37,  1.65s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:48<00:36,  1.74s/it]\u001b[A\n",
            " 60% 30/50 [00:50<00:36,  1.81s/it]\u001b[A\n",
            " 62% 31/50 [00:52<00:34,  1.80s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:33,  1.85s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:31,  1.85s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:28,  1.79s/it]\u001b[A\n",
            " 70% 35/50 [00:59<00:26,  1.78s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:24,  1.74s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:22,  1.71s/it]\u001b[A\n",
            " 76% 38/50 [01:05<00:22,  1.85s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:18,  1.72s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:17,  1.76s/it]\u001b[A\n",
            " 82% 41/50 [01:09<00:15,  1.67s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.72s/it]\u001b[A\n",
            " 86% 43/50 [01:13<00:12,  1.77s/it]\u001b[A\n",
            " 88% 44/50 [01:15<00:10,  1.75s/it]\u001b[A\n",
            " 90% 45/50 [01:17<00:08,  1.73s/it]\u001b[A\n",
            " 92% 46/50 [01:18<00:06,  1.70s/it]\u001b[A\n",
            " 94% 47/50 [01:20<00:05,  1.72s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.63s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.67s/it]\u001b[A\n",
            "100% 50/50 [01:25<00:00,  1.62s/it]\u001b[A\n",
            "{'eval_loss': 2.0728156566619873, 'eval_rouge1': 38.1625, 'eval_rouge2': 15.3522, 'eval_rougeL': 30.8552, 'eval_rougeLsum': 30.8322, 'eval_gen_len': 26.2, 'epoch': 2.303921568627451}\n",
            "\n",
            " 77% 23500/30600 [4:57:35<1:08:08,  1.74it/s]\n",
            "{'loss': 1.1087265625, 'learning_rate': 6.470588235294118e-06, 'epoch': 2.3529411764705883}\n",
            " 78% 24000/30600 [5:02:32<1:06:03,  1.67it/s][INFO|trainer.py:1387] 2020-11-17 20:01:06,862 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:01:06,862 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:01:06,862 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:42,  1.14it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.16s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.34s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.49s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.58s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:07,  1.68s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:09,  1.78s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:11,  1.88s/it]\u001b[A\n",
            " 26% 13/50 [00:21<01:08,  1.84s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:03,  1.76s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:03,  1.82s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:56,  1.65s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:56,  1.72s/it]\u001b[A\n",
            " 36% 18/50 [00:29<00:51,  1.62s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:51,  1.67s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:49,  1.65s/it]\u001b[A\n",
            " 42% 21/50 [00:34<00:48,  1.68s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:45,  1.62s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:43,  1.62s/it]\u001b[A\n",
            " 48% 24/50 [00:39<00:44,  1.71s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:41,  1.67s/it]\u001b[A\n",
            " 52% 26/50 [00:42<00:40,  1.70s/it]\u001b[A\n",
            " 54% 27/50 [00:44<00:38,  1.65s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:35,  1.62s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:36,  1.75s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:36,  1.81s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:33,  1.78s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:31,  1.76s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:31,  1.83s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.78s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:26,  1.75s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:24,  1.76s/it]\u001b[A\n",
            " 74% 37/50 [01:02<00:22,  1.76s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:19,  1.66s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:17,  1.56s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:15,  1.57s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:13,  1.55s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.72s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:12,  1.75s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.76s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.69s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.69s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.69s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.66s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.68s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  2.09s/it]\u001b[A\n",
            "{'eval_loss': 2.0813841819763184, 'eval_rouge1': 38.6351, 'eval_rouge2': 15.1329, 'eval_rougeL': 29.946, 'eval_rougeLsum': 29.8768, 'eval_gen_len': 26.2, 'epoch': 2.3529411764705883}\n",
            "\n",
            " 78% 24000/30600 [5:04:00<1:06:03,  1.67it/s]\n",
            "{'loss': 1.1015234375, 'learning_rate': 5.980392156862746e-06, 'epoch': 2.4019607843137254}\n",
            " 80% 24500/30600 [5:08:53<59:02,  1.72it/s][INFO|trainer.py:1387] 2020-11-17 20:07:28,377 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:07:28,377 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:07:28,377 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.11it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.16s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.34s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:07,  1.49s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.45s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:02,  1.46s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.49s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:02,  1.54s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:05,  1.63s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:07,  1.73s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:11,  1.87s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.76s/it]\u001b[A\n",
            " 28% 14/50 [00:23<01:17,  2.14s/it]\u001b[A\n",
            " 30% 15/50 [00:25<01:09,  2.00s/it]\u001b[A\n",
            " 32% 16/50 [00:26<01:00,  1.77s/it]\u001b[A\n",
            " 34% 17/50 [00:28<00:58,  1.77s/it]\u001b[A\n",
            " 36% 18/50 [00:29<00:52,  1.64s/it]\u001b[A\n",
            " 38% 19/50 [00:31<00:51,  1.65s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:49,  1.65s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:51,  1.77s/it]\u001b[A\n",
            " 44% 22/50 [00:36<00:46,  1.67s/it]\u001b[A\n",
            " 46% 23/50 [00:38<00:44,  1.65s/it]\u001b[A\n",
            " 48% 24/50 [00:39<00:43,  1.67s/it]\u001b[A\n",
            " 50% 25/50 [00:41<00:42,  1.70s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:40,  1.67s/it]\u001b[A\n",
            " 54% 27/50 [00:44<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:34,  1.55s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:33,  1.58s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:33,  1.66s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:30,  1.62s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:30,  1.72s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:30,  1.82s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:29,  1.84s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:26,  1.79s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:24,  1.76s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:22,  1.73s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:19,  1.66s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:17,  1.57s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:15,  1.54s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.65s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.73s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.65s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.69s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.74s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:07,  1.78s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.80s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.77s/it]\u001b[A\n",
            " 98% 49/50 [01:22<00:01,  1.76s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.72s/it]\u001b[A\n",
            "{'eval_loss': 2.074784994125366, 'eval_rouge1': 38.3667, 'eval_rouge2': 15.4177, 'eval_rougeL': 30.2901, 'eval_rougeLsum': 30.3096, 'eval_gen_len': 26.0, 'epoch': 2.4019607843137254}\n",
            "\n",
            " 80% 24500/30600 [5:10:20<59:02,  1.72it/s]\n",
            "{'loss': 1.1093125, 'learning_rate': 5.490196078431373e-06, 'epoch': 2.450980392156863}\n",
            " 82% 25000/30600 [5:15:09<54:12,  1.72it/s][INFO|trainer.py:1387] 2020-11-17 20:13:43,937 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:13:43,937 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:13:43,937 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:02<00:59,  1.24s/it]\u001b[A\n",
            "  6% 3/50 [00:04<01:04,  1.38s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:08,  1.49s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:09,  1.54s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:04,  1.46s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.48s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:02,  1.53s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:01,  1.53s/it]\u001b[A\n",
            " 22% 11/50 [00:16<00:59,  1.52s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:01,  1.62s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:01,  1.66s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:59,  1.66s/it]\u001b[A\n",
            " 30% 15/50 [00:23<01:03,  1.81s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:55,  1.64s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:54,  1.67s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:52,  1.64s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:51,  1.65s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:48,  1.63s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:49,  1.72s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:44,  1.59s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:42,  1.59s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:42,  1.71s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:41,  1.75s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:36,  1.60s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:33,  1.60s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:32,  1.64s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:32,  1.73s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:31,  1.75s/it]\u001b[A\n",
            " 66% 33/50 [00:53<00:30,  1.78s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:27,  1.71s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:24,  1.64s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:24,  1.74s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.69s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:22,  1.85s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:18,  1.70s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:14,  1.66s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:14,  1.78s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:12,  1.76s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.75s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.71s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.72s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.76s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.75s/it]\u001b[A\n",
            " 98% 49/50 [01:22<00:01,  1.98s/it]\u001b[A\n",
            "100% 50/50 [01:25<00:00,  2.33s/it]\u001b[A\n",
            "{'eval_loss': 2.0784924030303955, 'eval_rouge1': 38.8799, 'eval_rouge2': 15.6326, 'eval_rougeL': 30.6703, 'eval_rougeLsum': 30.4659, 'eval_gen_len': 25.7, 'epoch': 2.450980392156863}\n",
            "\n",
            " 82% 25000/30600 [5:16:38<54:12,  1.72it/s]\n",
            "{'loss': 1.095296875, 'learning_rate': 4.9999999999999996e-06, 'epoch': 2.5}\n",
            " 83% 25500/30600 [5:21:25<49:05,  1.73it/s][INFO|trainer.py:1387] 2020-11-17 20:20:00,079 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:20:00,079 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:20:00,079 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:41,  1.16it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:57,  1.23s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:05,  1.46s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.43s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.47s/it]\u001b[A\n",
            " 18% 9/50 [00:12<00:59,  1.45s/it]\u001b[A\n",
            " 20% 10/50 [00:14<00:59,  1.49s/it]\u001b[A\n",
            " 22% 11/50 [00:16<00:59,  1.53s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:04,  1.69s/it]\u001b[A\n",
            " 26% 13/50 [00:19<01:02,  1.69s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:59,  1.65s/it]\u001b[A\n",
            " 30% 15/50 [00:23<01:00,  1.72s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:53,  1.58s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:52,  1.60s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:48,  1.51s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:47,  1.55s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:45,  1.53s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:45,  1.58s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:42,  1.53s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:43,  1.63s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:42,  1.64s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:42,  1.72s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:40,  1.69s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.59s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:34,  1.58s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:33,  1.58s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:31,  1.59s/it]\u001b[A\n",
            " 62% 31/50 [00:48<00:30,  1.61s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:29,  1.65s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:29,  1.71s/it]\u001b[A\n",
            " 68% 34/50 [00:53<00:25,  1.61s/it]\u001b[A\n",
            " 70% 35/50 [00:55<00:25,  1.67s/it]\u001b[A\n",
            " 72% 36/50 [00:57<00:23,  1.71s/it]\u001b[A\n",
            " 74% 37/50 [00:58<00:21,  1.65s/it]\u001b[A\n",
            " 76% 38/50 [01:00<00:21,  1.76s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:18,  1.69s/it]\u001b[A\n",
            " 80% 40/50 [01:03<00:16,  1.61s/it]\u001b[A\n",
            " 82% 41/50 [01:04<00:13,  1.53s/it]\u001b[A\n",
            " 84% 42/50 [01:06<00:13,  1.64s/it]\u001b[A\n",
            " 86% 43/50 [01:08<00:11,  1.65s/it]\u001b[A\n",
            " 88% 44/50 [01:10<00:09,  1.66s/it]\u001b[A\n",
            " 90% 45/50 [01:11<00:08,  1.65s/it]\u001b[A\n",
            " 92% 46/50 [01:13<00:06,  1.69s/it]\u001b[A\n",
            " 94% 47/50 [01:15<00:05,  1.75s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.73s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.76s/it]\u001b[A\n",
            "100% 50/50 [01:20<00:00,  1.75s/it]\u001b[A\n",
            "{'eval_loss': 2.0649449825286865, 'eval_rouge1': 38.8418, 'eval_rouge2': 15.7828, 'eval_rougeL': 30.733, 'eval_rougeLsum': 30.8582, 'eval_gen_len': 25.5, 'epoch': 2.5}\n",
            "\n",
            " 83% 25500/30600 [5:22:49<49:05,  1.73it/s]\n",
            "{'loss': 1.1066640625, 'learning_rate': 4.509803921568627e-06, 'epoch': 2.549019607843137}\n",
            " 85% 26000/30600 [5:27:42<42:34,  1.80it/s][INFO|trainer.py:1387] 2020-11-17 20:26:17,178 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:26:17,178 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:26:17,178 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:02<00:49,  1.03s/it]\u001b[A\n",
            "  6% 3/50 [00:03<01:00,  1.30s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:05,  1.43s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:05,  1.46s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:02,  1.45s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.50s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:02,  1.51s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:04,  1.62s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:03,  1.63s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:05,  1.71s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:03,  1.71s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:02,  1.75s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:57,  1.65s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:52,  1.55s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:53,  1.63s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.57s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:49,  1.60s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:49,  1.72s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:45,  1.63s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:44,  1.63s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:42,  1.63s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:40,  1.63s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:36,  1.54s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:35,  1.54s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:32,  1.57s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:31,  1.59s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:31,  1.64s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:30,  1.71s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:29,  1.76s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:27,  1.69s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:34,  2.29s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:31,  2.25s/it]\u001b[A\n",
            " 74% 37/50 [01:02<00:27,  2.08s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:23,  1.95s/it]\u001b[A\n",
            " 78% 39/50 [01:05<00:20,  1.83s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:17,  1.73s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.63s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.72s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.68s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.67s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.63s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.70s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.76s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.72s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.73s/it]\u001b[A\n",
            "100% 50/50 [01:23<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.0557656288146973, 'eval_rouge1': 39.1207, 'eval_rouge2': 15.2657, 'eval_rougeL': 30.4561, 'eval_rougeLsum': 30.4353, 'eval_gen_len': 25.3, 'epoch': 2.549019607843137}\n",
            "\n",
            " 85% 26000/30600 [5:29:08<42:34,  1.80it/s]\n",
            "{'loss': 1.1278125, 'learning_rate': 4.019607843137255e-06, 'epoch': 2.5980392156862746}\n",
            " 87% 26500/30600 [5:34:06<41:00,  1.67it/s][INFO|trainer.py:1387] 2020-11-17 20:32:40,910 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:32:40,910 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:32:40,910 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:44,  1.09it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:56,  1.20s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.35s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:02,  1.42s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.45s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.49s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.57s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:03,  1.59s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:04,  1.65s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:06,  1.75s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:08,  1.86s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:04,  1.81s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:59,  1.69s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:53,  1.56s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:53,  1.63s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:50,  1.59s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.56s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:47,  1.58s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:47,  1.62s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:44,  1.61s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:43,  1.61s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:42,  1.62s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:42,  1.71s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:35,  1.53s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:35,  1.60s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:34,  1.63s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:33,  1.68s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:33,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:51<00:31,  1.78s/it]\u001b[A\n",
            " 66% 33/50 [00:53<00:31,  1.85s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:29,  1.84s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:27,  1.82s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:25,  1.85s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:22,  1.76s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:20,  1.69s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:17,  1.62s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:15,  1.59s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:14,  1.62s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.71s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.70s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.69s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.74s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.71s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.77s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.72s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.72s/it]\u001b[A\n",
            "100% 50/50 [01:22<00:00,  1.68s/it]\u001b[A\n",
            "{'eval_loss': 2.048461675643921, 'eval_rouge1': 39.2956, 'eval_rouge2': 15.73, 'eval_rougeL': 29.9688, 'eval_rougeLsum': 30.0269, 'eval_gen_len': 25.7, 'epoch': 2.5980392156862746}\n",
            "\n",
            " 87% 26500/30600 [5:35:31<41:00,  1.67it/s]\n",
            "{'loss': 1.1191171875, 'learning_rate': 3.5294117647058825e-06, 'epoch': 2.6470588235294117}\n",
            " 88% 27000/30600 [5:40:29<35:43,  1.68it/s][INFO|trainer.py:1387] 2020-11-17 20:39:04,243 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:39:04,244 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:39:04,244 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:42,  1.13it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:55,  1.18s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.35s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.41s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.40s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.45s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:02,  1.49s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.59s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.71s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:04,  1.66s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:07,  1.77s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.78s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:02,  1.75s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:03,  1.81s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:55,  1.64s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:55,  1.67s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:51,  1.62s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:50,  1.62s/it]\u001b[A\n",
            " 40% 20/50 [00:32<00:48,  1.62s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:49,  1.70s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:46,  1.68s/it]\u001b[A\n",
            " 46% 23/50 [00:37<00:46,  1.72s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:44,  1.69s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:37,  1.58s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:34,  1.51s/it]\u001b[A\n",
            " 56% 28/50 [00:45<00:35,  1.61s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:34,  1.66s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:34,  1.71s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:33,  1.77s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:32,  1.83s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:31,  1.85s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:27,  1.74s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:25,  1.73s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:24,  1.74s/it]\u001b[A\n",
            " 74% 37/50 [01:00<00:21,  1.65s/it]\u001b[A\n",
            " 76% 38/50 [01:02<00:18,  1.57s/it]\u001b[A\n",
            " 78% 39/50 [01:03<00:16,  1.50s/it]\u001b[A\n",
            " 80% 40/50 [01:05<00:15,  1.52s/it]\u001b[A\n",
            " 82% 41/50 [01:06<00:13,  1.49s/it]\u001b[A\n",
            " 84% 42/50 [01:08<00:12,  1.57s/it]\u001b[A\n",
            " 86% 43/50 [01:10<00:11,  1.64s/it]\u001b[A\n",
            " 88% 44/50 [01:11<00:09,  1.65s/it]\u001b[A\n",
            " 90% 45/50 [01:13<00:08,  1.71s/it]\u001b[A\n",
            " 92% 46/50 [01:15<00:06,  1.72s/it]\u001b[A\n",
            " 94% 47/50 [01:17<00:05,  1.73s/it]\u001b[A\n",
            " 96% 48/50 [01:18<00:03,  1.65s/it]\u001b[A\n",
            " 98% 49/50 [01:20<00:01,  1.70s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.63s/it]\u001b[A\n",
            "{'eval_loss': 2.064262628555298, 'eval_rouge1': 37.9491, 'eval_rouge2': 15.1541, 'eval_rougeL': 30.4904, 'eval_rougeLsum': 30.4347, 'eval_gen_len': 25.6, 'epoch': 2.6470588235294117}\n",
            "\n",
            " 88% 27000/30600 [5:41:53<35:43,  1.68it/s]\n",
            "{'loss': 1.1092265625, 'learning_rate': 3.03921568627451e-06, 'epoch': 2.696078431372549}\n",
            " 90% 27500/30600 [5:46:47<31:07,  1.66it/s][INFO|trainer.py:1387] 2020-11-17 20:45:22,275 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:45:22,275 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:45:22,275 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.11it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:54,  1.17s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:01,  1.34s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.40s/it]\u001b[A\n",
            " 14% 7/50 [00:09<00:59,  1.38s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:00,  1.43s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:03,  1.54s/it]\u001b[A\n",
            " 20% 10/50 [00:17<01:32,  2.30s/it]\u001b[A\n",
            " 22% 11/50 [00:18<01:23,  2.13s/it]\u001b[A\n",
            " 24% 12/50 [00:20<01:19,  2.10s/it]\u001b[A\n",
            " 26% 13/50 [00:22<01:12,  1.95s/it]\u001b[A\n",
            " 28% 14/50 [00:24<01:06,  1.85s/it]\u001b[A\n",
            " 30% 15/50 [00:25<01:04,  1.85s/it]\u001b[A\n",
            " 32% 16/50 [00:27<00:57,  1.68s/it]\u001b[A\n",
            " 34% 17/50 [00:28<00:56,  1.72s/it]\u001b[A\n",
            " 36% 18/50 [00:30<00:51,  1.62s/it]\u001b[A\n",
            " 38% 19/50 [00:31<00:49,  1.59s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:47,  1.59s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:36<00:44,  1.60s/it]\u001b[A\n",
            " 46% 23/50 [00:38<00:45,  1.67s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:43,  1.68s/it]\u001b[A\n",
            " 50% 25/50 [00:41<00:41,  1.64s/it]\u001b[A\n",
            " 52% 26/50 [00:43<00:37,  1.56s/it]\u001b[A\n",
            " 54% 27/50 [00:44<00:33,  1.45s/it]\u001b[A\n",
            " 56% 28/50 [00:46<00:33,  1.52s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:33,  1.60s/it]\u001b[A\n",
            " 60% 30/50 [00:49<00:34,  1.72s/it]\u001b[A\n",
            " 62% 31/50 [00:51<00:32,  1.69s/it]\u001b[A\n",
            " 64% 32/50 [00:53<00:31,  1.75s/it]\u001b[A\n",
            " 66% 33/50 [00:55<00:30,  1.77s/it]\u001b[A\n",
            " 68% 34/50 [00:56<00:28,  1.77s/it]\u001b[A\n",
            " 70% 35/50 [00:58<00:26,  1.74s/it]\u001b[A\n",
            " 72% 36/50 [01:00<00:25,  1.79s/it]\u001b[A\n",
            " 74% 37/50 [01:02<00:22,  1.75s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:19,  1.64s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:17,  1.57s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:15,  1.54s/it]\u001b[A\n",
            " 82% 41/50 [01:08<00:14,  1.63s/it]\u001b[A\n",
            " 84% 42/50 [01:10<00:13,  1.73s/it]\u001b[A\n",
            " 86% 43/50 [01:12<00:12,  1.75s/it]\u001b[A\n",
            " 88% 44/50 [01:13<00:10,  1.70s/it]\u001b[A\n",
            " 90% 45/50 [01:15<00:08,  1.65s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.65s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:04,  1.66s/it]\u001b[A\n",
            " 96% 48/50 [01:20<00:03,  1.64s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.69s/it]\u001b[A\n",
            "100% 50/50 [01:23<00:00,  1.60s/it]\u001b[A\n",
            "{'eval_loss': 2.059765338897705, 'eval_rouge1': 38.3414, 'eval_rouge2': 15.242, 'eval_rougeL': 30.5637, 'eval_rougeLsum': 30.3824, 'eval_gen_len': 25.0, 'epoch': 2.696078431372549}\n",
            "\n",
            " 90% 27500/30600 [5:48:13<31:07,  1.66it/s]\n",
            "{'loss': 1.1117265625, 'learning_rate': 2.5490196078431376e-06, 'epoch': 2.7450980392156863}\n",
            " 92% 28000/30600 [5:53:08<26:06,  1.66it/s][INFO|trainer.py:1387] 2020-11-17 20:51:43,589 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:51:43,589 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:51:43,589 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:02<00:48,  1.02s/it]\u001b[A\n",
            "  6% 3/50 [00:03<00:58,  1.24s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:03,  1.38s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.39s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:00,  1.41s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.46s/it]\u001b[A\n",
            " 18% 9/50 [00:12<01:00,  1.48s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:03,  1.60s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.58s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:03,  1.68s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:07,  1.82s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:03,  1.77s/it]\u001b[A\n",
            " 30% 15/50 [00:23<01:03,  1.81s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:55,  1.65s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:56,  1.71s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:52,  1.63s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:50,  1.62s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:48,  1.62s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:48,  1.66s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:44,  1.59s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:42,  1.57s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:39,  1.59s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.62s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.61s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.66s/it]\u001b[A\n",
            " 58% 29/50 [00:46<00:34,  1.65s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:34,  1.74s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:33,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:52<00:34,  1.90s/it]\u001b[A\n",
            " 66% 33/50 [00:54<00:32,  1.88s/it]\u001b[A\n",
            " 68% 34/50 [00:55<00:28,  1.77s/it]\u001b[A\n",
            " 70% 35/50 [00:57<00:26,  1.75s/it]\u001b[A\n",
            " 72% 36/50 [00:59<00:25,  1.81s/it]\u001b[A\n",
            " 74% 37/50 [01:01<00:23,  1.78s/it]\u001b[A\n",
            " 76% 38/50 [01:03<00:22,  1.89s/it]\u001b[A\n",
            " 78% 39/50 [01:04<00:19,  1.75s/it]\u001b[A\n",
            " 80% 40/50 [01:06<00:16,  1.69s/it]\u001b[A\n",
            " 82% 41/50 [01:07<00:14,  1.59s/it]\u001b[A\n",
            " 84% 42/50 [01:09<00:13,  1.63s/it]\u001b[A\n",
            " 86% 43/50 [01:11<00:11,  1.70s/it]\u001b[A\n",
            " 88% 44/50 [01:12<00:10,  1.72s/it]\u001b[A\n",
            " 90% 45/50 [01:14<00:08,  1.67s/it]\u001b[A\n",
            " 92% 46/50 [01:16<00:06,  1.67s/it]\u001b[A\n",
            " 94% 47/50 [01:18<00:05,  1.75s/it]\u001b[A\n",
            " 96% 48/50 [01:19<00:03,  1.74s/it]\u001b[A\n",
            " 98% 49/50 [01:21<00:01,  1.76s/it]\u001b[A\n",
            "100% 50/50 [01:23<00:00,  1.67s/it]\u001b[A\n",
            "{'eval_loss': 2.0610649585723877, 'eval_rouge1': 39.2436, 'eval_rouge2': 15.7457, 'eval_rougeL': 31.4509, 'eval_rougeLsum': 31.4012, 'eval_gen_len': 25.8, 'epoch': 2.7450980392156863}\n",
            "\n",
            " 92% 28000/30600 [5:54:34<26:06,  1.66it/s]\n",
            "{'loss': 1.084109375, 'learning_rate': 2.058823529411765e-06, 'epoch': 2.7941176470588234}\n",
            " 93% 28500/30600 [5:59:28<20:19,  1.72it/s][INFO|trainer.py:1387] 2020-11-17 20:58:02,972 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 20:58:02,972 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 20:58:02,972 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:43,  1.10it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:04,  1.43s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.39s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.45s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.51s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.61s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:08,  1.71s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:06,  1.69s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:06,  1.75s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:05,  1.76s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:02,  1.74s/it]\u001b[A\n",
            " 30% 15/50 [00:24<01:03,  1.81s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:55,  1.64s/it]\u001b[A\n",
            " 34% 17/50 [00:27<00:55,  1.67s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:51,  1.60s/it]\u001b[A\n",
            " 38% 19/50 [00:30<00:49,  1.61s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:47,  1.59s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:35<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:43,  1.59s/it]\u001b[A\n",
            " 48% 24/50 [00:38<00:42,  1.63s/it]\u001b[A\n",
            " 50% 25/50 [00:40<00:41,  1.67s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:39,  1.63s/it]\u001b[A\n",
            " 54% 27/50 [00:43<00:35,  1.56s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:35,  1.62s/it]\u001b[A\n",
            " 58% 29/50 [00:47<00:37,  1.78s/it]\u001b[A\n",
            " 60% 30/50 [00:48<00:36,  1.81s/it]\u001b[A\n",
            " 62% 31/50 [00:50<00:33,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:43,  2.43s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:40,  2.40s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:35,  2.25s/it]\u001b[A\n",
            " 70% 35/50 [01:00<00:31,  2.12s/it]\u001b[A\n",
            " 72% 36/50 [01:02<00:29,  2.07s/it]\u001b[A\n",
            " 74% 37/50 [01:04<00:25,  1.99s/it]\u001b[A\n",
            " 76% 38/50 [01:05<00:21,  1.81s/it]\u001b[A\n",
            " 78% 39/50 [01:07<00:18,  1.69s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:16,  1.63s/it]\u001b[A\n",
            " 82% 41/50 [01:10<00:14,  1.58s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.66s/it]\u001b[A\n",
            " 86% 43/50 [01:13<00:11,  1.67s/it]\u001b[A\n",
            " 88% 44/50 [01:15<00:09,  1.66s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:08,  1.62s/it]\u001b[A\n",
            " 92% 46/50 [01:18<00:06,  1.64s/it]\u001b[A\n",
            " 94% 47/50 [01:20<00:05,  1.73s/it]\u001b[A\n",
            " 96% 48/50 [01:22<00:03,  1.72s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.74s/it]\u001b[A\n",
            "100% 50/50 [01:25<00:00,  1.67s/it]\u001b[A\n",
            "{'eval_loss': 2.055527925491333, 'eval_rouge1': 38.9266, 'eval_rouge2': 15.8003, 'eval_rougeL': 30.8054, 'eval_rougeLsum': 30.9385, 'eval_gen_len': 25.8, 'epoch': 2.7941176470588234}\n",
            "\n",
            " 93% 28500/30600 [6:00:56<20:19,  1.72it/s]\n",
            "{'loss': 1.0733046875, 'learning_rate': 1.5686274509803923e-06, 'epoch': 2.843137254901961}\n",
            " 95% 29000/30600 [6:05:49<15:37,  1.71it/s][INFO|trainer.py:1387] 2020-11-17 21:04:24,081 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:04:24,081 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:04:24,081 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:44,  1.08it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:56,  1.21s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.35s/it]\u001b[A\n",
            " 10% 5/50 [00:06<01:03,  1.40s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.39s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:01,  1.43s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.47s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:04,  1.57s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:06,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.57s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:03,  1.67s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:02,  1.69s/it]\u001b[A\n",
            " 28% 14/50 [00:21<01:00,  1.68s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:55,  1.60s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:50,  1.50s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:51,  1.56s/it]\u001b[A\n",
            " 36% 18/50 [00:27<00:49,  1.54s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:48,  1.57s/it]\u001b[A\n",
            " 40% 20/50 [00:30<00:46,  1.56s/it]\u001b[A\n",
            " 42% 21/50 [00:32<00:46,  1.61s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:43,  1.55s/it]\u001b[A\n",
            " 46% 23/50 [00:35<00:42,  1.56s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:38<00:39,  1.59s/it]\u001b[A\n",
            " 52% 26/50 [00:40<00:37,  1.58s/it]\u001b[A\n",
            " 54% 27/50 [00:41<00:35,  1.53s/it]\u001b[A\n",
            " 56% 28/50 [00:43<00:33,  1.54s/it]\u001b[A\n",
            " 58% 29/50 [00:45<00:35,  1.71s/it]\u001b[A\n",
            " 60% 30/50 [00:47<00:35,  1.78s/it]\u001b[A\n",
            " 62% 31/50 [00:49<00:33,  1.79s/it]\u001b[A\n",
            " 64% 32/50 [00:50<00:32,  1.80s/it]\u001b[A\n",
            " 66% 33/50 [00:52<00:30,  1.80s/it]\u001b[A\n",
            " 68% 34/50 [00:54<00:28,  1.79s/it]\u001b[A\n",
            " 70% 35/50 [00:56<00:27,  1.82s/it]\u001b[A\n",
            " 72% 36/50 [00:58<00:25,  1.84s/it]\u001b[A\n",
            " 74% 37/50 [00:59<00:22,  1.76s/it]\u001b[A\n",
            " 76% 38/50 [01:01<00:19,  1.66s/it]\u001b[A\n",
            " 78% 39/50 [01:02<00:17,  1.56s/it]\u001b[A\n",
            " 80% 40/50 [01:04<00:15,  1.56s/it]\u001b[A\n",
            " 82% 41/50 [01:05<00:14,  1.57s/it]\u001b[A\n",
            " 84% 42/50 [01:07<00:13,  1.68s/it]\u001b[A\n",
            " 86% 43/50 [01:09<00:11,  1.68s/it]\u001b[A\n",
            " 88% 44/50 [01:11<00:10,  1.68s/it]\u001b[A\n",
            " 90% 45/50 [01:12<00:08,  1.64s/it]\u001b[A\n",
            " 92% 46/50 [01:14<00:06,  1.67s/it]\u001b[A\n",
            " 94% 47/50 [01:16<00:05,  1.74s/it]\u001b[A\n",
            " 96% 48/50 [01:17<00:03,  1.72s/it]\u001b[A\n",
            " 98% 49/50 [01:19<00:01,  1.77s/it]\u001b[A\n",
            "100% 50/50 [01:21<00:00,  1.66s/it]\u001b[A\n",
            "{'eval_loss': 2.0544517040252686, 'eval_rouge1': 39.017, 'eval_rouge2': 15.7428, 'eval_rougeL': 30.7204, 'eval_rougeLsum': 30.7774, 'eval_gen_len': 25.5, 'epoch': 2.843137254901961}\n",
            "\n",
            " 95% 29000/30600 [6:07:12<15:37,  1.71it/s]\n",
            "{'loss': 1.1090390625, 'learning_rate': 1.0784313725490197e-06, 'epoch': 2.892156862745098}\n",
            " 96% 29500/30600 [6:12:08<11:08,  1.65it/s][INFO|trainer.py:1387] 2020-11-17 21:10:43,054 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:10:43,054 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:10:43,054 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:44,  1.09it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:58,  1.24s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:03,  1.38s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:06,  1.48s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:03,  1.44s/it]\u001b[A\n",
            " 14% 7/50 [00:10<01:04,  1.49s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:03,  1.51s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:07,  1.63s/it]\u001b[A\n",
            " 20% 10/50 [00:15<01:06,  1.65s/it]\u001b[A\n",
            " 22% 11/50 [00:17<01:04,  1.64s/it]\u001b[A\n",
            " 24% 12/50 [00:19<01:08,  1.81s/it]\u001b[A\n",
            " 26% 13/50 [00:20<01:06,  1.78s/it]\u001b[A\n",
            " 28% 14/50 [00:22<01:02,  1.74s/it]\u001b[A\n",
            " 30% 15/50 [00:23<00:56,  1.62s/it]\u001b[A\n",
            " 32% 16/50 [00:25<00:51,  1.52s/it]\u001b[A\n",
            " 34% 17/50 [00:26<00:51,  1.57s/it]\u001b[A\n",
            " 36% 18/50 [00:28<00:47,  1.48s/it]\u001b[A\n",
            " 38% 19/50 [00:29<00:46,  1.49s/it]\u001b[A\n",
            " 40% 20/50 [00:31<00:45,  1.52s/it]\u001b[A\n",
            " 42% 21/50 [00:33<00:47,  1.63s/it]\u001b[A\n",
            " 44% 22/50 [00:34<00:44,  1.58s/it]\u001b[A\n",
            " 46% 23/50 [00:36<00:42,  1.57s/it]\u001b[A\n",
            " 48% 24/50 [00:37<00:41,  1.61s/it]\u001b[A\n",
            " 50% 25/50 [00:39<00:41,  1.66s/it]\u001b[A\n",
            " 52% 26/50 [00:41<00:38,  1.59s/it]\u001b[A\n",
            " 54% 27/50 [00:42<00:36,  1.59s/it]\u001b[A\n",
            " 56% 28/50 [00:44<00:36,  1.65s/it]\u001b[A\n",
            " 58% 29/50 [00:48<00:50,  2.38s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:49,  2.49s/it]\u001b[A\n",
            " 62% 31/50 [00:52<00:42,  2.23s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:38,  2.13s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:35,  2.06s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:31,  1.96s/it]\u001b[A\n",
            " 70% 35/50 [01:00<00:29,  1.94s/it]\u001b[A\n",
            " 72% 36/50 [01:02<00:26,  1.92s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:23,  1.83s/it]\u001b[A\n",
            " 76% 38/50 [01:05<00:20,  1.74s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:17,  1.62s/it]\u001b[A\n",
            " 80% 40/50 [01:08<00:15,  1.58s/it]\u001b[A\n",
            " 82% 41/50 [01:09<00:13,  1.51s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:13,  1.63s/it]\u001b[A\n",
            " 86% 43/50 [01:13<00:11,  1.65s/it]\u001b[A\n",
            " 88% 44/50 [01:14<00:09,  1.65s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:08,  1.60s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:06,  1.62s/it]\u001b[A\n",
            " 94% 47/50 [01:19<00:05,  1.68s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.70s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.73s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.059567928314209, 'eval_rouge1': 39.7092, 'eval_rouge2': 15.8494, 'eval_rougeL': 31.3195, 'eval_rougeLsum': 31.3299, 'eval_gen_len': 25.8, 'epoch': 2.892156862745098}\n",
            "\n",
            " 96% 29500/30600 [6:13:35<11:08,  1.65it/s]\n",
            "{'loss': 1.0952265625, 'learning_rate': 5.882352941176471e-07, 'epoch': 2.9411764705882355}\n",
            " 98% 30000/30600 [6:18:23<05:10,  1.93it/s][INFO|trainer.py:1387] 2020-11-17 21:16:58,556 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:16:58,556 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:16:58,556 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:36,  1.30it/s]\u001b[A\n",
            "  6% 3/50 [00:03<00:53,  1.14s/it]\u001b[A\n",
            "  8% 4/50 [00:05<01:02,  1.36s/it]\u001b[A\n",
            " 10% 5/50 [00:07<01:04,  1.44s/it]\u001b[A\n",
            " 12% 6/50 [00:08<01:01,  1.40s/it]\u001b[A\n",
            " 14% 7/50 [00:09<01:02,  1.44s/it]\u001b[A\n",
            " 16% 8/50 [00:11<01:01,  1.46s/it]\u001b[A\n",
            " 18% 9/50 [00:13<01:05,  1.59s/it]\u001b[A\n",
            " 20% 10/50 [00:14<01:04,  1.61s/it]\u001b[A\n",
            " 22% 11/50 [00:16<01:01,  1.59s/it]\u001b[A\n",
            " 24% 12/50 [00:18<01:02,  1.65s/it]\u001b[A\n",
            " 26% 13/50 [00:19<00:59,  1.62s/it]\u001b[A\n",
            " 28% 14/50 [00:21<00:58,  1.63s/it]\u001b[A\n",
            " 30% 15/50 [00:22<00:54,  1.55s/it]\u001b[A\n",
            " 32% 16/50 [00:24<00:49,  1.45s/it]\u001b[A\n",
            " 34% 17/50 [00:25<00:50,  1.52s/it]\u001b[A\n",
            " 36% 18/50 [00:26<00:45,  1.43s/it]\u001b[A\n",
            " 38% 19/50 [00:28<00:44,  1.45s/it]\u001b[A\n",
            " 40% 20/50 [00:29<00:43,  1.46s/it]\u001b[A\n",
            " 42% 21/50 [00:31<00:44,  1.54s/it]\u001b[A\n",
            " 44% 22/50 [00:33<00:41,  1.48s/it]\u001b[A\n",
            " 46% 23/50 [00:34<00:40,  1.49s/it]\u001b[A\n",
            " 48% 24/50 [00:36<00:39,  1.53s/it]\u001b[A\n",
            " 50% 25/50 [00:37<00:39,  1.57s/it]\u001b[A\n",
            " 52% 26/50 [00:39<00:37,  1.57s/it]\u001b[A\n",
            " 54% 27/50 [00:40<00:34,  1.50s/it]\u001b[A\n",
            " 56% 28/50 [00:42<00:34,  1.56s/it]\u001b[A\n",
            " 58% 29/50 [00:44<00:35,  1.69s/it]\u001b[A\n",
            " 60% 30/50 [00:46<00:34,  1.75s/it]\u001b[A\n",
            " 62% 31/50 [00:47<00:32,  1.69s/it]\u001b[A\n",
            " 64% 32/50 [00:49<00:30,  1.71s/it]\u001b[A\n",
            " 66% 33/50 [00:51<00:29,  1.72s/it]\u001b[A\n",
            " 68% 34/50 [00:52<00:26,  1.66s/it]\u001b[A\n",
            " 70% 35/50 [00:54<00:24,  1.66s/it]\u001b[A\n",
            " 72% 36/50 [00:56<00:23,  1.71s/it]\u001b[A\n",
            " 74% 37/50 [00:57<00:20,  1.61s/it]\u001b[A\n",
            " 76% 38/50 [00:59<00:18,  1.53s/it]\u001b[A\n",
            " 78% 39/50 [01:00<00:16,  1.47s/it]\u001b[A\n",
            " 80% 40/50 [01:01<00:14,  1.47s/it]\u001b[A\n",
            " 82% 41/50 [01:03<00:12,  1.41s/it]\u001b[A\n",
            " 84% 42/50 [01:04<00:12,  1.52s/it]\u001b[A\n",
            " 86% 43/50 [01:06<00:10,  1.54s/it]\u001b[A\n",
            " 88% 44/50 [01:08<00:09,  1.55s/it]\u001b[A\n",
            " 90% 45/50 [01:09<00:07,  1.57s/it]\u001b[A\n",
            " 92% 46/50 [01:11<00:06,  1.57s/it]\u001b[A\n",
            " 94% 47/50 [01:13<00:04,  1.62s/it]\u001b[A\n",
            " 96% 48/50 [01:14<00:03,  1.62s/it]\u001b[A\n",
            " 98% 49/50 [01:16<00:01,  1.64s/it]\u001b[A\n",
            "100% 50/50 [01:17<00:00,  1.56s/it]\u001b[A\n",
            "{'eval_loss': 2.056570529937744, 'eval_rouge1': 40.1666, 'eval_rouge2': 16.4195, 'eval_rougeL': 31.5833, 'eval_rougeLsum': 31.4883, 'eval_gen_len': 25.7, 'epoch': 2.9411764705882355}\n",
            "\n",
            " 98% 30000/30600 [6:19:43<05:10,  1.93it/s]\n",
            "                                   \u001b[A[INFO|trainer.py:1222] 2020-11-17 21:18:18,405 >> Saving model checkpoint to prophetnet_large_uncased_xsum/checkpoint-30000\n",
            "[INFO|configuration_utils.py:282] 2020-11-17 21:18:18,408 >> Configuration saved in prophetnet_large_uncased_xsum/checkpoint-30000/config.json\n",
            "[INFO|modeling_utils.py:740] 2020-11-17 21:18:26,784 >> Model weights saved in prophetnet_large_uncased_xsum/checkpoint-30000/pytorch_model.bin\n",
            "{'loss': 1.0988828125, 'learning_rate': 9.803921568627452e-08, 'epoch': 2.9901960784313726}\n",
            "100% 30500/30600 [6:26:02<00:54,  1.83it/s][INFO|trainer.py:1387] 2020-11-17 21:24:37,315 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:24:37,315 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:24:37,315 >>   Batch size = 2\n",
            "\n",
            "  0% 0/50 [00:00<?, ?it/s]\u001b[A\n",
            "  4% 2/50 [00:01<00:39,  1.22it/s]\u001b[A\n",
            "  6% 3/50 [00:05<01:20,  1.71s/it]\u001b[A\n",
            "  8% 4/50 [00:08<01:31,  1.99s/it]\u001b[A\n",
            " 10% 5/50 [00:09<01:23,  1.85s/it]\u001b[A\n",
            " 12% 6/50 [00:11<01:15,  1.73s/it]\u001b[A\n",
            " 14% 7/50 [00:12<01:10,  1.64s/it]\u001b[A\n",
            " 16% 8/50 [00:14<01:08,  1.62s/it]\u001b[A\n",
            " 18% 9/50 [00:16<01:10,  1.73s/it]\u001b[A\n",
            " 20% 10/50 [00:17<01:09,  1.73s/it]\u001b[A\n",
            " 22% 11/50 [00:19<01:05,  1.67s/it]\u001b[A\n",
            " 24% 12/50 [00:21<01:06,  1.75s/it]\u001b[A\n",
            " 26% 13/50 [00:23<01:05,  1.76s/it]\u001b[A\n",
            " 28% 14/50 [00:24<01:02,  1.74s/it]\u001b[A\n",
            " 30% 15/50 [00:26<00:59,  1.71s/it]\u001b[A\n",
            " 32% 16/50 [00:27<00:54,  1.59s/it]\u001b[A\n",
            " 34% 17/50 [00:29<00:53,  1.63s/it]\u001b[A\n",
            " 36% 18/50 [00:30<00:48,  1.53s/it]\u001b[A\n",
            " 38% 19/50 [00:32<00:47,  1.55s/it]\u001b[A\n",
            " 40% 20/50 [00:33<00:46,  1.56s/it]\u001b[A\n",
            " 42% 21/50 [00:35<00:47,  1.64s/it]\u001b[A\n",
            " 44% 22/50 [00:37<00:44,  1.59s/it]\u001b[A\n",
            " 46% 23/50 [00:38<00:43,  1.61s/it]\u001b[A\n",
            " 48% 24/50 [00:40<00:43,  1.66s/it]\u001b[A\n",
            " 50% 25/50 [00:42<00:42,  1.70s/it]\u001b[A\n",
            " 52% 26/50 [00:44<00:40,  1.70s/it]\u001b[A\n",
            " 54% 27/50 [00:45<00:38,  1.66s/it]\u001b[A\n",
            " 56% 28/50 [00:47<00:36,  1.64s/it]\u001b[A\n",
            " 58% 29/50 [00:49<00:37,  1.78s/it]\u001b[A\n",
            " 60% 30/50 [00:51<00:36,  1.84s/it]\u001b[A\n",
            " 62% 31/50 [00:52<00:33,  1.74s/it]\u001b[A\n",
            " 64% 32/50 [00:54<00:31,  1.78s/it]\u001b[A\n",
            " 66% 33/50 [00:56<00:30,  1.81s/it]\u001b[A\n",
            " 68% 34/50 [00:58<00:28,  1.78s/it]\u001b[A\n",
            " 70% 35/50 [01:00<00:26,  1.78s/it]\u001b[A\n",
            " 72% 36/50 [01:01<00:25,  1.81s/it]\u001b[A\n",
            " 74% 37/50 [01:03<00:22,  1.74s/it]\u001b[A\n",
            " 76% 38/50 [01:04<00:19,  1.63s/it]\u001b[A\n",
            " 78% 39/50 [01:06<00:17,  1.57s/it]\u001b[A\n",
            " 80% 40/50 [01:07<00:15,  1.56s/it]\u001b[A\n",
            " 82% 41/50 [01:09<00:13,  1.49s/it]\u001b[A\n",
            " 84% 42/50 [01:11<00:12,  1.62s/it]\u001b[A\n",
            " 86% 43/50 [01:12<00:11,  1.64s/it]\u001b[A\n",
            " 88% 44/50 [01:14<00:09,  1.66s/it]\u001b[A\n",
            " 90% 45/50 [01:16<00:08,  1.67s/it]\u001b[A\n",
            " 92% 46/50 [01:17<00:06,  1.67s/it]\u001b[A\n",
            " 94% 47/50 [01:19<00:05,  1.71s/it]\u001b[A\n",
            " 96% 48/50 [01:21<00:03,  1.70s/it]\u001b[A\n",
            " 98% 49/50 [01:23<00:01,  1.73s/it]\u001b[A\n",
            "100% 50/50 [01:24<00:00,  1.64s/it]\u001b[A\n",
            "{'eval_loss': 2.0574538707733154, 'eval_rouge1': 40.0368, 'eval_rouge2': 15.8821, 'eval_rougeL': 31.1158, 'eval_rougeLsum': 31.1397, 'eval_gen_len': 25.6, 'epoch': 2.9901960784313726}\n",
            "\n",
            "100% 30500/30600 [6:27:29<00:54,  1.83it/s]\n",
            "100% 30600/30600 [6:28:28<00:00,  1.72it/s][INFO|trainer.py:829] 2020-11-17 21:27:03,009 >> \n",
            "\n",
            "Training completed. Do not forget to share your model on huggingface.co/models =)\n",
            "\n",
            "\n",
            "{'epoch': 3.0}\n",
            "100% 30600/30600 [6:28:28<00:00,  1.31it/s]\n",
            "[INFO|trainer.py:1222] 2020-11-17 21:27:03,011 >> Saving model checkpoint to prophetnet_large_uncased_xsum\n",
            "[INFO|configuration_utils.py:282] 2020-11-17 21:27:03,013 >> Configuration saved in prophetnet_large_uncased_xsum/config.json\n",
            "[INFO|modeling_utils.py:740] 2020-11-17 21:27:11,398 >> Model weights saved in prophetnet_large_uncased_xsum/pytorch_model.bin\n",
            "11/17/2020 21:27:11 - INFO - __main__ -   *** Evaluate ***\n",
            "[INFO|trainer.py:1387] 2020-11-17 21:27:11,423 >> ***** Running Evaluation *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:27:11,423 >>   Num examples = 100\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:27:11,423 >>   Batch size = 2\n",
            "100% 50/50 [01:22<00:00,  1.65s/it]\n",
            "11/17/2020 21:28:35 - INFO - __main__ -   ***** Eval results *****\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_loss = 2.057460308074951\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_rouge1 = 40.0447\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_rouge2 = 15.8487\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_rougeL = 31.2542\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_rougeLsum = 31.2476\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     eval_gen_len = 25.6\n",
            "11/17/2020 21:28:35 - INFO - __main__ -     epoch = 3.0\n",
            "11/17/2020 21:28:35 - INFO - root -   *** Test ***\n",
            "[INFO|trainer.py:1387] 2020-11-17 21:28:35,447 >> ***** Running Prediction *****\n",
            "[INFO|trainer.py:1388] 2020-11-17 21:28:35,448 >>   Num examples = 11332\n",
            "[INFO|trainer.py:1389] 2020-11-17 21:28:35,448 >>   Batch size = 2\n",
            "100% 5666/5666 [2:36:39<00:00,  1.67s/it]11/18/2020 00:06:13 - INFO - __main__ -   ***** Test results *****\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_loss = 2.11389422416687\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_rouge1 = 38.5525\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_rouge2 = 15.35\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_rougeL = 30.6488\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_rougeLsum = 30.7002\n",
            "11/18/2020 00:06:13 - INFO - __main__ -     test_gen_len = 26.0\n",
            "100% 5666/5666 [2:37:53<00:00,  1.67s/it]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8TqUwvnhW5zU",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17
        },
        "outputId": "2f20eee7-4047-4dec-9c04-884963e39295"
      },
      "source": [
        "from google.colab import files\n",
        "import glob\n",
        "\n",
        "output_files = glob.glob(\"/content/transformers/examples/seq2seq/prophetnet_large_uncased_xsum/*\")\n",
        "\n",
        "for filename in output_files:\n",
        "  if \"results\" in filename or \"generation\" in filename:\n",
        "    files.download(filename)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_067686b5-32ae-490d-a62b-74f5b114de12\", \"test_results.json\", 176)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_76421ca0-86e1-4322-a7b7-c5057fea8230\", \"test_generations.txt\", 1309233)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "\n",
              "    async function download(id, filename, size) {\n",
              "      if (!google.colab.kernel.accessAllowed) {\n",
              "        return;\n",
              "      }\n",
              "      const div = document.createElement('div');\n",
              "      const label = document.createElement('label');\n",
              "      label.textContent = `Downloading \"${filename}\": `;\n",
              "      div.appendChild(label);\n",
              "      const progress = document.createElement('progress');\n",
              "      progress.max = size;\n",
              "      div.appendChild(progress);\n",
              "      document.body.appendChild(div);\n",
              "\n",
              "      const buffers = [];\n",
              "      let downloaded = 0;\n",
              "\n",
              "      const channel = await google.colab.kernel.comms.open(id);\n",
              "      // Send a message to notify the kernel that we're ready.\n",
              "      channel.send({})\n",
              "\n",
              "      for await (const message of channel.messages) {\n",
              "        // Send a message to notify the kernel that we're ready.\n",
              "        channel.send({})\n",
              "        if (message.buffers) {\n",
              "          for (const buffer of message.buffers) {\n",
              "            buffers.push(buffer);\n",
              "            downloaded += buffer.byteLength;\n",
              "            progress.value = downloaded;\n",
              "          }\n",
              "        }\n",
              "      }\n",
              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
              "      const a = document.createElement('a');\n",
              "      a.href = window.URL.createObjectURL(blob);\n",
              "      a.download = filename;\n",
              "      div.appendChild(a);\n",
              "      a.click();\n",
              "      div.remove();\n",
              "    }\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "application/javascript": [
              "download(\"download_3e474725-9c47-4f49-9bb1-94b3a7fe2687\", \"eval_results.json\", 197)"
            ],
            "text/plain": [
              "<IPython.core.display.Javascript object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "I8WqDU4YdJlu"
      },
      "source": [
        "## Run BART-base"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n9-NwSpEaKlZ"
      },
      "source": [
        "!python finetune_trainer.py \\\n",
        "    --learning_rate=3e-5 \\\n",
        "    --fp16 \\\n",
        "    --do_train --do_eval --do_predict --evaluate_during_training \\\n",
        "    --predict_with_generate \\\n",
        "    --max_source_length 510 \\\n",
        "    --per_device_train_batch_size 2 \\\n",
        "    --per_device_eval_batch_size 2 \\\n",
        "    --n_train 20400 \\\n",
        "    --n_val 100 \\\n",
        "    --model_name_or_path facebook/bart-base \\\n",
        "    --data_dir $XSUM_DIR \\\n",
        "    --output_dir bart_base_xsum \\\n",
        "    --save_steps 10000 \\\n",
        "    --save_total_limit 1 \\\n",
        "    --overwrite_output_dir "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "us8S5L-uFW77"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}