challenging-america-word-gap-prediction-bigram/zad7.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Collecting torchtext\n",
      "  Downloading torchtext-0.15.2-cp310-cp310-manylinux1_x86_64.whl (2.0 MB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.0/2.0 MB\u001b[0m \u001b[31m1.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m[36m0:00:01\u001b[0m[36m0:00:01\u001b[0m:01\u001b[0m\n",
      "\u001b[?25hCollecting tqdm\n",
      "  Using cached tqdm-4.65.0-py3-none-any.whl (77 kB)\n",
      "Requirement already satisfied: numpy in /home/gedin/.local/lib/python3.10/site-packages (from torchtext) (1.24.3)\n",
      "Collecting torchdata==0.6.1\n",
      "  Downloading torchdata-0.6.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (4.6 MB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m4.6/4.6 MB\u001b[0m \u001b[31m1.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mm eta \u001b[36m0:00:01\u001b[0m[36m0:00:01\u001b[0m\n",
      "\u001b[?25hRequirement already satisfied: requests in /usr/lib/python3/dist-packages (from torchtext) (2.25.1)\n",
      "Collecting torch==2.0.1\n",
      "  Downloading torch-2.0.1-cp310-cp310-manylinux1_x86_64.whl (619.9 MB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m619.9/619.9 MB\u001b[0m \u001b[31m1.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mm eta \u001b[36m0:00:01\u001b[0m[36m0:00:09\u001b[0m\n",
      "\u001b[?25hCollecting sympy\n",
      "  Downloading sympy-1.12-py3-none-any.whl (5.7 MB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m5.7/5.7 MB\u001b[0m \u001b[31m1.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mm eta \u001b[36m0:00:01\u001b[0m[36m0:00:01\u001b[0m\n",
      "\u001b[?25hCollecting nvidia-cudnn-cu11==8.5.0.96\n",
      "  Using cached nvidia_cudnn_cu11-8.5.0.96-2-py3-none-manylinux1_x86_64.whl (557.1 MB)\n",
      "Collecting nvidia-cuda-cupti-cu11==11.7.101\n",
      "  Using cached nvidia_cuda_cupti_cu11-11.7.101-py3-none-manylinux1_x86_64.whl (11.8 MB)\n",
      "Collecting nvidia-cusparse-cu11==11.7.4.91\n",
      "  Using cached nvidia_cusparse_cu11-11.7.4.91-py3-none-manylinux1_x86_64.whl (173.2 MB)\n",
      "Collecting networkx\n",
      "  Using cached networkx-3.1-py3-none-any.whl (2.1 MB)\n",
      "Collecting nvidia-cufft-cu11==10.9.0.58\n",
      "  Using cached nvidia_cufft_cu11-10.9.0.58-py3-none-manylinux1_x86_64.whl (168.4 MB)\n",
      "Collecting filelock\n",
      "  Downloading filelock-3.12.0-py3-none-any.whl (10 kB)\n",
      "Collecting nvidia-cuda-runtime-cu11==11.7.99\n",
      "  Using cached nvidia_cuda_runtime_cu11-11.7.99-py3-none-manylinux1_x86_64.whl (849 kB)\n",
      "Collecting triton==2.0.0\n",
      "  Downloading triton-2.0.0-1-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (63.3 MB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m63.3/63.3 MB\u001b[0m \u001b[31m1.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mm eta \u001b[36m0:00:01\u001b[0m[36m0:00:02\u001b[0m\n",
      "\u001b[?25hCollecting nvidia-cusolver-cu11==11.4.0.1\n",
      "  Using cached nvidia_cusolver_cu11-11.4.0.1-2-py3-none-manylinux1_x86_64.whl (102.6 MB)\n",
      "Requirement already satisfied: jinja2 in /home/gedin/.local/lib/python3.10/site-packages (from torch==2.0.1->torchtext) (3.1.2)\n",
      "Collecting nvidia-cublas-cu11==11.10.3.66\n",
      "  Using cached nvidia_cublas_cu11-11.10.3.66-py3-none-manylinux1_x86_64.whl (317.1 MB)\n",
      "Collecting typing-extensions\n",
      "  Downloading typing_extensions-4.6.3-py3-none-any.whl (31 kB)\n",
      "Collecting nvidia-nccl-cu11==2.14.3\n",
      "  Using cached nvidia_nccl_cu11-2.14.3-py3-none-manylinux1_x86_64.whl (177.1 MB)\n",
      "Collecting nvidia-cuda-nvrtc-cu11==11.7.99\n",
      "  Using cached nvidia_cuda_nvrtc_cu11-11.7.99-2-py3-none-manylinux1_x86_64.whl (21.0 MB)\n",
      "Collecting nvidia-curand-cu11==10.2.10.91\n",
      "  Using cached nvidia_curand_cu11-10.2.10.91-py3-none-manylinux1_x86_64.whl (54.6 MB)\n",
      "Collecting nvidia-nvtx-cu11==11.7.91\n",
      "  Using cached nvidia_nvtx_cu11-11.7.91-py3-none-manylinux1_x86_64.whl (98 kB)\n",
      "Requirement already satisfied: urllib3>=1.25 in /usr/lib/python3/dist-packages (from torchdata==0.6.1->torchtext) (1.26.5)\n",
      "Requirement already satisfied: wheel in /usr/lib/python3/dist-packages (from nvidia-cublas-cu11==11.10.3.66->torch==2.0.1->torchtext) (0.37.1)\n",
      "Requirement already satisfied: setuptools in /usr/lib/python3/dist-packages (from nvidia-cublas-cu11==11.10.3.66->torch==2.0.1->torchtext) (59.6.0)\n",
      "Collecting lit\n",
      "  Downloading lit-16.0.5.tar.gz (138 kB)\n",
      "\u001b[2K     \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m138.0/138.0 KB\u001b[0m \u001b[31m1.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m[31m1.6 MB/s\u001b[0m eta \u001b[36m0:00:01\u001b[0m\n",
      "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25hCollecting cmake\n",
      "  Using cached cmake-3.26.3-py2.py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (24.0 MB)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in /usr/lib/python3/dist-packages (from jinja2->torch==2.0.1->torchtext) (2.0.1)\n",
      "Collecting mpmath>=0.19\n",
      "  Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)\n",
      "Building wheels for collected packages: lit\n",
      "  Building wheel for lit (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25h  Created wheel for lit: filename=lit-16.0.5-py3-none-any.whl size=88192 sha256=f6c57a31a147cbfe0af3d6bf4b856390ad14c28a9ddb38c8044ec29331b35c26\n",
      "  Stored in directory: /home/gedin/.cache/pip/wheels/eb/02/84/d82f0b1a6098209edf7e3607be6cc592ebbc015a8a3127c68d\n",
      "Successfully built lit\n",
      "Installing collected packages: mpmath, lit, cmake, typing-extensions, tqdm, sympy, nvidia-nvtx-cu11, nvidia-nccl-cu11, nvidia-cusparse-cu11, nvidia-curand-cu11, nvidia-cufft-cu11, nvidia-cuda-runtime-cu11, nvidia-cuda-nvrtc-cu11, nvidia-cuda-cupti-cu11, nvidia-cublas-cu11, networkx, filelock, nvidia-cusolver-cu11, nvidia-cudnn-cu11, triton, torch, torchdata, torchtext\n",
      "Successfully installed cmake-3.26.3 filelock-3.12.0 lit-16.0.5 mpmath-1.3.0 networkx-3.1 nvidia-cublas-cu11-11.10.3.66 nvidia-cuda-cupti-cu11-11.7.101 nvidia-cuda-nvrtc-cu11-11.7.99 nvidia-cuda-runtime-cu11-11.7.99 nvidia-cudnn-cu11-8.5.0.96 nvidia-cufft-cu11-10.9.0.58 nvidia-curand-cu11-10.2.10.91 nvidia-cusolver-cu11-11.4.0.1 nvidia-cusparse-cu11-11.7.4.91 nvidia-nccl-cu11-2.14.3 nvidia-nvtx-cu11-11.7.91 sympy-1.12 torch-2.0.1 torchdata-0.6.1 torchtext-0.15.2 tqdm-4.65.0 triton-2.0.0 typing-extensions-4.6.3\n"
     ]
    }
   ],
   "source": [
    "!pip install torchtext"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_file ='train/in.tsv.xz'\n",
    "test_file = 'dev-0/in.tsv.xz'\n",
    "out_file = 'dev-0/out.tsv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import islice\n",
    "import regex as re\n",
    "import sys\n",
    "from torchtext.vocab import build_vocab_from_iterator\n",
    "import lzma\n",
    "import pickle\n",
    "import re\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch.utils.data import IterableDataset\n",
    "import itertools\n",
    "from torch.utils.data import DataLoader\n",
    "import gc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "embed_size = 300\n",
    "device = 'cuda'\n",
    "vocab_size = 25000\n",
    "batch_s = 3200\n",
    "learning_rate = 0.0001\n",
    "epochs = 4\n",
    "k = 20 #top k words\n",
    "wildcard_minweight = 0.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "###preprocessing\n",
    "def preprocess(line):\n",
    "    line = get_rid_of_header(line)\n",
    "    line = replace_endline(line)\n",
    "    return line\n",
    "\n",
    "def get_rid_of_header(line):\n",
    "    line = line.split('\\t')[6:]\n",
    "    return \"\".join(line)\n",
    "    \n",
    "def replace_endline(line):\n",
    "    line = line.replace(\"\\\\n\", \" \")\n",
    "    return line\n",
    "\n",
    "\n",
    "def get_last_word(text):\n",
    "    \"\"\"Return the last word of a string.\"\"\"\n",
    "    last_word = \"\"\n",
    "    for i in range(len(text)-1, -1, -1):\n",
    "        if text[i] == ' ':\n",
    "            return last_word[::-1].rstrip()\n",
    "        else:\n",
    "            last_word += text[i]\n",
    "    return last_word[::-1].rstrip()\n",
    "\n",
    "def get_first_word(text):\n",
    "    \"\"\"Return the first word of a string.\"\"\"\n",
    "    word = \"\"\n",
    "    for i in range(len(text)-1):\n",
    "        if text[i] == ' ':\n",
    "            return word\n",
    "        else:\n",
    "            word += text[i]\n",
    "    return word\n",
    "\n",
    "\n",
    "def get_words_from_line(line):\n",
    "    line = line.rstrip()\n",
    "    yield '<s>'\n",
    "    line = preprocess(line)\n",
    "    for t in line.split(' '):\n",
    "        yield t\n",
    "    yield '</s>'\n",
    "\n",
    "\n",
    "def get_word_lines_from_file(file_name):\n",
    "    n = 0\n",
    "    with lzma.open(file_name, 'r') as fh:\n",
    "        for line in fh:\n",
    "            n+=1\n",
    "            if n%1000==0:\n",
    "                print(n)\n",
    "            yield get_words_from_line(line.decode('utf-8'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000\n",
      "2000\n",
      "3000\n",
      "4000\n",
      "5000\n",
      "6000\n",
      "7000\n",
      "8000\n",
      "9000\n",
      "10000\n",
      "11000\n",
      "12000\n",
      "13000\n",
      "14000\n",
      "15000\n",
      "16000\n",
      "17000\n",
      "18000\n",
      "19000\n",
      "20000\n",
      "21000\n",
      "22000\n",
      "23000\n",
      "24000\n",
      "25000\n",
      "26000\n",
      "27000\n",
      "28000\n",
      "29000\n",
      "30000\n",
      "31000\n",
      "32000\n",
      "33000\n",
      "34000\n",
      "35000\n",
      "36000\n",
      "37000\n",
      "38000\n",
      "39000\n",
      "40000\n",
      "41000\n",
      "42000\n",
      "43000\n",
      "44000\n",
      "45000\n",
      "46000\n",
      "47000\n",
      "48000\n",
      "49000\n",
      "50000\n",
      "51000\n",
      "52000\n",
      "53000\n",
      "54000\n",
      "55000\n",
      "56000\n",
      "57000\n",
      "58000\n",
      "59000\n",
      "60000\n",
      "61000\n",
      "62000\n",
      "63000\n",
      "64000\n",
      "65000\n",
      "66000\n",
      "67000\n",
      "68000\n",
      "69000\n",
      "70000\n",
      "71000\n",
      "72000\n",
      "73000\n",
      "74000\n",
      "75000\n",
      "76000\n",
      "77000\n",
      "78000\n",
      "79000\n",
      "80000\n",
      "81000\n",
      "82000\n",
      "83000\n",
      "84000\n",
      "85000\n",
      "86000\n",
      "87000\n",
      "88000\n",
      "89000\n",
      "90000\n",
      "91000\n",
      "92000\n",
      "93000\n",
      "94000\n",
      "95000\n",
      "96000\n",
      "97000\n",
      "98000\n",
      "99000\n",
      "100000\n",
      "101000\n",
      "102000\n",
      "103000\n",
      "104000\n",
      "105000\n",
      "106000\n",
      "107000\n",
      "108000\n",
      "109000\n",
      "110000\n",
      "111000\n",
      "112000\n",
      "113000\n",
      "114000\n",
      "115000\n",
      "116000\n",
      "117000\n",
      "118000\n",
      "119000\n",
      "120000\n",
      "121000\n",
      "122000\n",
      "123000\n",
      "124000\n",
      "125000\n",
      "126000\n",
      "127000\n",
      "128000\n",
      "129000\n",
      "130000\n",
      "131000\n",
      "132000\n",
      "133000\n",
      "134000\n",
      "135000\n",
      "136000\n",
      "137000\n",
      "138000\n",
      "139000\n",
      "140000\n",
      "141000\n",
      "142000\n",
      "143000\n",
      "144000\n",
      "145000\n",
      "146000\n",
      "147000\n",
      "148000\n",
      "149000\n",
      "150000\n",
      "151000\n",
      "152000\n",
      "153000\n",
      "154000\n",
      "155000\n",
      "156000\n",
      "157000\n",
      "158000\n",
      "159000\n",
      "160000\n",
      "161000\n",
      "162000\n",
      "163000\n",
      "164000\n",
      "165000\n",
      "166000\n",
      "167000\n",
      "168000\n",
      "169000\n",
      "170000\n",
      "171000\n",
      "172000\n",
      "173000\n",
      "174000\n",
      "175000\n",
      "176000\n",
      "177000\n",
      "178000\n",
      "179000\n",
      "180000\n",
      "181000\n",
      "182000\n",
      "183000\n",
      "184000\n",
      "185000\n",
      "186000\n",
      "187000\n",
      "188000\n",
      "189000\n",
      "190000\n",
      "191000\n",
      "192000\n",
      "193000\n",
      "194000\n",
      "195000\n",
      "196000\n",
      "197000\n",
      "198000\n",
      "199000\n",
      "200000\n",
      "201000\n",
      "202000\n",
      "203000\n",
      "204000\n",
      "205000\n",
      "206000\n",
      "207000\n",
      "208000\n",
      "209000\n",
      "210000\n",
      "211000\n",
      "212000\n",
      "213000\n",
      "214000\n",
      "215000\n",
      "216000\n",
      "217000\n",
      "218000\n",
      "219000\n",
      "220000\n",
      "221000\n",
      "222000\n",
      "223000\n",
      "224000\n",
      "225000\n",
      "226000\n",
      "227000\n",
      "228000\n",
      "229000\n",
      "230000\n",
      "231000\n",
      "232000\n",
      "233000\n",
      "234000\n",
      "235000\n",
      "236000\n",
      "237000\n",
      "238000\n",
      "239000\n",
      "240000\n",
      "241000\n",
      "242000\n",
      "243000\n",
      "244000\n",
      "245000\n",
      "246000\n",
      "247000\n",
      "248000\n",
      "249000\n",
      "250000\n",
      "251000\n",
      "252000\n",
      "253000\n",
      "254000\n",
      "255000\n",
      "256000\n",
      "257000\n",
      "258000\n",
      "259000\n",
      "260000\n",
      "261000\n",
      "262000\n",
      "263000\n",
      "264000\n",
      "265000\n",
      "266000\n",
      "267000\n",
      "268000\n",
      "269000\n",
      "270000\n",
      "271000\n",
      "272000\n",
      "273000\n",
      "274000\n",
      "275000\n",
      "276000\n",
      "277000\n",
      "278000\n",
      "279000\n",
      "280000\n",
      "281000\n",
      "282000\n",
      "283000\n",
      "284000\n",
      "285000\n",
      "286000\n",
      "287000\n",
      "288000\n",
      "289000\n",
      "290000\n",
      "291000\n",
      "292000\n",
      "293000\n",
      "294000\n",
      "295000\n",
      "296000\n",
      "297000\n",
      "298000\n",
      "299000\n",
      "300000\n",
      "301000\n",
      "302000\n",
      "303000\n",
      "304000\n",
      "305000\n",
      "306000\n",
      "307000\n",
      "308000\n",
      "309000\n",
      "310000\n",
      "311000\n",
      "312000\n",
      "313000\n",
      "314000\n",
      "315000\n",
      "316000\n",
      "317000\n",
      "318000\n",
      "319000\n",
      "320000\n",
      "321000\n",
      "322000\n",
      "323000\n",
      "324000\n",
      "325000\n",
      "326000\n",
      "327000\n",
      "328000\n",
      "329000\n",
      "330000\n",
      "331000\n",
      "332000\n",
      "333000\n",
      "334000\n",
      "335000\n",
      "336000\n",
      "337000\n",
      "338000\n",
      "339000\n",
      "340000\n",
      "341000\n",
      "342000\n",
      "343000\n",
      "344000\n",
      "345000\n",
      "346000\n",
      "347000\n",
      "348000\n",
      "349000\n",
      "350000\n",
      "351000\n",
      "352000\n",
      "353000\n",
      "354000\n",
      "355000\n",
      "356000\n",
      "357000\n",
      "358000\n",
      "359000\n",
      "360000\n",
      "361000\n",
      "362000\n",
      "363000\n",
      "364000\n",
      "365000\n",
      "366000\n",
      "367000\n",
      "368000\n",
      "369000\n",
      "370000\n",
      "371000\n",
      "372000\n",
      "373000\n",
      "374000\n",
      "375000\n",
      "376000\n",
      "377000\n",
      "378000\n",
      "379000\n",
      "380000\n",
      "381000\n",
      "382000\n",
      "383000\n",
      "384000\n",
      "385000\n",
      "386000\n",
      "387000\n",
      "388000\n",
      "389000\n",
      "390000\n",
      "391000\n",
      "392000\n",
      "393000\n",
      "394000\n",
      "395000\n",
      "396000\n",
      "397000\n",
      "398000\n",
      "399000\n",
      "400000\n",
      "401000\n",
      "402000\n",
      "403000\n",
      "404000\n",
      "405000\n",
      "406000\n",
      "407000\n",
      "408000\n",
      "409000\n",
      "410000\n",
      "411000\n",
      "412000\n",
      "413000\n",
      "414000\n",
      "415000\n",
      "416000\n",
      "417000\n",
      "418000\n",
      "419000\n",
      "420000\n",
      "421000\n",
      "422000\n",
      "423000\n",
      "424000\n",
      "425000\n",
      "426000\n",
      "427000\n",
      "428000\n",
      "429000\n",
      "430000\n",
      "431000\n",
      "432000\n"
     ]
    }
   ],
   "source": [
    "vocab = build_vocab_from_iterator(\n",
    "    get_word_lines_from_file(train_file),\n",
    "    max_tokens = vocab_size,\n",
    "    specials = ['<unk>'])\n",
    "\n",
    "with open('filename.pickle', 'wb') as handle:\n",
    "    pickle.dump(vocab, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['<unk>', 'the', 'of', 'was', 'ladies']"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab.lookup_tokens([0, 1, 2, 10, 2000])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Definicja sieci\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Naszą prostą sieć neuronową zaimplementujemy używając frameworku PyTorch.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "class SimpleBigramNeuralLanguageModel(nn.Module):\n",
    "  def __init__(self, vocabulary_size, embedding_size):\n",
    "      super(SimpleBigramNeuralLanguageModel, self).__init__()\n",
    "      self.model = nn.Sequential(\n",
    "          nn.Embedding(vocabulary_size, embedding_size),\n",
    "          nn.Linear(embedding_size, vocabulary_size),\n",
    "          nn.Softmax()\n",
    "      )\n",
    "    \n",
    "  def forward(self, x):\n",
    "      return self.model(x)\n",
    "\n",
    "with open('filename.pickle','rb') as handle:\n",
    "    vocab = pickle.load(handle)\n",
    "\n",
    "vocab.set_default_index(vocab['<unk>'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on Vocab in module torchtext.vocab.vocab object:\n",
      "\n",
      "class Vocab(torch.nn.modules.module.Module)\n",
      " |  Vocab(vocab) -> None\n",
      " |  \n",
      " |  Base class for all neural network modules.\n",
      " |  \n",
      " |  Your models should also subclass this class.\n",
      " |  \n",
      " |  Modules can also contain other Modules, allowing to nest them in\n",
      " |  a tree structure. You can assign the submodules as regular attributes::\n",
      " |  \n",
      " |      import torch.nn as nn\n",
      " |      import torch.nn.functional as F\n",
      " |  \n",
      " |      class Model(nn.Module):\n",
      " |          def __init__(self):\n",
      " |              super().__init__()\n",
      " |              self.conv1 = nn.Conv2d(1, 20, 5)\n",
      " |              self.conv2 = nn.Conv2d(20, 20, 5)\n",
      " |  \n",
      " |          def forward(self, x):\n",
      " |              x = F.relu(self.conv1(x))\n",
      " |              return F.relu(self.conv2(x))\n",
      " |  \n",
      " |  Submodules assigned in this way will be registered, and will have their\n",
      " |  parameters converted too when you call :meth:`to`, etc.\n",
      " |  \n",
      " |  .. note::\n",
      " |      As per the example above, an ``__init__()`` call to the parent class\n",
      " |      must be made before assignment on the child.\n",
      " |  \n",
      " |  :ivar training: Boolean represents whether this module is in training or\n",
      " |                  evaluation mode.\n",
      " |  :vartype training: bool\n",
      " |  \n",
      " |  Method resolution order:\n",
      " |      Vocab\n",
      " |      torch.nn.modules.module.Module\n",
      " |      builtins.object\n",
      " |  \n",
      " |  Methods defined here:\n",
      " |  \n",
      " |  __contains__(self, token: str) -> bool\n",
      " |      Args:\n",
      " |          token: The token for which to check the membership.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Whether the token is member of vocab or not.\n",
      " |  \n",
      " |  __getitem__(self, token: str) -> int\n",
      " |      Args:\n",
      " |          token: The token used to lookup the corresponding index.\n",
      " |      \n",
      " |      Returns:\n",
      " |          The index corresponding to the associated token.\n",
      " |  \n",
      " |  __init__(self, vocab) -> None\n",
      " |      Initializes internal Module state, shared by both nn.Module and ScriptModule.\n",
      " |  \n",
      " |  __len__(self) -> int\n",
      " |      Returns:\n",
      " |          The length of the vocab.\n",
      " |  \n",
      " |  __prepare_scriptable__(self)\n",
      " |      Return a JITable Vocab.\n",
      " |  \n",
      " |  append_token(self, token: str) -> None\n",
      " |      Args:\n",
      " |          token: The token used to lookup the corresponding index.\n",
      " |      \n",
      " |      Raises:\n",
      " |          RuntimeError: If `token` already exists in the vocab\n",
      " |  \n",
      " |  forward(self, tokens: List[str]) -> List[int]\n",
      " |      Calls the `lookup_indices` method\n",
      " |      \n",
      " |      Args:\n",
      " |          tokens: a list of tokens used to lookup their corresponding `indices`.\n",
      " |      \n",
      " |      Returns:\n",
      " |          The indices associated with a list of `tokens`.\n",
      " |  \n",
      " |  get_default_index(self) -> Union[int, NoneType]\n",
      " |      Returns:\n",
      " |          Value of default index if it is set.\n",
      " |  \n",
      " |  get_itos(self) -> List[str]\n",
      " |      Returns:\n",
      " |          List mapping indices to tokens.\n",
      " |  \n",
      " |  get_stoi(self) -> Dict[str, int]\n",
      " |      Returns:\n",
      " |          Dictionary mapping tokens to indices.\n",
      " |  \n",
      " |  insert_token(self, token: str, index: int) -> None\n",
      " |      Args:\n",
      " |          token: The token used to lookup the corresponding index.\n",
      " |          index: The index corresponding to the associated token.\n",
      " |      Raises:\n",
      " |          RuntimeError: If `index` is not in range [0, Vocab.size()] or if `token` already exists in the vocab.\n",
      " |  \n",
      " |  lookup_indices(self, tokens: List[str]) -> List[int]\n",
      " |      Args:\n",
      " |          tokens: the tokens used to lookup their corresponding `indices`.\n",
      " |      \n",
      " |      Returns:\n",
      " |          The 'indices` associated with `tokens`.\n",
      " |  \n",
      " |  lookup_token(self, index: int) -> str\n",
      " |      Args:\n",
      " |          index: The index corresponding to the associated token.\n",
      " |      \n",
      " |      Returns:\n",
      " |          token: The token used to lookup the corresponding index.\n",
      " |      \n",
      " |      Raises:\n",
      " |          RuntimeError: If `index` not in range [0, itos.size()).\n",
      " |  \n",
      " |  lookup_tokens(self, indices: List[int]) -> List[str]\n",
      " |      Args:\n",
      " |          indices: The `indices` used to lookup their corresponding`tokens`.\n",
      " |      \n",
      " |      Returns:\n",
      " |          The `tokens` associated with `indices`.\n",
      " |      \n",
      " |      Raises:\n",
      " |          RuntimeError: If an index within `indices` is not int range [0, itos.size()).\n",
      " |  \n",
      " |  set_default_index(self, index: Union[int, NoneType]) -> None\n",
      " |      Args:\n",
      " |          index: Value of default index. This index will be returned when OOV token is queried.\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Readonly properties defined here:\n",
      " |  \n",
      " |  is_jitable\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Data and other attributes defined here:\n",
      " |  \n",
      " |  __jit_unused_properties__ = ['is_jitable']\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Methods inherited from torch.nn.modules.module.Module:\n",
      " |  \n",
      " |  __call__ = _call_impl(self, *args, **kwargs)\n",
      " |  \n",
      " |  __delattr__(self, name)\n",
      " |      Implement delattr(self, name).\n",
      " |  \n",
      " |  __dir__(self)\n",
      " |      Default dir() implementation.\n",
      " |  \n",
      " |  __getattr__(self, name: str) -> Union[torch.Tensor, ForwardRef('Module')]\n",
      " |  \n",
      " |  __repr__(self)\n",
      " |      Return repr(self).\n",
      " |  \n",
      " |  __setattr__(self, name: str, value: Union[torch.Tensor, ForwardRef('Module')]) -> None\n",
      " |      Implement setattr(self, name, value).\n",
      " |  \n",
      " |  __setstate__(self, state)\n",
      " |  \n",
      " |  add_module(self, name: str, module: Union[ForwardRef('Module'), NoneType]) -> None\n",
      " |      Adds a child module to the current module.\n",
      " |      \n",
      " |      The module can be accessed as an attribute using the given name.\n",
      " |      \n",
      " |      Args:\n",
      " |          name (str): name of the child module. The child module can be\n",
      " |              accessed from this module using the given name\n",
      " |          module (Module): child module to be added to the module.\n",
      " |  \n",
      " |  apply(self: ~T, fn: Callable[[ForwardRef('Module')], NoneType]) -> ~T\n",
      " |      Applies ``fn`` recursively to every submodule (as returned by ``.children()``)\n",
      " |      as well as self. Typical use includes initializing the parameters of a model\n",
      " |      (see also :ref:`nn-init-doc`).\n",
      " |      \n",
      " |      Args:\n",
      " |          fn (:class:`Module` -> None): function to be applied to each submodule\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> @torch.no_grad()\n",
      " |          >>> def init_weights(m):\n",
      " |          >>>     print(m)\n",
      " |          >>>     if type(m) == nn.Linear:\n",
      " |          >>>         m.weight.fill_(1.0)\n",
      " |          >>>         print(m.weight)\n",
      " |          >>> net = nn.Sequential(nn.Linear(2, 2), nn.Linear(2, 2))\n",
      " |          >>> net.apply(init_weights)\n",
      " |          Linear(in_features=2, out_features=2, bias=True)\n",
      " |          Parameter containing:\n",
      " |          tensor([[1., 1.],\n",
      " |                  [1., 1.]], requires_grad=True)\n",
      " |          Linear(in_features=2, out_features=2, bias=True)\n",
      " |          Parameter containing:\n",
      " |          tensor([[1., 1.],\n",
      " |                  [1., 1.]], requires_grad=True)\n",
      " |          Sequential(\n",
      " |            (0): Linear(in_features=2, out_features=2, bias=True)\n",
      " |            (1): Linear(in_features=2, out_features=2, bias=True)\n",
      " |          )\n",
      " |  \n",
      " |  bfloat16(self: ~T) -> ~T\n",
      " |      Casts all floating point parameters and buffers to ``bfloat16`` datatype.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  buffers(self, recurse: bool = True) -> Iterator[torch.Tensor]\n",
      " |      Returns an iterator over module buffers.\n",
      " |      \n",
      " |      Args:\n",
      " |          recurse (bool): if True, then yields buffers of this module\n",
      " |              and all submodules. Otherwise, yields only buffers that\n",
      " |              are direct members of this module.\n",
      " |      \n",
      " |      Yields:\n",
      " |          torch.Tensor: module buffer\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> for buf in model.buffers():\n",
      " |          >>>     print(type(buf), buf.size())\n",
      " |          <class 'torch.Tensor'> (20L,)\n",
      " |          <class 'torch.Tensor'> (20L, 1L, 5L, 5L)\n",
      " |  \n",
      " |  children(self) -> Iterator[ForwardRef('Module')]\n",
      " |      Returns an iterator over immediate children modules.\n",
      " |      \n",
      " |      Yields:\n",
      " |          Module: a child module\n",
      " |  \n",
      " |  cpu(self: ~T) -> ~T\n",
      " |      Moves all model parameters and buffers to the CPU.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  cuda(self: ~T, device: Union[int, torch.device, NoneType] = None) -> ~T\n",
      " |      Moves all model parameters and buffers to the GPU.\n",
      " |      \n",
      " |      This also makes associated parameters and buffers different objects. So\n",
      " |      it should be called before constructing optimizer if the module will\n",
      " |      live on GPU while being optimized.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Args:\n",
      " |          device (int, optional): if specified, all parameters will be\n",
      " |              copied to that device\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  double(self: ~T) -> ~T\n",
      " |      Casts all floating point parameters and buffers to ``double`` datatype.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  eval(self: ~T) -> ~T\n",
      " |      Sets the module in evaluation mode.\n",
      " |      \n",
      " |      This has any effect only on certain modules. See documentations of\n",
      " |      particular modules for details of their behaviors in training/evaluation\n",
      " |      mode, if they are affected, e.g. :class:`Dropout`, :class:`BatchNorm`,\n",
      " |      etc.\n",
      " |      \n",
      " |      This is equivalent with :meth:`self.train(False) <torch.nn.Module.train>`.\n",
      " |      \n",
      " |      See :ref:`locally-disable-grad-doc` for a comparison between\n",
      " |      `.eval()` and several similar mechanisms that may be confused with it.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  extra_repr(self) -> str\n",
      " |      Set the extra representation of the module\n",
      " |      \n",
      " |      To print customized extra information, you should re-implement\n",
      " |      this method in your own modules. Both single-line and multi-line\n",
      " |      strings are acceptable.\n",
      " |  \n",
      " |  float(self: ~T) -> ~T\n",
      " |      Casts all floating point parameters and buffers to ``float`` datatype.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  get_buffer(self, target: str) -> 'Tensor'\n",
      " |      Returns the buffer given by ``target`` if it exists,\n",
      " |      otherwise throws an error.\n",
      " |      \n",
      " |      See the docstring for ``get_submodule`` for a more detailed\n",
      " |      explanation of this method's functionality as well as how to\n",
      " |      correctly specify ``target``.\n",
      " |      \n",
      " |      Args:\n",
      " |          target: The fully-qualified string name of the buffer\n",
      " |              to look for. (See ``get_submodule`` for how to specify a\n",
      " |              fully-qualified string.)\n",
      " |      \n",
      " |      Returns:\n",
      " |          torch.Tensor: The buffer referenced by ``target``\n",
      " |      \n",
      " |      Raises:\n",
      " |          AttributeError: If the target string references an invalid\n",
      " |              path or resolves to something that is not a\n",
      " |              buffer\n",
      " |  \n",
      " |  get_extra_state(self) -> Any\n",
      " |      Returns any extra state to include in the module's state_dict.\n",
      " |      Implement this and a corresponding :func:`set_extra_state` for your module\n",
      " |      if you need to store extra state. This function is called when building the\n",
      " |      module's `state_dict()`.\n",
      " |      \n",
      " |      Note that extra state should be picklable to ensure working serialization\n",
      " |      of the state_dict. We only provide provide backwards compatibility guarantees\n",
      " |      for serializing Tensors; other objects may break backwards compatibility if\n",
      " |      their serialized pickled form changes.\n",
      " |      \n",
      " |      Returns:\n",
      " |          object: Any extra state to store in the module's state_dict\n",
      " |  \n",
      " |  get_parameter(self, target: str) -> 'Parameter'\n",
      " |      Returns the parameter given by ``target`` if it exists,\n",
      " |      otherwise throws an error.\n",
      " |      \n",
      " |      See the docstring for ``get_submodule`` for a more detailed\n",
      " |      explanation of this method's functionality as well as how to\n",
      " |      correctly specify ``target``.\n",
      " |      \n",
      " |      Args:\n",
      " |          target: The fully-qualified string name of the Parameter\n",
      " |              to look for. (See ``get_submodule`` for how to specify a\n",
      " |              fully-qualified string.)\n",
      " |      \n",
      " |      Returns:\n",
      " |          torch.nn.Parameter: The Parameter referenced by ``target``\n",
      " |      \n",
      " |      Raises:\n",
      " |          AttributeError: If the target string references an invalid\n",
      " |              path or resolves to something that is not an\n",
      " |              ``nn.Parameter``\n",
      " |  \n",
      " |  get_submodule(self, target: str) -> 'Module'\n",
      " |      Returns the submodule given by ``target`` if it exists,\n",
      " |      otherwise throws an error.\n",
      " |      \n",
      " |      For example, let's say you have an ``nn.Module`` ``A`` that\n",
      " |      looks like this:\n",
      " |      \n",
      " |      .. code-block:: text\n",
      " |      \n",
      " |          A(\n",
      " |              (net_b): Module(\n",
      " |                  (net_c): Module(\n",
      " |                      (conv): Conv2d(16, 33, kernel_size=(3, 3), stride=(2, 2))\n",
      " |                  )\n",
      " |                  (linear): Linear(in_features=100, out_features=200, bias=True)\n",
      " |              )\n",
      " |          )\n",
      " |      \n",
      " |      (The diagram shows an ``nn.Module`` ``A``. ``A`` has a nested\n",
      " |      submodule ``net_b``, which itself has two submodules ``net_c``\n",
      " |      and ``linear``. ``net_c`` then has a submodule ``conv``.)\n",
      " |      \n",
      " |      To check whether or not we have the ``linear`` submodule, we\n",
      " |      would call ``get_submodule(\"net_b.linear\")``. To check whether\n",
      " |      we have the ``conv`` submodule, we would call\n",
      " |      ``get_submodule(\"net_b.net_c.conv\")``.\n",
      " |      \n",
      " |      The runtime of ``get_submodule`` is bounded by the degree\n",
      " |      of module nesting in ``target``. A query against\n",
      " |      ``named_modules`` achieves the same result, but it is O(N) in\n",
      " |      the number of transitive modules. So, for a simple check to see\n",
      " |      if some submodule exists, ``get_submodule`` should always be\n",
      " |      used.\n",
      " |      \n",
      " |      Args:\n",
      " |          target: The fully-qualified string name of the submodule\n",
      " |              to look for. (See above example for how to specify a\n",
      " |              fully-qualified string.)\n",
      " |      \n",
      " |      Returns:\n",
      " |          torch.nn.Module: The submodule referenced by ``target``\n",
      " |      \n",
      " |      Raises:\n",
      " |          AttributeError: If the target string references an invalid\n",
      " |              path or resolves to something that is not an\n",
      " |              ``nn.Module``\n",
      " |  \n",
      " |  half(self: ~T) -> ~T\n",
      " |      Casts all floating point parameters and buffers to ``half`` datatype.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  ipu(self: ~T, device: Union[int, torch.device, NoneType] = None) -> ~T\n",
      " |      Moves all model parameters and buffers to the IPU.\n",
      " |      \n",
      " |      This also makes associated parameters and buffers different objects. So\n",
      " |      it should be called before constructing optimizer if the module will\n",
      " |      live on IPU while being optimized.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Arguments:\n",
      " |          device (int, optional): if specified, all parameters will be\n",
      " |              copied to that device\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  load_state_dict(self, state_dict: Mapping[str, Any], strict: bool = True)\n",
      " |      Copies parameters and buffers from :attr:`state_dict` into\n",
      " |      this module and its descendants. If :attr:`strict` is ``True``, then\n",
      " |      the keys of :attr:`state_dict` must exactly match the keys returned\n",
      " |      by this module's :meth:`~torch.nn.Module.state_dict` function.\n",
      " |      \n",
      " |      Args:\n",
      " |          state_dict (dict): a dict containing parameters and\n",
      " |              persistent buffers.\n",
      " |          strict (bool, optional): whether to strictly enforce that the keys\n",
      " |              in :attr:`state_dict` match the keys returned by this module's\n",
      " |              :meth:`~torch.nn.Module.state_dict` function. Default: ``True``\n",
      " |      \n",
      " |      Returns:\n",
      " |          ``NamedTuple`` with ``missing_keys`` and ``unexpected_keys`` fields:\n",
      " |              * **missing_keys** is a list of str containing the missing keys\n",
      " |              * **unexpected_keys** is a list of str containing the unexpected keys\n",
      " |      \n",
      " |      Note:\n",
      " |          If a parameter or buffer is registered as ``None`` and its corresponding key\n",
      " |          exists in :attr:`state_dict`, :meth:`load_state_dict` will raise a\n",
      " |          ``RuntimeError``.\n",
      " |  \n",
      " |  modules(self) -> Iterator[ForwardRef('Module')]\n",
      " |      Returns an iterator over all modules in the network.\n",
      " |      \n",
      " |      Yields:\n",
      " |          Module: a module in the network\n",
      " |      \n",
      " |      Note:\n",
      " |          Duplicate modules are returned only once. In the following\n",
      " |          example, ``l`` will be returned only once.\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> l = nn.Linear(2, 2)\n",
      " |          >>> net = nn.Sequential(l, l)\n",
      " |          >>> for idx, m in enumerate(net.modules()):\n",
      " |          ...     print(idx, '->', m)\n",
      " |      \n",
      " |          0 -> Sequential(\n",
      " |            (0): Linear(in_features=2, out_features=2, bias=True)\n",
      " |            (1): Linear(in_features=2, out_features=2, bias=True)\n",
      " |          )\n",
      " |          1 -> Linear(in_features=2, out_features=2, bias=True)\n",
      " |  \n",
      " |  named_buffers(self, prefix: str = '', recurse: bool = True, remove_duplicate: bool = True) -> Iterator[Tuple[str, torch.Tensor]]\n",
      " |      Returns an iterator over module buffers, yielding both the\n",
      " |      name of the buffer as well as the buffer itself.\n",
      " |      \n",
      " |      Args:\n",
      " |          prefix (str): prefix to prepend to all buffer names.\n",
      " |          recurse (bool, optional): if True, then yields buffers of this module\n",
      " |              and all submodules. Otherwise, yields only buffers that\n",
      " |              are direct members of this module. Defaults to True.\n",
      " |          remove_duplicate (bool, optional): whether to remove the duplicated buffers in the result. Defaults to True.\n",
      " |      \n",
      " |      Yields:\n",
      " |          (str, torch.Tensor): Tuple containing the name and buffer\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> for name, buf in self.named_buffers():\n",
      " |          >>>     if name in ['running_var']:\n",
      " |          >>>         print(buf.size())\n",
      " |  \n",
      " |  named_children(self) -> Iterator[Tuple[str, ForwardRef('Module')]]\n",
      " |      Returns an iterator over immediate children modules, yielding both\n",
      " |      the name of the module as well as the module itself.\n",
      " |      \n",
      " |      Yields:\n",
      " |          (str, Module): Tuple containing a name and child module\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> for name, module in model.named_children():\n",
      " |          >>>     if name in ['conv4', 'conv5']:\n",
      " |          >>>         print(module)\n",
      " |  \n",
      " |  named_modules(self, memo: Union[Set[ForwardRef('Module')], NoneType] = None, prefix: str = '', remove_duplicate: bool = True)\n",
      " |      Returns an iterator over all modules in the network, yielding\n",
      " |      both the name of the module as well as the module itself.\n",
      " |      \n",
      " |      Args:\n",
      " |          memo: a memo to store the set of modules already added to the result\n",
      " |          prefix: a prefix that will be added to the name of the module\n",
      " |          remove_duplicate: whether to remove the duplicated module instances in the result\n",
      " |              or not\n",
      " |      \n",
      " |      Yields:\n",
      " |          (str, Module): Tuple of name and module\n",
      " |      \n",
      " |      Note:\n",
      " |          Duplicate modules are returned only once. In the following\n",
      " |          example, ``l`` will be returned only once.\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> l = nn.Linear(2, 2)\n",
      " |          >>> net = nn.Sequential(l, l)\n",
      " |          >>> for idx, m in enumerate(net.named_modules()):\n",
      " |          ...     print(idx, '->', m)\n",
      " |      \n",
      " |          0 -> ('', Sequential(\n",
      " |            (0): Linear(in_features=2, out_features=2, bias=True)\n",
      " |            (1): Linear(in_features=2, out_features=2, bias=True)\n",
      " |          ))\n",
      " |          1 -> ('0', Linear(in_features=2, out_features=2, bias=True))\n",
      " |  \n",
      " |  named_parameters(self, prefix: str = '', recurse: bool = True, remove_duplicate: bool = True) -> Iterator[Tuple[str, torch.nn.parameter.Parameter]]\n",
      " |      Returns an iterator over module parameters, yielding both the\n",
      " |      name of the parameter as well as the parameter itself.\n",
      " |      \n",
      " |      Args:\n",
      " |          prefix (str): prefix to prepend to all parameter names.\n",
      " |          recurse (bool): if True, then yields parameters of this module\n",
      " |              and all submodules. Otherwise, yields only parameters that\n",
      " |              are direct members of this module.\n",
      " |          remove_duplicate (bool, optional): whether to remove the duplicated\n",
      " |              parameters in the result. Defaults to True.\n",
      " |      \n",
      " |      Yields:\n",
      " |          (str, Parameter): Tuple containing the name and parameter\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> for name, param in self.named_parameters():\n",
      " |          >>>     if name in ['bias']:\n",
      " |          >>>         print(param.size())\n",
      " |  \n",
      " |  parameters(self, recurse: bool = True) -> Iterator[torch.nn.parameter.Parameter]\n",
      " |      Returns an iterator over module parameters.\n",
      " |      \n",
      " |      This is typically passed to an optimizer.\n",
      " |      \n",
      " |      Args:\n",
      " |          recurse (bool): if True, then yields parameters of this module\n",
      " |              and all submodules. Otherwise, yields only parameters that\n",
      " |              are direct members of this module.\n",
      " |      \n",
      " |      Yields:\n",
      " |          Parameter: module parameter\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> for param in model.parameters():\n",
      " |          >>>     print(type(param), param.size())\n",
      " |          <class 'torch.Tensor'> (20L,)\n",
      " |          <class 'torch.Tensor'> (20L, 1L, 5L, 5L)\n",
      " |  \n",
      " |  register_backward_hook(self, hook: Callable[[ForwardRef('Module'), Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[NoneType, Tuple[torch.Tensor, ...], torch.Tensor]]) -> torch.utils.hooks.RemovableHandle\n",
      " |      Registers a backward hook on the module.\n",
      " |      \n",
      " |      This function is deprecated in favor of :meth:`~torch.nn.Module.register_full_backward_hook` and\n",
      " |      the behavior of this function will change in future versions.\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_buffer(self, name: str, tensor: Union[torch.Tensor, NoneType], persistent: bool = True) -> None\n",
      " |      Adds a buffer to the module.\n",
      " |      \n",
      " |      This is typically used to register a buffer that should not to be\n",
      " |      considered a model parameter. For example, BatchNorm's ``running_mean``\n",
      " |      is not a parameter, but is part of the module's state. Buffers, by\n",
      " |      default, are persistent and will be saved alongside parameters. This\n",
      " |      behavior can be changed by setting :attr:`persistent` to ``False``. The\n",
      " |      only difference between a persistent buffer and a non-persistent buffer\n",
      " |      is that the latter will not be a part of this module's\n",
      " |      :attr:`state_dict`.\n",
      " |      \n",
      " |      Buffers can be accessed as attributes using given names.\n",
      " |      \n",
      " |      Args:\n",
      " |          name (str): name of the buffer. The buffer can be accessed\n",
      " |              from this module using the given name\n",
      " |          tensor (Tensor or None): buffer to be registered. If ``None``, then operations\n",
      " |              that run on buffers, such as :attr:`cuda`, are ignored. If ``None``,\n",
      " |              the buffer is **not** included in the module's :attr:`state_dict`.\n",
      " |          persistent (bool): whether the buffer is part of this module's\n",
      " |              :attr:`state_dict`.\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> self.register_buffer('running_mean', torch.zeros(num_features))\n",
      " |  \n",
      " |  register_forward_hook(self, hook: Union[Callable[[~T, Tuple[Any, ...], Any], Union[Any, NoneType]], Callable[[~T, Tuple[Any, ...], Dict[str, Any], Any], Union[Any, NoneType]]], *, prepend: bool = False, with_kwargs: bool = False) -> torch.utils.hooks.RemovableHandle\n",
      " |      Registers a forward hook on the module.\n",
      " |      \n",
      " |      The hook will be called every time after :func:`forward` has computed an output.\n",
      " |      \n",
      " |      If ``with_kwargs`` is ``False`` or not specified, the input contains only\n",
      " |      the positional arguments given to the module. Keyword arguments won't be\n",
      " |      passed to the hooks and only to the ``forward``. The hook can modify the\n",
      " |      output. It can modify the input inplace but it will not have effect on\n",
      " |      forward since this is called after :func:`forward` is called. The hook\n",
      " |      should have the following signature::\n",
      " |      \n",
      " |          hook(module, args, output) -> None or modified output\n",
      " |      \n",
      " |      If ``with_kwargs`` is ``True``, the forward hook will be passed the\n",
      " |      ``kwargs`` given to the forward function and be expected to return the\n",
      " |      output possibly modified. The hook should have the following signature::\n",
      " |      \n",
      " |          hook(module, args, kwargs, output) -> None or modified output\n",
      " |      \n",
      " |      Args:\n",
      " |          hook (Callable): The user defined hook to be registered.\n",
      " |          prepend (bool): If ``True``, the provided ``hook`` will be fired\n",
      " |              before all existing ``forward`` hooks on this\n",
      " |              :class:`torch.nn.modules.Module`. Otherwise, the provided\n",
      " |              ``hook`` will be fired after all existing ``forward`` hooks on\n",
      " |              this :class:`torch.nn.modules.Module`. Note that global\n",
      " |              ``forward`` hooks registered with\n",
      " |              :func:`register_module_forward_hook` will fire before all hooks\n",
      " |              registered by this method.\n",
      " |              Default: ``False``\n",
      " |          with_kwargs (bool): If ``True``, the ``hook`` will be passed the\n",
      " |              kwargs given to the forward function.\n",
      " |              Default: ``False``\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_forward_pre_hook(self, hook: Union[Callable[[~T, Tuple[Any, ...]], Union[Any, NoneType]], Callable[[~T, Tuple[Any, ...], Dict[str, Any]], Union[Tuple[Any, Dict[str, Any]], NoneType]]], *, prepend: bool = False, with_kwargs: bool = False) -> torch.utils.hooks.RemovableHandle\n",
      " |      Registers a forward pre-hook on the module.\n",
      " |      \n",
      " |      The hook will be called every time before :func:`forward` is invoked.\n",
      " |      \n",
      " |      \n",
      " |      If ``with_kwargs`` is false or not specified, the input contains only\n",
      " |      the positional arguments given to the module. Keyword arguments won't be\n",
      " |      passed to the hooks and only to the ``forward``. The hook can modify the\n",
      " |      input. User can either return a tuple or a single modified value in the\n",
      " |      hook. We will wrap the value into a tuple if a single value is returned\n",
      " |      (unless that value is already a tuple). The hook should have the\n",
      " |      following signature::\n",
      " |      \n",
      " |          hook(module, args) -> None or modified input\n",
      " |      \n",
      " |      If ``with_kwargs`` is true, the forward pre-hook will be passed the\n",
      " |      kwargs given to the forward function. And if the hook modifies the\n",
      " |      input, both the args and kwargs should be returned. The hook should have\n",
      " |      the following signature::\n",
      " |      \n",
      " |          hook(module, args, kwargs) -> None or a tuple of modified input and kwargs\n",
      " |      \n",
      " |      Args:\n",
      " |          hook (Callable): The user defined hook to be registered.\n",
      " |          prepend (bool): If true, the provided ``hook`` will be fired before\n",
      " |              all existing ``forward_pre`` hooks on this\n",
      " |              :class:`torch.nn.modules.Module`. Otherwise, the provided\n",
      " |              ``hook`` will be fired after all existing ``forward_pre`` hooks\n",
      " |              on this :class:`torch.nn.modules.Module`. Note that global\n",
      " |              ``forward_pre`` hooks registered with\n",
      " |              :func:`register_module_forward_pre_hook` will fire before all\n",
      " |              hooks registered by this method.\n",
      " |              Default: ``False``\n",
      " |          with_kwargs (bool): If true, the ``hook`` will be passed the kwargs\n",
      " |              given to the forward function.\n",
      " |              Default: ``False``\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_full_backward_hook(self, hook: Callable[[ForwardRef('Module'), Union[Tuple[torch.Tensor, ...], torch.Tensor], Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[NoneType, Tuple[torch.Tensor, ...], torch.Tensor]], prepend: bool = False) -> torch.utils.hooks.RemovableHandle\n",
      " |      Registers a backward hook on the module.\n",
      " |      \n",
      " |      The hook will be called every time the gradients with respect to a module\n",
      " |      are computed, i.e. the hook will execute if and only if the gradients with\n",
      " |      respect to module outputs are computed. The hook should have the following\n",
      " |      signature::\n",
      " |      \n",
      " |          hook(module, grad_input, grad_output) -> tuple(Tensor) or None\n",
      " |      \n",
      " |      The :attr:`grad_input` and :attr:`grad_output` are tuples that contain the gradients\n",
      " |      with respect to the inputs and outputs respectively. The hook should\n",
      " |      not modify its arguments, but it can optionally return a new gradient with\n",
      " |      respect to the input that will be used in place of :attr:`grad_input` in\n",
      " |      subsequent computations. :attr:`grad_input` will only correspond to the inputs given\n",
      " |      as positional arguments and all kwarg arguments are ignored. Entries\n",
      " |      in :attr:`grad_input` and :attr:`grad_output` will be ``None`` for all non-Tensor\n",
      " |      arguments.\n",
      " |      \n",
      " |      For technical reasons, when this hook is applied to a Module, its forward function will\n",
      " |      receive a view of each Tensor passed to the Module. Similarly the caller will receive a view\n",
      " |      of each Tensor returned by the Module's forward function.\n",
      " |      \n",
      " |      .. warning ::\n",
      " |          Modifying inputs or outputs inplace is not allowed when using backward hooks and\n",
      " |          will raise an error.\n",
      " |      \n",
      " |      Args:\n",
      " |          hook (Callable): The user-defined hook to be registered.\n",
      " |          prepend (bool): If true, the provided ``hook`` will be fired before\n",
      " |              all existing ``backward`` hooks on this\n",
      " |              :class:`torch.nn.modules.Module`. Otherwise, the provided\n",
      " |              ``hook`` will be fired after all existing ``backward`` hooks on\n",
      " |              this :class:`torch.nn.modules.Module`. Note that global\n",
      " |              ``backward`` hooks registered with\n",
      " |              :func:`register_module_full_backward_hook` will fire before\n",
      " |              all hooks registered by this method.\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_full_backward_pre_hook(self, hook: Callable[[ForwardRef('Module'), Union[Tuple[torch.Tensor, ...], torch.Tensor]], Union[NoneType, Tuple[torch.Tensor, ...], torch.Tensor]], prepend: bool = False) -> torch.utils.hooks.RemovableHandle\n",
      " |      Registers a backward pre-hook on the module.\n",
      " |      \n",
      " |      The hook will be called every time the gradients for the module are computed.\n",
      " |      The hook should have the following signature::\n",
      " |      \n",
      " |          hook(module, grad_output) -> Tensor or None\n",
      " |      \n",
      " |      The :attr:`grad_output` is a tuple. The hook should\n",
      " |      not modify its arguments, but it can optionally return a new gradient with\n",
      " |      respect to the output that will be used in place of :attr:`grad_output` in\n",
      " |      subsequent computations. Entries in :attr:`grad_output` will be ``None`` for\n",
      " |      all non-Tensor arguments.\n",
      " |      \n",
      " |      For technical reasons, when this hook is applied to a Module, its forward function will\n",
      " |      receive a view of each Tensor passed to the Module. Similarly the caller will receive a view\n",
      " |      of each Tensor returned by the Module's forward function.\n",
      " |      \n",
      " |      .. warning ::\n",
      " |          Modifying inputs inplace is not allowed when using backward hooks and\n",
      " |          will raise an error.\n",
      " |      \n",
      " |      Args:\n",
      " |          hook (Callable): The user-defined hook to be registered.\n",
      " |          prepend (bool): If true, the provided ``hook`` will be fired before\n",
      " |              all existing ``backward_pre`` hooks on this\n",
      " |              :class:`torch.nn.modules.Module`. Otherwise, the provided\n",
      " |              ``hook`` will be fired after all existing ``backward_pre`` hooks\n",
      " |              on this :class:`torch.nn.modules.Module`. Note that global\n",
      " |              ``backward_pre`` hooks registered with\n",
      " |              :func:`register_module_full_backward_pre_hook` will fire before\n",
      " |              all hooks registered by this method.\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_load_state_dict_post_hook(self, hook)\n",
      " |      Registers a post hook to be run after module's ``load_state_dict``\n",
      " |      is called.\n",
      " |      \n",
      " |      It should have the following signature::\n",
      " |          hook(module, incompatible_keys) -> None\n",
      " |      \n",
      " |      The ``module`` argument is the current module that this hook is registered\n",
      " |      on, and the ``incompatible_keys`` argument is a ``NamedTuple`` consisting\n",
      " |      of attributes ``missing_keys`` and ``unexpected_keys``. ``missing_keys``\n",
      " |      is a ``list`` of ``str`` containing the missing keys and\n",
      " |      ``unexpected_keys`` is a ``list`` of ``str`` containing the unexpected keys.\n",
      " |      \n",
      " |      The given incompatible_keys can be modified inplace if needed.\n",
      " |      \n",
      " |      Note that the checks performed when calling :func:`load_state_dict` with\n",
      " |      ``strict=True`` are affected by modifications the hook makes to\n",
      " |      ``missing_keys`` or ``unexpected_keys``, as expected. Additions to either\n",
      " |      set of keys will result in an error being thrown when ``strict=True``, and\n",
      " |      clearing out both missing and unexpected keys will avoid an error.\n",
      " |      \n",
      " |      Returns:\n",
      " |          :class:`torch.utils.hooks.RemovableHandle`:\n",
      " |              a handle that can be used to remove the added hook by calling\n",
      " |              ``handle.remove()``\n",
      " |  \n",
      " |  register_module(self, name: str, module: Union[ForwardRef('Module'), NoneType]) -> None\n",
      " |      Alias for :func:`add_module`.\n",
      " |  \n",
      " |  register_parameter(self, name: str, param: Union[torch.nn.parameter.Parameter, NoneType]) -> None\n",
      " |      Adds a parameter to the module.\n",
      " |      \n",
      " |      The parameter can be accessed as an attribute using given name.\n",
      " |      \n",
      " |      Args:\n",
      " |          name (str): name of the parameter. The parameter can be accessed\n",
      " |              from this module using the given name\n",
      " |          param (Parameter or None): parameter to be added to the module. If\n",
      " |              ``None``, then operations that run on parameters, such as :attr:`cuda`,\n",
      " |              are ignored. If ``None``, the parameter is **not** included in the\n",
      " |              module's :attr:`state_dict`.\n",
      " |  \n",
      " |  register_state_dict_pre_hook(self, hook)\n",
      " |      These hooks will be called with arguments: ``self``, ``prefix``,\n",
      " |      and ``keep_vars`` before calling ``state_dict`` on ``self``. The registered\n",
      " |      hooks can be used to perform pre-processing before the ``state_dict``\n",
      " |      call is made.\n",
      " |  \n",
      " |  requires_grad_(self: ~T, requires_grad: bool = True) -> ~T\n",
      " |      Change if autograd should record operations on parameters in this\n",
      " |      module.\n",
      " |      \n",
      " |      This method sets the parameters' :attr:`requires_grad` attributes\n",
      " |      in-place.\n",
      " |      \n",
      " |      This method is helpful for freezing part of the module for finetuning\n",
      " |      or training parts of a model individually (e.g., GAN training).\n",
      " |      \n",
      " |      See :ref:`locally-disable-grad-doc` for a comparison between\n",
      " |      `.requires_grad_()` and several similar mechanisms that may be confused with it.\n",
      " |      \n",
      " |      Args:\n",
      " |          requires_grad (bool): whether autograd should record operations on\n",
      " |                                parameters in this module. Default: ``True``.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  set_extra_state(self, state: Any)\n",
      " |      This function is called from :func:`load_state_dict` to handle any extra state\n",
      " |      found within the `state_dict`. Implement this function and a corresponding\n",
      " |      :func:`get_extra_state` for your module if you need to store extra state within its\n",
      " |      `state_dict`.\n",
      " |      \n",
      " |      Args:\n",
      " |          state (dict): Extra state from the `state_dict`\n",
      " |  \n",
      " |  share_memory(self: ~T) -> ~T\n",
      " |      See :meth:`torch.Tensor.share_memory_`\n",
      " |  \n",
      " |  state_dict(self, *args, destination=None, prefix='', keep_vars=False)\n",
      " |      Returns a dictionary containing references to the whole state of the module.\n",
      " |      \n",
      " |      Both parameters and persistent buffers (e.g. running averages) are\n",
      " |      included. Keys are corresponding parameter and buffer names.\n",
      " |      Parameters and buffers set to ``None`` are not included.\n",
      " |      \n",
      " |      .. note::\n",
      " |          The returned object is a shallow copy. It contains references\n",
      " |          to the module's parameters and buffers.\n",
      " |      \n",
      " |      .. warning::\n",
      " |          Currently ``state_dict()`` also accepts positional arguments for\n",
      " |          ``destination``, ``prefix`` and ``keep_vars`` in order. However,\n",
      " |          this is being deprecated and keyword arguments will be enforced in\n",
      " |          future releases.\n",
      " |      \n",
      " |      .. warning::\n",
      " |          Please avoid the use of argument ``destination`` as it is not\n",
      " |          designed for end-users.\n",
      " |      \n",
      " |      Args:\n",
      " |          destination (dict, optional): If provided, the state of module will\n",
      " |              be updated into the dict and the same object is returned.\n",
      " |              Otherwise, an ``OrderedDict`` will be created and returned.\n",
      " |              Default: ``None``.\n",
      " |          prefix (str, optional): a prefix added to parameter and buffer\n",
      " |              names to compose the keys in state_dict. Default: ``''``.\n",
      " |          keep_vars (bool, optional): by default the :class:`~torch.Tensor` s\n",
      " |              returned in the state dict are detached from autograd. If it's\n",
      " |              set to ``True``, detaching will not be performed.\n",
      " |              Default: ``False``.\n",
      " |      \n",
      " |      Returns:\n",
      " |          dict:\n",
      " |              a dictionary containing a whole state of the module\n",
      " |      \n",
      " |      Example::\n",
      " |      \n",
      " |          >>> # xdoctest: +SKIP(\"undefined vars\")\n",
      " |          >>> module.state_dict().keys()\n",
      " |          ['bias', 'weight']\n",
      " |  \n",
      " |  to(self, *args, **kwargs)\n",
      " |      Moves and/or casts the parameters and buffers.\n",
      " |      \n",
      " |      This can be called as\n",
      " |      \n",
      " |      .. function:: to(device=None, dtype=None, non_blocking=False)\n",
      " |         :noindex:\n",
      " |      \n",
      " |      .. function:: to(dtype, non_blocking=False)\n",
      " |         :noindex:\n",
      " |      \n",
      " |      .. function:: to(tensor, non_blocking=False)\n",
      " |         :noindex:\n",
      " |      \n",
      " |      .. function:: to(memory_format=torch.channels_last)\n",
      " |         :noindex:\n",
      " |      \n",
      " |      Its signature is similar to :meth:`torch.Tensor.to`, but only accepts\n",
      " |      floating point or complex :attr:`dtype`\\ s. In addition, this method will\n",
      " |      only cast the floating point or complex parameters and buffers to :attr:`dtype`\n",
      " |      (if given). The integral parameters and buffers will be moved\n",
      " |      :attr:`device`, if that is given, but with dtypes unchanged. When\n",
      " |      :attr:`non_blocking` is set, it tries to convert/move asynchronously\n",
      " |      with respect to the host if possible, e.g., moving CPU Tensors with\n",
      " |      pinned memory to CUDA devices.\n",
      " |      \n",
      " |      See below for examples.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Args:\n",
      " |          device (:class:`torch.device`): the desired device of the parameters\n",
      " |              and buffers in this module\n",
      " |          dtype (:class:`torch.dtype`): the desired floating point or complex dtype of\n",
      " |              the parameters and buffers in this module\n",
      " |          tensor (torch.Tensor): Tensor whose dtype and device are the desired\n",
      " |              dtype and device for all parameters and buffers in this module\n",
      " |          memory_format (:class:`torch.memory_format`): the desired memory\n",
      " |              format for 4D parameters and buffers in this module (keyword\n",
      " |              only argument)\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |      \n",
      " |      Examples::\n",
      " |      \n",
      " |          >>> # xdoctest: +IGNORE_WANT(\"non-deterministic\")\n",
      " |          >>> linear = nn.Linear(2, 2)\n",
      " |          >>> linear.weight\n",
      " |          Parameter containing:\n",
      " |          tensor([[ 0.1913, -0.3420],\n",
      " |                  [-0.5113, -0.2325]])\n",
      " |          >>> linear.to(torch.double)\n",
      " |          Linear(in_features=2, out_features=2, bias=True)\n",
      " |          >>> linear.weight\n",
      " |          Parameter containing:\n",
      " |          tensor([[ 0.1913, -0.3420],\n",
      " |                  [-0.5113, -0.2325]], dtype=torch.float64)\n",
      " |          >>> # xdoctest: +REQUIRES(env:TORCH_DOCTEST_CUDA1)\n",
      " |          >>> gpu1 = torch.device(\"cuda:1\")\n",
      " |          >>> linear.to(gpu1, dtype=torch.half, non_blocking=True)\n",
      " |          Linear(in_features=2, out_features=2, bias=True)\n",
      " |          >>> linear.weight\n",
      " |          Parameter containing:\n",
      " |          tensor([[ 0.1914, -0.3420],\n",
      " |                  [-0.5112, -0.2324]], dtype=torch.float16, device='cuda:1')\n",
      " |          >>> cpu = torch.device(\"cpu\")\n",
      " |          >>> linear.to(cpu)\n",
      " |          Linear(in_features=2, out_features=2, bias=True)\n",
      " |          >>> linear.weight\n",
      " |          Parameter containing:\n",
      " |          tensor([[ 0.1914, -0.3420],\n",
      " |                  [-0.5112, -0.2324]], dtype=torch.float16)\n",
      " |      \n",
      " |          >>> linear = nn.Linear(2, 2, bias=None).to(torch.cdouble)\n",
      " |          >>> linear.weight\n",
      " |          Parameter containing:\n",
      " |          tensor([[ 0.3741+0.j,  0.2382+0.j],\n",
      " |                  [ 0.5593+0.j, -0.4443+0.j]], dtype=torch.complex128)\n",
      " |          >>> linear(torch.ones(3, 2, dtype=torch.cdouble))\n",
      " |          tensor([[0.6122+0.j, 0.1150+0.j],\n",
      " |                  [0.6122+0.j, 0.1150+0.j],\n",
      " |                  [0.6122+0.j, 0.1150+0.j]], dtype=torch.complex128)\n",
      " |  \n",
      " |  to_empty(self: ~T, *, device: Union[str, torch.device]) -> ~T\n",
      " |      Moves the parameters and buffers to the specified device without copying storage.\n",
      " |      \n",
      " |      Args:\n",
      " |          device (:class:`torch.device`): The desired device of the parameters\n",
      " |              and buffers in this module.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  train(self: ~T, mode: bool = True) -> ~T\n",
      " |      Sets the module in training mode.\n",
      " |      \n",
      " |      This has any effect only on certain modules. See documentations of\n",
      " |      particular modules for details of their behaviors in training/evaluation\n",
      " |      mode, if they are affected, e.g. :class:`Dropout`, :class:`BatchNorm`,\n",
      " |      etc.\n",
      " |      \n",
      " |      Args:\n",
      " |          mode (bool): whether to set training mode (``True``) or evaluation\n",
      " |                       mode (``False``). Default: ``True``.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  type(self: ~T, dst_type: Union[torch.dtype, str]) -> ~T\n",
      " |      Casts all parameters and buffers to :attr:`dst_type`.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Args:\n",
      " |          dst_type (type or string): the desired type\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  xpu(self: ~T, device: Union[int, torch.device, NoneType] = None) -> ~T\n",
      " |      Moves all model parameters and buffers to the XPU.\n",
      " |      \n",
      " |      This also makes associated parameters and buffers different objects. So\n",
      " |      it should be called before constructing optimizer if the module will\n",
      " |      live on XPU while being optimized.\n",
      " |      \n",
      " |      .. note::\n",
      " |          This method modifies the module in-place.\n",
      " |      \n",
      " |      Arguments:\n",
      " |          device (int, optional): if specified, all parameters will be\n",
      " |              copied to that device\n",
      " |      \n",
      " |      Returns:\n",
      " |          Module: self\n",
      " |  \n",
      " |  zero_grad(self, set_to_none: bool = True) -> None\n",
      " |      Sets gradients of all model parameters to zero. See similar function\n",
      " |      under :class:`torch.optim.Optimizer` for more context.\n",
      " |      \n",
      " |      Args:\n",
      " |          set_to_none (bool): instead of setting to zero, set the grads to None.\n",
      " |              See :meth:`torch.optim.Optimizer.zero_grad` for details.\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Data descriptors inherited from torch.nn.modules.module.Module:\n",
      " |  \n",
      " |  __dict__\n",
      " |      dictionary for instance variables (if defined)\n",
      " |  \n",
      " |  __weakref__\n",
      " |      list of weak references to the object (if defined)\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Data and other attributes inherited from torch.nn.modules.module.Module:\n",
      " |  \n",
      " |  T_destination = ~T_destination\n",
      " |  \n",
      " |  __annotations__ = {'__call__': typing.Callable[..., typing.Any], '_bac...\n",
      " |  \n",
      " |  call_super_init = False\n",
      " |  \n",
      " |  dump_patches = False\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(vocab)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000\n",
      "2000\n",
      "3000\n",
      "4000\n",
      "5000\n",
      "6000\n",
      "7000\n",
      "8000\n",
      "9000\n",
      "10000\n",
      "11000\n",
      "12000\n",
      "13000\n",
      "14000\n",
      "15000\n",
      "16000\n",
      "17000\n",
      "18000\n",
      "19000\n",
      "20000\n",
      "21000\n",
      "22000\n",
      "23000\n",
      "24000\n",
      "25000\n",
      "26000\n",
      "27000\n",
      "28000\n",
      "29000\n",
      "30000\n",
      "31000\n",
      "32000\n",
      "33000\n",
      "34000\n",
      "35000\n",
      "36000\n",
      "37000\n",
      "38000\n",
      "39000\n",
      "40000\n",
      "41000\n",
      "42000\n",
      "43000\n",
      "44000\n",
      "45000\n",
      "46000\n",
      "47000\n",
      "48000\n",
      "49000\n",
      "50000\n",
      "51000\n",
      "52000\n",
      "53000\n",
      "54000\n",
      "55000\n",
      "56000\n",
      "57000\n",
      "58000\n",
      "59000\n",
      "60000\n",
      "61000\n",
      "62000\n",
      "63000\n",
      "64000\n",
      "65000\n",
      "66000\n",
      "67000\n",
      "68000\n",
      "69000\n",
      "70000\n",
      "71000\n",
      "72000\n",
      "73000\n",
      "74000\n",
      "75000\n",
      "76000\n",
      "77000\n",
      "78000\n",
      "79000\n",
      "80000\n",
      "81000\n",
      "82000\n",
      "83000\n",
      "84000\n",
      "85000\n",
      "86000\n",
      "87000\n",
      "88000\n",
      "89000\n",
      "90000\n",
      "91000\n",
      "92000\n",
      "93000\n",
      "94000\n",
      "95000\n",
      "96000\n",
      "97000\n",
      "98000\n",
      "99000\n",
      "100000\n",
      "101000\n",
      "102000\n",
      "103000\n",
      "104000\n",
      "105000\n",
      "106000\n",
      "107000\n",
      "108000\n",
      "109000\n",
      "110000\n",
      "111000\n",
      "112000\n",
      "113000\n",
      "114000\n",
      "115000\n",
      "116000\n",
      "117000\n",
      "118000\n",
      "119000\n",
      "120000\n",
      "121000\n",
      "122000\n",
      "123000\n",
      "124000\n",
      "125000\n",
      "126000\n",
      "127000\n",
      "128000\n",
      "129000\n",
      "130000\n",
      "131000\n",
      "132000\n",
      "133000\n",
      "134000\n",
      "135000\n",
      "136000\n",
      "137000\n",
      "138000\n",
      "139000\n",
      "140000\n",
      "141000\n",
      "142000\n",
      "143000\n",
      "144000\n",
      "145000\n",
      "146000\n",
      "147000\n",
      "148000\n",
      "149000\n",
      "150000\n",
      "151000\n",
      "152000\n",
      "153000\n",
      "154000\n",
      "155000\n",
      "156000\n",
      "157000\n",
      "158000\n",
      "159000\n",
      "160000\n",
      "161000\n",
      "162000\n",
      "163000\n",
      "164000\n",
      "165000\n",
      "166000\n",
      "167000\n",
      "168000\n",
      "169000\n",
      "170000\n",
      "171000\n",
      "172000\n",
      "173000\n",
      "174000\n",
      "175000\n",
      "176000\n",
      "177000\n",
      "178000\n",
      "179000\n",
      "180000\n",
      "181000\n",
      "182000\n",
      "183000\n",
      "184000\n",
      "185000\n",
      "186000\n",
      "187000\n",
      "188000\n",
      "189000\n",
      "190000\n",
      "191000\n",
      "192000\n",
      "193000\n",
      "194000\n",
      "195000\n",
      "196000\n",
      "197000\n",
      "198000\n",
      "199000\n",
      "200000\n",
      "201000\n",
      "202000\n",
      "203000\n",
      "204000\n",
      "205000\n",
      "206000\n",
      "207000\n",
      "208000\n",
      "209000\n",
      "210000\n",
      "211000\n",
      "212000\n",
      "213000\n",
      "214000\n",
      "215000\n",
      "216000\n",
      "217000\n",
      "218000\n",
      "219000\n",
      "220000\n",
      "221000\n",
      "222000\n",
      "223000\n",
      "224000\n",
      "225000\n",
      "226000\n",
      "227000\n",
      "228000\n",
      "229000\n",
      "230000\n",
      "231000\n",
      "232000\n",
      "233000\n",
      "234000\n",
      "235000\n",
      "236000\n",
      "237000\n",
      "238000\n",
      "239000\n",
      "240000\n",
      "241000\n",
      "242000\n",
      "243000\n",
      "244000\n",
      "245000\n",
      "246000\n",
      "247000\n",
      "248000\n",
      "249000\n",
      "250000\n",
      "251000\n",
      "252000\n",
      "253000\n",
      "254000\n",
      "255000\n",
      "256000\n",
      "257000\n",
      "258000\n",
      "259000\n",
      "260000\n",
      "261000\n",
      "262000\n",
      "263000\n",
      "264000\n",
      "265000\n",
      "266000\n",
      "267000\n",
      "268000\n",
      "269000\n",
      "270000\n",
      "271000\n",
      "272000\n",
      "273000\n",
      "274000\n",
      "275000\n",
      "276000\n",
      "277000\n",
      "278000\n",
      "279000\n",
      "280000\n",
      "281000\n",
      "282000\n",
      "283000\n",
      "284000\n",
      "285000\n",
      "286000\n",
      "287000\n",
      "288000\n",
      "289000\n",
      "290000\n",
      "291000\n",
      "292000\n",
      "293000\n",
      "294000\n",
      "295000\n",
      "296000\n",
      "297000\n",
      "298000\n",
      "299000\n",
      "300000\n",
      "301000\n",
      "302000\n",
      "303000\n",
      "304000\n",
      "305000\n",
      "306000\n",
      "307000\n",
      "308000\n",
      "309000\n",
      "310000\n",
      "311000\n",
      "312000\n",
      "313000\n",
      "314000\n",
      "315000\n",
      "316000\n",
      "317000\n",
      "318000\n",
      "319000\n",
      "320000\n",
      "321000\n",
      "322000\n",
      "323000\n",
      "324000\n",
      "325000\n",
      "326000\n",
      "327000\n",
      "328000\n",
      "329000\n",
      "330000\n",
      "331000\n",
      "332000\n",
      "333000\n",
      "334000\n",
      "335000\n",
      "336000\n",
      "337000\n",
      "338000\n",
      "339000\n",
      "340000\n",
      "341000\n",
      "342000\n",
      "343000\n",
      "344000\n",
      "345000\n",
      "346000\n",
      "347000\n",
      "348000\n",
      "349000\n",
      "350000\n",
      "351000\n",
      "352000\n",
      "353000\n",
      "354000\n",
      "355000\n",
      "356000\n",
      "357000\n",
      "358000\n",
      "359000\n",
      "360000\n",
      "361000\n",
      "362000\n",
      "363000\n",
      "364000\n",
      "365000\n",
      "366000\n",
      "367000\n",
      "368000\n",
      "369000\n",
      "370000\n",
      "371000\n",
      "372000\n",
      "373000\n",
      "374000\n",
      "375000\n",
      "376000\n",
      "377000\n",
      "378000\n",
      "379000\n",
      "380000\n",
      "381000\n",
      "382000\n",
      "383000\n",
      "384000\n",
      "385000\n",
      "386000\n",
      "387000\n",
      "388000\n",
      "389000\n",
      "390000\n",
      "391000\n",
      "392000\n",
      "393000\n",
      "394000\n",
      "395000\n",
      "396000\n",
      "397000\n",
      "398000\n",
      "399000\n",
      "400000\n",
      "401000\n",
      "402000\n",
      "403000\n",
      "404000\n",
      "405000\n",
      "406000\n",
      "407000\n",
      "408000\n",
      "409000\n",
      "410000\n",
      "411000\n",
      "412000\n",
      "413000\n",
      "414000\n",
      "415000\n",
      "416000\n",
      "417000\n",
      "418000\n",
      "419000\n",
      "420000\n",
      "421000\n",
      "422000\n",
      "423000\n",
      "424000\n",
      "425000\n",
      "426000\n",
      "427000\n",
      "428000\n",
      "429000\n",
      "430000\n",
      "431000\n",
      "432000\n"
     ]
    }
   ],
   "source": [
    "def look_ahead_iterator(gen):\n",
    "    prev = None\n",
    "    for item in gen:\n",
    "        if prev is not None:\n",
    "            yield (prev, item)\n",
    "        prev = item\n",
    "\n",
    "class Bigrams(IterableDataset):\n",
    "  def __init__(self, text_file, vocabulary_size):\n",
    "      self.vocab = build_vocab_from_iterator(\n",
    "         get_word_lines_from_file(text_file),\n",
    "         max_tokens = vocabulary_size,\n",
    "         specials = ['<unk>'])\n",
    "      self.vocab.set_default_index(self.vocab['<unk>'])\n",
    "      self.vocabulary_size = vocabulary_size\n",
    "      self.text_file = text_file\n",
    "\n",
    "  def __iter__(self):\n",
    "     return look_ahead_iterator(\n",
    "         (self.vocab[t] for t in itertools.chain.from_iterable(get_word_lines_from_file(self.text_file))))\n",
    "\n",
    "train_dataset = Bigrams(train_file, vocab_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<__main__.Bigrams object at 0x7fdd26d23940>\n"
     ]
    }
   ],
   "source": [
    "print(train_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'|===========================================================================|\\n|                  PyTorch CUDA memory summary, device ID 0                 |\\n|---------------------------------------------------------------------------|\\n|            CUDA OOMs: 1            |        cudaMalloc retries: 1         |\\n|===========================================================================|\\n|        Metric         | Cur Usage  | Peak Usage | Tot Alloc  | Tot Freed  |\\n|---------------------------------------------------------------------------|\\n| Allocated memory      | 699613 KiB |   1903 MiB |   3735 MiB |   3052 MiB |\\n|       from large pool | 699414 KiB |   1903 MiB |   3734 MiB |   3051 MiB |\\n|       from small pool |    199 KiB |      1 MiB |      1 MiB |      1 MiB |\\n|---------------------------------------------------------------------------|\\n| Active memory         | 699613 KiB |   1903 MiB |   3735 MiB |   3052 MiB |\\n|       from large pool | 699414 KiB |   1903 MiB |   3734 MiB |   3051 MiB |\\n|       from small pool |    199 KiB |      1 MiB |      1 MiB |      1 MiB |\\n|---------------------------------------------------------------------------|\\n| Requested memory      | 699611 KiB |   1903 MiB |   3735 MiB |   3052 MiB |\\n|       from large pool | 699413 KiB |   1903 MiB |   3734 MiB |   3051 MiB |\\n|       from small pool |    197 KiB |      1 MiB |      1 MiB |      1 MiB |\\n|---------------------------------------------------------------------------|\\n| GPU reserved memory   | 710656 KiB |   1918 MiB |   1918 MiB |   1224 MiB |\\n|       from large pool | 708608 KiB |   1916 MiB |   1916 MiB |   1224 MiB |\\n|       from small pool |   2048 KiB |      2 MiB |      2 MiB |      0 MiB |\\n|---------------------------------------------------------------------------|\\n| Non-releasable memory |  11043 KiB |  19364 KiB |  28939 KiB |  17896 KiB |\\n|       from large pool |   9194 KiB |  17514 KiB |  25954 KiB |  16760 KiB |\\n|       from small pool |   1849 KiB |   1950 KiB |   2985 KiB |   1136 KiB |\\n|---------------------------------------------------------------------------|\\n| Allocations           |      10    |      17    |      38    |      28    |\\n|       from large pool |       5    |       7    |      10    |       5    |\\n|       from small pool |       5    |      11    |      28    |      23    |\\n|---------------------------------------------------------------------------|\\n| Active allocs         |      10    |      17    |      38    |      28    |\\n|       from large pool |       5    |       7    |      10    |       5    |\\n|       from small pool |       5    |      11    |      28    |      23    |\\n|---------------------------------------------------------------------------|\\n| GPU reserved segments |       5    |       7    |       7    |       2    |\\n|       from large pool |       4    |       6    |       6    |       2    |\\n|       from small pool |       1    |       1    |       1    |       0    |\\n|---------------------------------------------------------------------------|\\n| Non-releasable allocs |       6    |       8    |      20    |      14    |\\n|       from large pool |       4    |       6    |       9    |       5    |\\n|       from small pool |       2    |       3    |      11    |       9    |\\n|---------------------------------------------------------------------------|\\n| Oversize allocations  |       0    |       0    |       0    |       0    |\\n|---------------------------------------------------------------------------|\\n| Oversize GPU segments |       0    |       0    |       0    |       0    |\\n|===========================================================================|\\n'"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.cuda.memory_summary(device=None, abbreviated=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "os.environ[\"PYTORCH_CUDA_ALLOC_CONF\"] = \"max_split_size_mb:256\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cuda'\n",
    "model = SimpleBigramNeuralLanguageModel(vocab_size, embed_size).to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch: = 1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/gedin/.local/lib/python3.8/site-packages/torch/nn/modules/container.py:217: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.\n",
      "  input = module(input)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 tensor(5.9599, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "1000\n",
      "100 tensor(6.1015, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200 tensor(5.9708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "2000\n",
      "300 tensor(6.2176, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "3000\n",
      "400 tensor(5.9401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "4000\n",
      "500 tensor(6.2084, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "5000\n",
      "600 tensor(5.9736, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "6000\n",
      "700 tensor(6.1423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "7000\n",
      "800 tensor(5.7344, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "8000\n",
      "900 tensor(6.0950, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9000\n",
      "1000 tensor(5.8473, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "10000\n",
      "1100 tensor(6.0612, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "11000\n",
      "1200 tensor(6.1509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "12000\n",
      "1300 tensor(6.0760, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "13000\n",
      "1400 tensor(6.2047, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "14000\n",
      "1500 tensor(6.1186, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "15000\n",
      "1600 tensor(5.8722, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "16000\n",
      "1700 tensor(5.8741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "17000\n",
      "1800 tensor(5.8971, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "18000\n",
      "1900 tensor(5.8521, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "19000\n",
      "2000 tensor(5.9434, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20000\n",
      "2100 tensor(6.0348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "21000\n",
      "2200 tensor(5.8840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "22000\n",
      "2300 tensor(5.8641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "23000\n",
      "2400 tensor(5.9068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "24000\n",
      "2500 tensor(5.9170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "25000\n",
      "2600 tensor(5.9812, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "26000\n",
      "2700 tensor(5.8985, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "27000\n",
      "2800 tensor(6.0008, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "28000\n",
      "2900 tensor(6.1230, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "29000\n",
      "3000 tensor(5.8770, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30000\n",
      "3100 tensor(5.9268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "31000\n",
      "3200 tensor(5.8530, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "32000\n",
      "3300 tensor(5.8436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "33000\n",
      "3400 tensor(5.7692, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "34000\n",
      "3500 tensor(5.8909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "35000\n",
      "3600 tensor(5.8325, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "36000\n",
      "3700 tensor(5.8082, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "37000\n",
      "3800 tensor(5.8106, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "38000\n",
      "3900 tensor(5.6382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "39000\n",
      "4000 tensor(5.6596, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40000\n",
      "4100 tensor(5.9587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "41000\n",
      "4200 tensor(5.8862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "42000\n",
      "4300 tensor(5.9541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "43000\n",
      "4400 tensor(5.8681, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "44000\n",
      "4500 tensor(5.6963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "45000\n",
      "4600 tensor(6.0707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "46000\n",
      "4700 tensor(5.7091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "47000\n",
      "4800 tensor(5.8139, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "48000\n",
      "4900 tensor(5.8696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "49000\n",
      "5000 tensor(5.8844, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "50000\n",
      "5100 tensor(5.9806, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "51000\n",
      "5200 tensor(6.0075, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "52000\n",
      "5300 tensor(6.0588, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "53000\n",
      "5400 tensor(5.8456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "54000\n",
      "5500 tensor(5.9166, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "55000\n",
      "5600 tensor(5.6528, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "56000\n",
      "5700 tensor(5.8988, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "57000\n",
      "5800 tensor(5.9132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "58000\n",
      "5900 tensor(5.9460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "59000\n",
      "6000 tensor(5.7543, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "60000\n",
      "6100 tensor(5.8256, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "61000\n",
      "6200 tensor(5.9448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "62000\n",
      "6300 tensor(5.7601, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "63000\n",
      "6400 tensor(5.7091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "64000\n",
      "6500 tensor(5.5621, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "65000\n",
      "6600 tensor(5.7094, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "66000\n",
      "6700 tensor(5.6785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "67000\n",
      "6800 tensor(5.9249, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "68000\n",
      "6900 tensor(5.8775, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "69000\n",
      "7000 tensor(5.8075, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "70000\n",
      "7100 tensor(5.5748, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "71000\n",
      "7200 tensor(5.7217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "72000\n",
      "7300 tensor(5.9124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "73000\n",
      "7400 tensor(5.7197, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "74000\n",
      "7500 tensor(5.6429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "75000\n",
      "7600 tensor(5.6847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "76000\n",
      "7700 tensor(5.7197, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "77000\n",
      "7800 tensor(5.8559, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "78000\n",
      "7900 tensor(5.5600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "79000\n",
      "8000 tensor(5.6288, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "80000\n",
      "8100 tensor(5.7767, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "81000\n",
      "8200 tensor(5.8037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "82000\n",
      "8300 tensor(5.7344, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "83000\n",
      "8400 tensor(5.8092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "84000\n",
      "8500 tensor(5.8847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "85000\n",
      "8600 tensor(5.8754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "86000\n",
      "8700 tensor(5.9227, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "87000\n",
      "8800 tensor(5.8028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "88000\n",
      "8900 tensor(5.6476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "89000\n",
      "9000 tensor(5.7656, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "90000\n",
      "9100 tensor(5.7805, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "91000\n",
      "9200 tensor(5.6879, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "92000\n",
      "9300 tensor(5.7098, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "93000\n",
      "9400 tensor(5.5631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "94000\n",
      "9500 tensor(5.6497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "95000\n",
      "9600 tensor(5.7500, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "96000\n",
      "9700 tensor(5.6607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "97000\n",
      "9800 tensor(5.7196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9900 tensor(5.5987, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "98000\n",
      "10000 tensor(5.7795, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "99000\n",
      "10100 tensor(5.6980, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "100000\n",
      "10200 tensor(5.6093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "101000\n",
      "10300 tensor(5.6792, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "102000\n",
      "10400 tensor(5.7035, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "103000\n",
      "10500 tensor(5.8282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "104000\n",
      "10600 tensor(5.8605, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "105000\n",
      "10700 tensor(5.7354, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "106000\n",
      "10800 tensor(5.8034, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "107000\n",
      "10900 tensor(5.6194, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "108000\n",
      "11000 tensor(5.8502, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "109000\n",
      "11100 tensor(5.4406, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "110000\n",
      "11200 tensor(5.6379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "111000\n",
      "11300 tensor(5.6668, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "112000\n",
      "11400 tensor(5.6140, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "113000\n",
      "11500 tensor(5.6565, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "114000\n",
      "11600 tensor(5.6308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "115000\n",
      "11700 tensor(5.5680, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "116000\n",
      "11800 tensor(5.7604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "117000\n",
      "11900 tensor(5.5792, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "118000\n",
      "12000 tensor(5.7329, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "119000\n",
      "12100 tensor(5.7726, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "120000\n",
      "12200 tensor(5.7151, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "121000\n",
      "12300 tensor(5.8561, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "122000\n",
      "12400 tensor(5.6791, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "123000\n",
      "12500 tensor(5.5574, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "124000\n",
      "12600 tensor(5.6817, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "125000\n",
      "12700 tensor(5.5375, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "126000\n",
      "12800 tensor(5.7270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "127000\n",
      "12900 tensor(5.6252, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "128000\n",
      "13000 tensor(5.4536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "129000\n",
      "13100 tensor(5.6091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "130000\n",
      "13200 tensor(5.7324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "131000\n",
      "13300 tensor(5.5253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "132000\n",
      "13400 tensor(5.6491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "133000\n",
      "13500 tensor(5.5728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "134000\n",
      "13600 tensor(5.6632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "135000\n",
      "13700 tensor(5.6678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "136000\n",
      "13800 tensor(5.6112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "137000\n",
      "13900 tensor(5.4884, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "138000\n",
      "14000 tensor(5.7304, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "139000\n",
      "14100 tensor(5.4326, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "140000\n",
      "14200 tensor(5.7188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "141000\n",
      "14300 tensor(5.6519, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "142000\n",
      "14400 tensor(5.5892, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "143000\n",
      "14500 tensor(5.7225, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "144000\n",
      "14600 tensor(5.7216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "145000\n",
      "14700 tensor(5.5748, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "146000\n",
      "14800 tensor(6.0184, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "147000\n",
      "14900 tensor(5.6781, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "148000\n",
      "15000 tensor(5.6038, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "149000\n",
      "15100 tensor(5.7875, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "150000\n",
      "15200 tensor(5.6485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "151000\n",
      "15300 tensor(5.5927, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "152000\n",
      "15400 tensor(5.5156, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "153000\n",
      "15500 tensor(5.6556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "154000\n",
      "15600 tensor(5.6485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "155000\n",
      "15700 tensor(5.5904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "156000\n",
      "15800 tensor(5.4613, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "157000\n",
      "15900 tensor(5.6254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "158000\n",
      "16000 tensor(5.4349, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "159000\n",
      "16100 tensor(5.5205, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "160000\n",
      "16200 tensor(5.8051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "161000\n",
      "16300 tensor(5.6452, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "162000\n",
      "16400 tensor(5.6071, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "163000\n",
      "16500 tensor(5.7237, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "164000\n",
      "16600 tensor(5.5771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "165000\n",
      "16700 tensor(5.5355, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "166000\n",
      "16800 tensor(5.6363, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "167000\n",
      "16900 tensor(5.3746, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "168000\n",
      "17000 tensor(5.6707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "169000\n",
      "17100 tensor(5.5359, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "170000\n",
      "17200 tensor(5.6118, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "171000\n",
      "17300 tensor(5.6740, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "172000\n",
      "17400 tensor(5.4438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "173000\n",
      "17500 tensor(5.5001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "174000\n",
      "17600 tensor(5.4953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "175000\n",
      "17700 tensor(5.5398, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "176000\n",
      "17800 tensor(5.6053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "177000\n",
      "17900 tensor(5.4726, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "178000\n",
      "18000 tensor(5.6747, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "179000\n",
      "18100 tensor(5.6238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "180000\n",
      "18200 tensor(5.5469, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "181000\n",
      "18300 tensor(5.5299, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "182000\n",
      "18400 tensor(5.6323, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "183000\n",
      "18500 tensor(5.5893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "184000\n",
      "18600 tensor(5.7452, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "185000\n",
      "18700 tensor(5.5576, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "186000\n",
      "18800 tensor(5.7439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "187000\n",
      "18900 tensor(5.6106, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "188000\n",
      "19000 tensor(5.6647, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "189000\n",
      "19100 tensor(5.7728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "190000\n",
      "19200 tensor(5.6169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "191000\n",
      "19300 tensor(5.7852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "192000\n",
      "19400 tensor(5.5627, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "193000\n",
      "19500 tensor(5.5682, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "194000\n",
      "19600 tensor(5.5978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "195000\n",
      "19700 tensor(5.6453, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "196000\n",
      "19800 tensor(5.4786, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "197000\n",
      "19900 tensor(5.4894, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "198000\n",
      "20000 tensor(5.4999, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "199000\n",
      "20100 tensor(5.4881, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200000\n",
      "20200 tensor(5.3915, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "201000\n",
      "20300 tensor(5.5216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20400 tensor(5.5761, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "202000\n",
      "20500 tensor(5.5586, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "203000\n",
      "20600 tensor(5.7870, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "204000\n",
      "20700 tensor(5.5776, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "205000\n",
      "20800 tensor(5.4417, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "206000\n",
      "20900 tensor(5.7186, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "207000\n",
      "21000 tensor(5.5415, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "208000\n",
      "21100 tensor(5.5141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "209000\n",
      "21200 tensor(5.4401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "210000\n",
      "21300 tensor(5.6511, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "211000\n",
      "21400 tensor(5.6474, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "212000\n",
      "21500 tensor(5.3946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "213000\n",
      "21600 tensor(5.3958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "214000\n",
      "21700 tensor(5.4040, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "215000\n",
      "21800 tensor(5.5745, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "216000\n",
      "21900 tensor(5.4996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "217000\n",
      "22000 tensor(5.5234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "218000\n",
      "22100 tensor(5.3870, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "219000\n",
      "22200 tensor(5.2661, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "220000\n",
      "22300 tensor(5.7031, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "221000\n",
      "22400 tensor(5.3633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "222000\n",
      "22500 tensor(5.4404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "223000\n",
      "22600 tensor(5.5951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "224000\n",
      "22700 tensor(5.3901, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "225000\n",
      "22800 tensor(5.6404, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "226000\n",
      "22900 tensor(5.6646, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "227000\n",
      "23000 tensor(5.5949, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "228000\n",
      "23100 tensor(5.5284, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "229000\n",
      "23200 tensor(5.5617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "230000\n",
      "23300 tensor(5.6426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "231000\n",
      "23400 tensor(5.7283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "232000\n",
      "23500 tensor(5.4558, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "233000\n",
      "23600 tensor(5.4600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "234000\n",
      "23700 tensor(5.4961, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "235000\n",
      "23800 tensor(5.3373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "236000\n",
      "23900 tensor(5.4470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "237000\n",
      "24000 tensor(5.4346, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "238000\n",
      "24100 tensor(5.5112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "239000\n",
      "24200 tensor(5.6918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "240000\n",
      "24300 tensor(5.6115, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "241000\n",
      "24400 tensor(5.7404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "242000\n",
      "24500 tensor(5.4982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "243000\n",
      "24600 tensor(5.6136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "244000\n",
      "24700 tensor(5.5225, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "245000\n",
      "24800 tensor(5.5563, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "246000\n",
      "24900 tensor(5.6283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "247000\n",
      "25000 tensor(5.6176, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "248000\n",
      "25100 tensor(5.5795, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "249000\n",
      "25200 tensor(5.5831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "250000\n",
      "25300 tensor(5.5894, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "251000\n",
      "25400 tensor(5.5670, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "252000\n",
      "25500 tensor(5.5016, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "253000\n",
      "25600 tensor(5.7909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "254000\n",
      "25700 tensor(5.5229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "255000\n",
      "25800 tensor(5.6035, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "256000\n",
      "25900 tensor(5.5293, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "257000\n",
      "26000 tensor(5.5553, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "258000\n",
      "26100 tensor(5.4476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "259000\n",
      "26200 tensor(5.3721, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "260000\n",
      "26300 tensor(5.6142, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "261000\n",
      "26400 tensor(5.6202, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "262000\n",
      "26500 tensor(5.3529, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "263000\n",
      "26600 tensor(5.7148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "264000\n",
      "26700 tensor(5.5755, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "265000\n",
      "26800 tensor(5.7480, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "266000\n",
      "26900 tensor(5.5025, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "267000\n",
      "27000 tensor(5.4017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "268000\n",
      "27100 tensor(5.3996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "269000\n",
      "27200 tensor(5.4862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "270000\n",
      "27300 tensor(5.6392, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "271000\n",
      "27400 tensor(5.5634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "272000\n",
      "27500 tensor(5.4420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "273000\n",
      "27600 tensor(5.7835, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "274000\n",
      "27700 tensor(5.5555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "275000\n",
      "27800 tensor(5.5381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "276000\n",
      "27900 tensor(5.6515, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "277000\n",
      "28000 tensor(5.5254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "278000\n",
      "28100 tensor(5.4929, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "279000\n",
      "28200 tensor(5.6218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "280000\n",
      "28300 tensor(5.2878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "281000\n",
      "28400 tensor(5.7112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "282000\n",
      "28500 tensor(5.5490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "283000\n",
      "28600 tensor(5.4572, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "284000\n",
      "28700 tensor(5.6349, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "285000\n",
      "28800 tensor(5.6607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "286000\n",
      "28900 tensor(5.5422, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "287000\n",
      "29000 tensor(5.4277, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "288000\n",
      "29100 tensor(5.1870, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "289000\n",
      "29200 tensor(5.3593, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "290000\n",
      "29300 tensor(5.6512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "291000\n",
      "29400 tensor(5.8051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "292000\n",
      "29500 tensor(5.5308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "293000\n",
      "29600 tensor(5.3791, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "294000\n",
      "29700 tensor(5.6108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "295000\n",
      "29800 tensor(5.4015, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "296000\n",
      "29900 tensor(5.6953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "297000\n",
      "30000 tensor(5.3925, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "298000\n",
      "30100 tensor(5.4241, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "299000\n",
      "30200 tensor(5.4216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "300000\n",
      "30300 tensor(5.5074, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "301000\n",
      "30400 tensor(5.3631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "302000\n",
      "30500 tensor(5.5690, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30600 tensor(5.4734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "303000\n",
      "30700 tensor(5.5061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "304000\n",
      "30800 tensor(5.5709, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "305000\n",
      "30900 tensor(5.5478, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "306000\n",
      "31000 tensor(5.6687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "307000\n",
      "31100 tensor(5.2899, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "308000\n",
      "31200 tensor(5.3663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "309000\n",
      "31300 tensor(5.6274, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "310000\n",
      "31400 tensor(5.4358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "311000\n",
      "31500 tensor(5.5738, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "312000\n",
      "31600 tensor(5.5612, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "313000\n",
      "31700 tensor(5.5104, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "314000\n",
      "31800 tensor(5.6343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "315000\n",
      "31900 tensor(5.2243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "316000\n",
      "32000 tensor(5.4320, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "317000\n",
      "32100 tensor(5.3344, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "318000\n",
      "32200 tensor(5.6543, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "319000\n",
      "32300 tensor(5.6512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "320000\n",
      "32400 tensor(5.6237, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "321000\n",
      "32500 tensor(5.4246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "322000\n",
      "32600 tensor(5.5469, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "323000\n",
      "32700 tensor(5.5338, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "324000\n",
      "32800 tensor(5.6954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "325000\n",
      "32900 tensor(5.5754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "326000\n",
      "33000 tensor(5.3334, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "327000\n",
      "33100 tensor(5.5284, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "328000\n",
      "33200 tensor(5.6350, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "329000\n",
      "33300 tensor(5.4312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "330000\n",
      "33400 tensor(5.6854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "331000\n",
      "33500 tensor(5.4921, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "332000\n",
      "33600 tensor(5.4345, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "333000\n",
      "33700 tensor(5.4950, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "334000\n",
      "33800 tensor(5.5757, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "335000\n",
      "33900 tensor(5.3466, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "336000\n",
      "34000 tensor(5.5373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "337000\n",
      "34100 tensor(5.5144, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "338000\n",
      "34200 tensor(5.5543, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "339000\n",
      "34300 tensor(5.3564, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "340000\n",
      "34400 tensor(5.8091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "341000\n",
      "34500 tensor(5.6699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "342000\n",
      "34600 tensor(5.5536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "343000\n",
      "34700 tensor(5.6261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "344000\n",
      "34800 tensor(5.6504, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "345000\n",
      "34900 tensor(5.7067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "346000\n",
      "35000 tensor(5.7307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "347000\n",
      "35100 tensor(5.4831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "348000\n",
      "35200 tensor(5.4367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "349000\n",
      "35300 tensor(5.6503, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "350000\n",
      "35400 tensor(5.2892, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "351000\n",
      "35500 tensor(5.4198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "352000\n",
      "35600 tensor(5.4870, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "353000\n",
      "35700 tensor(5.4489, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "354000\n",
      "35800 tensor(5.5170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "355000\n",
      "35900 tensor(5.4699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "356000\n",
      "36000 tensor(5.2451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "357000\n",
      "36100 tensor(5.6311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "358000\n",
      "36200 tensor(5.5157, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "359000\n",
      "36300 tensor(5.7751, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "360000\n",
      "36400 tensor(5.4740, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "361000\n",
      "36500 tensor(5.4746, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "362000\n",
      "36600 tensor(5.5244, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "363000\n",
      "36700 tensor(5.3037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "364000\n",
      "36800 tensor(5.4238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "365000\n",
      "36900 tensor(5.5203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "366000\n",
      "37000 tensor(5.4431, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "367000\n",
      "37100 tensor(5.4286, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "368000\n",
      "37200 tensor(5.5108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "369000\n",
      "37300 tensor(5.4229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "370000\n",
      "37400 tensor(5.8406, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "371000\n",
      "37500 tensor(5.4602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "372000\n",
      "37600 tensor(5.4417, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "373000\n",
      "37700 tensor(5.6200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "374000\n",
      "37800 tensor(5.4527, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "375000\n",
      "37900 tensor(5.4631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "376000\n",
      "38000 tensor(5.5196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "377000\n",
      "38100 tensor(5.5436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "378000\n",
      "38200 tensor(5.5269, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "379000\n",
      "38300 tensor(5.4716, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "380000\n",
      "38400 tensor(5.5081, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "381000\n",
      "38500 tensor(5.5249, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "382000\n",
      "38600 tensor(5.5018, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "383000\n",
      "38700 tensor(5.4845, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "384000\n",
      "38800 tensor(5.5505, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "385000\n",
      "38900 tensor(5.6658, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "386000\n",
      "39000 tensor(5.3333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "387000\n",
      "39100 tensor(5.5598, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "388000\n",
      "39200 tensor(5.6624, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "389000\n",
      "39300 tensor(5.4714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "390000\n",
      "39400 tensor(5.5470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "391000\n",
      "39500 tensor(5.6905, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "392000\n",
      "39600 tensor(5.3592, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "393000\n",
      "39700 tensor(5.3170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "394000\n",
      "39800 tensor(5.4491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "395000\n",
      "39900 tensor(5.2872, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "396000\n",
      "40000 tensor(5.3865, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "397000\n",
      "40100 tensor(5.4536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "398000\n",
      "40200 tensor(5.4382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "399000\n",
      "40300 tensor(5.4819, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40400 tensor(5.5250, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "400000\n",
      "40500 tensor(5.4396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "401000\n",
      "40600 tensor(5.5062, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "402000\n",
      "40700 tensor(5.5362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "403000\n",
      "40800 tensor(5.5015, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "404000\n",
      "40900 tensor(5.4610, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "405000\n",
      "41000 tensor(5.5083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "406000\n",
      "41100 tensor(5.4346, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "407000\n",
      "41200 tensor(5.3340, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "408000\n",
      "41300 tensor(5.4608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "409000\n",
      "41400 tensor(5.3758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "410000\n",
      "41500 tensor(5.5160, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "411000\n",
      "41600 tensor(5.4290, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "412000\n",
      "41700 tensor(5.4426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "413000\n",
      "41800 tensor(5.4764, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "414000\n",
      "41900 tensor(5.4730, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "415000\n",
      "42000 tensor(5.6150, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "416000\n",
      "42100 tensor(5.3622, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "417000\n",
      "42200 tensor(5.4380, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "418000\n",
      "42300 tensor(5.5031, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "419000\n",
      "42400 tensor(5.3124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "420000\n",
      "42500 tensor(5.4812, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "421000\n",
      "42600 tensor(5.2723, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "422000\n",
      "42700 tensor(5.5998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "423000\n",
      "42800 tensor(5.5254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "424000\n",
      "42900 tensor(5.3716, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "425000\n",
      "43000 tensor(5.5020, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "426000\n",
      "43100 tensor(5.5091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "427000\n",
      "43200 tensor(5.3182, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "428000\n",
      "43300 tensor(5.4001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "429000\n",
      "43400 tensor(5.5150, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "430000\n",
      "43500 tensor(5.2440, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "431000\n",
      "43600 tensor(5.4439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "432000\n",
      "epoch: = 2\n",
      "0 tensor(5.3953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "1000\n",
      "100 tensor(5.4847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200 tensor(5.3626, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "2000\n",
      "300 tensor(5.4127, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "3000\n",
      "400 tensor(5.3734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "4000\n",
      "500 tensor(5.5564, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "5000\n",
      "600 tensor(5.3391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "6000\n",
      "700 tensor(5.6198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "7000\n",
      "800 tensor(5.2255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "8000\n",
      "900 tensor(5.5161, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9000\n",
      "1000 tensor(5.3517, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "10000\n",
      "1100 tensor(5.5420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "11000\n",
      "1200 tensor(5.6031, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "12000\n",
      "1300 tensor(5.5343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "13000\n",
      "1400 tensor(5.5547, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "14000\n",
      "1500 tensor(5.6080, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "15000\n",
      "1600 tensor(5.2940, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "16000\n",
      "1700 tensor(5.3671, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "17000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1800 tensor(5.3777, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "18000\n",
      "1900 tensor(5.3593, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "19000\n",
      "2000 tensor(5.4348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20000\n",
      "2100 tensor(5.5513, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "21000\n",
      "2200 tensor(5.3939, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "22000\n",
      "2300 tensor(5.4063, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "23000\n",
      "2400 tensor(5.4092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "24000\n",
      "2500 tensor(5.4460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "25000\n",
      "2600 tensor(5.4738, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "26000\n",
      "2700 tensor(5.4848, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "27000\n",
      "2800 tensor(5.5244, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "28000\n",
      "2900 tensor(5.6711, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "29000\n",
      "3000 tensor(5.4024, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30000\n",
      "3100 tensor(5.4842, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "31000\n",
      "3200 tensor(5.4863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "32000\n",
      "3300 tensor(5.4114, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "33000\n",
      "3400 tensor(5.3231, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "34000\n",
      "3500 tensor(5.4598, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "35000\n",
      "3600 tensor(5.4579, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "36000\n",
      "3700 tensor(5.3890, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "37000\n",
      "3800 tensor(5.4162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "38000\n",
      "3900 tensor(5.2854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "39000\n",
      "4000 tensor(5.3370, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40000\n",
      "4100 tensor(5.5078, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "41000\n",
      "4200 tensor(5.5341, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "42000\n",
      "4300 tensor(5.4704, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "43000\n",
      "4400 tensor(5.4990, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "44000\n",
      "4500 tensor(5.3300, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "45000\n",
      "4600 tensor(5.6674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "46000\n",
      "4700 tensor(5.3622, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "47000\n",
      "4800 tensor(5.4762, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "48000\n",
      "4900 tensor(5.5403, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "49000\n",
      "5000 tensor(5.5359, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "50000\n",
      "5100 tensor(5.6058, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "51000\n",
      "5200 tensor(5.6209, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "52000\n",
      "5300 tensor(5.6273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "53000\n",
      "5400 tensor(5.4695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "54000\n",
      "5500 tensor(5.5771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "55000\n",
      "5600 tensor(5.3552, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "56000\n",
      "5700 tensor(5.5957, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "57000\n",
      "5800 tensor(5.5952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "58000\n",
      "5900 tensor(5.5643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "59000\n",
      "6000 tensor(5.4346, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "60000\n",
      "6100 tensor(5.4620, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "61000\n",
      "6200 tensor(5.6256, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "62000\n",
      "6300 tensor(5.4832, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "63000\n",
      "6400 tensor(5.4063, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "64000\n",
      "6500 tensor(5.2587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "65000\n",
      "6600 tensor(5.4320, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "66000\n",
      "6700 tensor(5.3770, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "67000\n",
      "6800 tensor(5.6077, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "68000\n",
      "6900 tensor(5.5788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "69000\n",
      "7000 tensor(5.4929, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "70000\n",
      "7100 tensor(5.2828, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "71000\n",
      "7200 tensor(5.3992, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "72000\n",
      "7300 tensor(5.6273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "73000\n",
      "7400 tensor(5.4385, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "74000\n",
      "7500 tensor(5.3176, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "75000\n",
      "7600 tensor(5.3834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "76000\n",
      "7700 tensor(5.4532, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "77000\n",
      "7800 tensor(5.5669, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "78000\n",
      "7900 tensor(5.2508, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "79000\n",
      "8000 tensor(5.3027, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "80000\n",
      "8100 tensor(5.4813, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "81000\n",
      "8200 tensor(5.4822, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "82000\n",
      "8300 tensor(5.4510, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "83000\n",
      "8400 tensor(5.5712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "84000\n",
      "8500 tensor(5.5634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "85000\n",
      "8600 tensor(5.5616, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "86000\n",
      "8700 tensor(5.6568, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "87000\n",
      "8800 tensor(5.5397, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "88000\n",
      "8900 tensor(5.3852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "89000\n",
      "9000 tensor(5.5022, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "90000\n",
      "9100 tensor(5.5088, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "91000\n",
      "9200 tensor(5.4214, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "92000\n",
      "9300 tensor(5.4641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "93000\n",
      "9400 tensor(5.3085, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "94000\n",
      "9500 tensor(5.3852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "95000\n",
      "9600 tensor(5.5097, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "96000\n",
      "9700 tensor(5.4373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "97000\n",
      "9800 tensor(5.4786, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9900 tensor(5.3198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "98000\n",
      "10000 tensor(5.5310, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "99000\n",
      "10100 tensor(5.4341, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "100000\n",
      "10200 tensor(5.3571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "101000\n",
      "10300 tensor(5.4712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "102000\n",
      "10400 tensor(5.4810, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "103000\n",
      "10500 tensor(5.5463, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "104000\n",
      "10600 tensor(5.6233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "105000\n",
      "10700 tensor(5.4678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "106000\n",
      "10800 tensor(5.5040, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "107000\n",
      "10900 tensor(5.3963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "108000\n",
      "11000 tensor(5.6295, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "109000\n",
      "11100 tensor(5.2378, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "110000\n",
      "11200 tensor(5.4184, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "111000\n",
      "11300 tensor(5.4404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "112000\n",
      "11400 tensor(5.3875, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "113000\n",
      "11500 tensor(5.4523, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "114000\n",
      "11600 tensor(5.4418, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "115000\n",
      "11700 tensor(5.3604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "116000\n",
      "11800 tensor(5.5647, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "117000\n",
      "11900 tensor(5.3936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "118000\n",
      "12000 tensor(5.4823, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "119000\n",
      "12100 tensor(5.5069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "120000\n",
      "12200 tensor(5.4983, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "121000\n",
      "12300 tensor(5.6030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "122000\n",
      "12400 tensor(5.4763, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "123000\n",
      "12500 tensor(5.3718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "124000\n",
      "12600 tensor(5.4416, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "125000\n",
      "12700 tensor(5.3554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "126000\n",
      "12800 tensor(5.5392, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "127000\n",
      "12900 tensor(5.4164, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "128000\n",
      "13000 tensor(5.2286, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "129000\n",
      "13100 tensor(5.4288, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "130000\n",
      "13200 tensor(5.4770, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "131000\n",
      "13300 tensor(5.3352, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "132000\n",
      "13400 tensor(5.4349, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "133000\n",
      "13500 tensor(5.3860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "134000\n",
      "13600 tensor(5.4648, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "135000\n",
      "13700 tensor(5.4444, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "136000\n",
      "13800 tensor(5.4320, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "137000\n",
      "13900 tensor(5.2935, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "138000\n",
      "14000 tensor(5.5387, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "139000\n",
      "14100 tensor(5.2424, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "140000\n",
      "14200 tensor(5.5177, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "141000\n",
      "14300 tensor(5.4831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "142000\n",
      "14400 tensor(5.3877, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "143000\n",
      "14500 tensor(5.4919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "144000\n",
      "14600 tensor(5.5253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "145000\n",
      "14700 tensor(5.3948, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "146000\n",
      "14800 tensor(5.8442, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "147000\n",
      "14900 tensor(5.4967, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "148000\n",
      "15000 tensor(5.3788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "149000\n",
      "15100 tensor(5.5832, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "150000\n",
      "15200 tensor(5.4482, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "151000\n",
      "15300 tensor(5.4260, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "152000\n",
      "15400 tensor(5.3273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "153000\n",
      "15500 tensor(5.4840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "154000\n",
      "15600 tensor(5.4851, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "155000\n",
      "15700 tensor(5.3871, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "156000\n",
      "15800 tensor(5.2933, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "157000\n",
      "15900 tensor(5.4374, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "158000\n",
      "16000 tensor(5.2555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "159000\n",
      "16100 tensor(5.3127, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "160000\n",
      "16200 tensor(5.6423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "161000\n",
      "16300 tensor(5.4702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "162000\n",
      "16400 tensor(5.4419, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "163000\n",
      "16500 tensor(5.5640, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "164000\n",
      "16600 tensor(5.4099, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "165000\n",
      "16700 tensor(5.3822, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "166000\n",
      "16800 tensor(5.4643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "167000\n",
      "16900 tensor(5.2234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "168000\n",
      "17000 tensor(5.5021, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "169000\n",
      "17100 tensor(5.3524, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "170000\n",
      "17200 tensor(5.4725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "171000\n",
      "17300 tensor(5.5034, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "172000\n",
      "17400 tensor(5.2911, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "173000\n",
      "17500 tensor(5.3147, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "174000\n",
      "17600 tensor(5.3426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "175000\n",
      "17700 tensor(5.3414, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "176000\n",
      "17800 tensor(5.3991, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "177000\n",
      "17900 tensor(5.2936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "178000\n",
      "18000 tensor(5.5238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "179000\n",
      "18100 tensor(5.4684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "180000\n",
      "18200 tensor(5.3916, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "181000\n",
      "18300 tensor(5.3888, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "182000\n",
      "18400 tensor(5.4299, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "183000\n",
      "18500 tensor(5.4103, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "184000\n",
      "18600 tensor(5.5980, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "185000\n",
      "18700 tensor(5.4135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "186000\n",
      "18800 tensor(5.5855, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "187000\n",
      "18900 tensor(5.4583, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "188000\n",
      "19000 tensor(5.4854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "189000\n",
      "19100 tensor(5.5879, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "190000\n",
      "19200 tensor(5.4675, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "191000\n",
      "19300 tensor(5.5741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "192000\n",
      "19400 tensor(5.3977, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "193000\n",
      "19500 tensor(5.4042, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "194000\n",
      "19600 tensor(5.4364, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "195000\n",
      "19700 tensor(5.4868, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "196000\n",
      "19800 tensor(5.3476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "197000\n",
      "19900 tensor(5.3553, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "198000\n",
      "20000 tensor(5.3707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "199000\n",
      "20100 tensor(5.3226, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200000\n",
      "20200 tensor(5.2488, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "201000\n",
      "20300 tensor(5.3648, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20400 tensor(5.4156, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "202000\n",
      "20500 tensor(5.4102, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "203000\n",
      "20600 tensor(5.6109, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "204000\n",
      "20700 tensor(5.4335, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "205000\n",
      "20800 tensor(5.2795, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "206000\n",
      "20900 tensor(5.5609, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "207000\n",
      "21000 tensor(5.3918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "208000\n",
      "21100 tensor(5.3831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "209000\n",
      "21200 tensor(5.2790, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "210000\n",
      "21300 tensor(5.4710, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "211000\n",
      "21400 tensor(5.5050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "212000\n",
      "21500 tensor(5.2692, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "213000\n",
      "21600 tensor(5.2668, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "214000\n",
      "21700 tensor(5.2633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "215000\n",
      "21800 tensor(5.4067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "216000\n",
      "21900 tensor(5.3829, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "217000\n",
      "22000 tensor(5.3773, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "218000\n",
      "22100 tensor(5.2472, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "219000\n",
      "22200 tensor(5.1171, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "220000\n",
      "22300 tensor(5.5545, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "221000\n",
      "22400 tensor(5.2499, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "222000\n",
      "22500 tensor(5.2943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "223000\n",
      "22600 tensor(5.4748, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "224000\n",
      "22700 tensor(5.2436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "225000\n",
      "22800 tensor(5.5053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "226000\n",
      "22900 tensor(5.5519, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "227000\n",
      "23000 tensor(5.4541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "228000\n",
      "23100 tensor(5.4279, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "229000\n",
      "23200 tensor(5.4286, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "230000\n",
      "23300 tensor(5.5179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "231000\n",
      "23400 tensor(5.5355, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "232000\n",
      "23500 tensor(5.3505, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "233000\n",
      "23600 tensor(5.3313, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "234000\n",
      "23700 tensor(5.3509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "235000\n",
      "23800 tensor(5.2170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "236000\n",
      "23900 tensor(5.3101, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "237000\n",
      "24000 tensor(5.2962, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "238000\n",
      "24100 tensor(5.3882, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "239000\n",
      "24200 tensor(5.5633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "240000\n",
      "24300 tensor(5.4595, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "241000\n",
      "24400 tensor(5.5932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "242000\n",
      "24500 tensor(5.3717, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "243000\n",
      "24600 tensor(5.4943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "244000\n",
      "24700 tensor(5.3985, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "245000\n",
      "24800 tensor(5.4347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "246000\n",
      "24900 tensor(5.5008, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "247000\n",
      "25000 tensor(5.5100, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "248000\n",
      "25100 tensor(5.4427, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "249000\n",
      "25200 tensor(5.4508, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "250000\n",
      "25300 tensor(5.4724, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "251000\n",
      "25400 tensor(5.4525, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "252000\n",
      "25500 tensor(5.3620, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "253000\n",
      "25600 tensor(5.6446, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "254000\n",
      "25700 tensor(5.3966, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "255000\n",
      "25800 tensor(5.4889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "256000\n",
      "25900 tensor(5.4251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "257000\n",
      "26000 tensor(5.4346, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "258000\n",
      "26100 tensor(5.3395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "259000\n",
      "26200 tensor(5.2695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "260000\n",
      "26300 tensor(5.4767, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "261000\n",
      "26400 tensor(5.5083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "262000\n",
      "26500 tensor(5.2347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "263000\n",
      "26600 tensor(5.5761, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "264000\n",
      "26700 tensor(5.4402, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "265000\n",
      "26800 tensor(5.6173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "266000\n",
      "26900 tensor(5.3775, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "267000\n",
      "27000 tensor(5.2863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "268000\n",
      "27100 tensor(5.3007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "269000\n",
      "27200 tensor(5.3551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "270000\n",
      "27300 tensor(5.5439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "271000\n",
      "27400 tensor(5.4334, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "272000\n",
      "27500 tensor(5.3266, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "273000\n",
      "27600 tensor(5.6412, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "274000\n",
      "27700 tensor(5.4420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "275000\n",
      "27800 tensor(5.4381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "276000\n",
      "27900 tensor(5.5550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "277000\n",
      "28000 tensor(5.4154, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "278000\n",
      "28100 tensor(5.3823, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "279000\n",
      "28200 tensor(5.5344, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "280000\n",
      "28300 tensor(5.1615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "281000\n",
      "28400 tensor(5.6069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "282000\n",
      "28500 tensor(5.4426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "283000\n",
      "28600 tensor(5.3672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "284000\n",
      "28700 tensor(5.5133, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "285000\n",
      "28800 tensor(5.5556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "286000\n",
      "28900 tensor(5.4294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "287000\n",
      "29000 tensor(5.3359, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "288000\n",
      "29100 tensor(5.0951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "289000\n",
      "29200 tensor(5.2511, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "290000\n",
      "29300 tensor(5.5364, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "291000\n",
      "29400 tensor(5.6708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "292000\n",
      "29500 tensor(5.4371, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "293000\n",
      "29600 tensor(5.2942, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "294000\n",
      "29700 tensor(5.4637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "295000\n",
      "29800 tensor(5.2914, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "296000\n",
      "29900 tensor(5.5562, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "297000\n",
      "30000 tensor(5.2833, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "298000\n",
      "30100 tensor(5.3481, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "299000\n",
      "30200 tensor(5.3122, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "300000\n",
      "30300 tensor(5.4103, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "301000\n",
      "30400 tensor(5.2480, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "302000\n",
      "30500 tensor(5.4258, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30600 tensor(5.3835, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "303000\n",
      "30700 tensor(5.4193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "304000\n",
      "30800 tensor(5.4438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "305000\n",
      "30900 tensor(5.4518, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "306000\n",
      "31000 tensor(5.5607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "307000\n",
      "31100 tensor(5.2059, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "308000\n",
      "31200 tensor(5.2571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "309000\n",
      "31300 tensor(5.5208, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "310000\n",
      "31400 tensor(5.3061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "311000\n",
      "31500 tensor(5.4834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "312000\n",
      "31600 tensor(5.4653, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "313000\n",
      "31700 tensor(5.4308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "314000\n",
      "31800 tensor(5.5400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "315000\n",
      "31900 tensor(5.1536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "316000\n",
      "32000 tensor(5.3460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "317000\n",
      "32100 tensor(5.2300, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "318000\n",
      "32200 tensor(5.5511, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "319000\n",
      "32300 tensor(5.5391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "320000\n",
      "32400 tensor(5.5157, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "321000\n",
      "32500 tensor(5.3336, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "322000\n",
      "32600 tensor(5.4475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "323000\n",
      "32700 tensor(5.3894, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "324000\n",
      "32800 tensor(5.6022, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "325000\n",
      "32900 tensor(5.4663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "326000\n",
      "33000 tensor(5.2387, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "327000\n",
      "33100 tensor(5.4446, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "328000\n",
      "33200 tensor(5.5450, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "329000\n",
      "33300 tensor(5.3179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "330000\n",
      "33400 tensor(5.5905, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "331000\n",
      "33500 tensor(5.4066, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "332000\n",
      "33600 tensor(5.3542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "333000\n",
      "33700 tensor(5.4097, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "334000\n",
      "33800 tensor(5.4912, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "335000\n",
      "33900 tensor(5.2358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "336000\n",
      "34000 tensor(5.4470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "337000\n",
      "34100 tensor(5.4207, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "338000\n",
      "34200 tensor(5.4651, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "339000\n",
      "34300 tensor(5.2545, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "340000\n",
      "34400 tensor(5.7106, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "341000\n",
      "34500 tensor(5.5699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "342000\n",
      "34600 tensor(5.4638, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "343000\n",
      "34700 tensor(5.5382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "344000\n",
      "34800 tensor(5.5603, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "345000\n",
      "34900 tensor(5.6072, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "346000\n",
      "35000 tensor(5.6037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "347000\n",
      "35100 tensor(5.4069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "348000\n",
      "35200 tensor(5.3398, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "349000\n",
      "35300 tensor(5.5607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "350000\n",
      "35400 tensor(5.2068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "351000\n",
      "35500 tensor(5.3112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "352000\n",
      "35600 tensor(5.4126, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "353000\n",
      "35700 tensor(5.3091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "354000\n",
      "35800 tensor(5.4252, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "355000\n",
      "35900 tensor(5.3956, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "356000\n",
      "36000 tensor(5.1705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "357000\n",
      "36100 tensor(5.5497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "358000\n",
      "36200 tensor(5.4066, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "359000\n",
      "36300 tensor(5.6858, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "360000\n",
      "36400 tensor(5.3812, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "361000\n",
      "36500 tensor(5.3990, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "362000\n",
      "36600 tensor(5.4302, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "363000\n",
      "36700 tensor(5.2253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "364000\n",
      "36800 tensor(5.3347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "365000\n",
      "36900 tensor(5.4426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "366000\n",
      "37000 tensor(5.3419, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "367000\n",
      "37100 tensor(5.3579, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "368000\n",
      "37200 tensor(5.4332, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "369000\n",
      "37300 tensor(5.3362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "370000\n",
      "37400 tensor(5.7100, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "371000\n",
      "37500 tensor(5.3742, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "372000\n",
      "37600 tensor(5.3615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "373000\n",
      "37700 tensor(5.5402, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "374000\n",
      "37800 tensor(5.3734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "375000\n",
      "37900 tensor(5.3621, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "376000\n",
      "38000 tensor(5.4380, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "377000\n",
      "38100 tensor(5.4513, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "378000\n",
      "38200 tensor(5.4554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "379000\n",
      "38300 tensor(5.3735, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "380000\n",
      "38400 tensor(5.4297, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "381000\n",
      "38500 tensor(5.4561, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "382000\n",
      "38600 tensor(5.4118, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "383000\n",
      "38700 tensor(5.3996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "384000\n",
      "38800 tensor(5.4825, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "385000\n",
      "38900 tensor(5.5692, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "386000\n",
      "39000 tensor(5.2573, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "387000\n",
      "39100 tensor(5.4847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "388000\n",
      "39200 tensor(5.5802, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "389000\n",
      "39300 tensor(5.3968, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "390000\n",
      "39400 tensor(5.4666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "391000\n",
      "39500 tensor(5.5847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "392000\n",
      "39600 tensor(5.2648, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "393000\n",
      "39700 tensor(5.2423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "394000\n",
      "39800 tensor(5.3731, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "395000\n",
      "39900 tensor(5.2014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "396000\n",
      "40000 tensor(5.2903, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "397000\n",
      "40100 tensor(5.3712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "398000\n",
      "40200 tensor(5.3557, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "399000\n",
      "40300 tensor(5.4151, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40400 tensor(5.4358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "400000\n",
      "40500 tensor(5.3498, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "401000\n",
      "40600 tensor(5.4152, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "402000\n",
      "40700 tensor(5.4551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "403000\n",
      "40800 tensor(5.4138, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "404000\n",
      "40900 tensor(5.3628, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "405000\n",
      "41000 tensor(5.4124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "406000\n",
      "41100 tensor(5.3750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "407000\n",
      "41200 tensor(5.2687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "408000\n",
      "41300 tensor(5.3987, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "409000\n",
      "41400 tensor(5.2976, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "410000\n",
      "41500 tensor(5.4418, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "411000\n",
      "41600 tensor(5.3558, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "412000\n",
      "41700 tensor(5.3767, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "413000\n",
      "41800 tensor(5.3836, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "414000\n",
      "41900 tensor(5.3904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "415000\n",
      "42000 tensor(5.5445, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "416000\n",
      "42100 tensor(5.2890, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "417000\n",
      "42200 tensor(5.3691, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "418000\n",
      "42300 tensor(5.4364, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "419000\n",
      "42400 tensor(5.2507, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "420000\n",
      "42500 tensor(5.4215, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "421000\n",
      "42600 tensor(5.2136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "422000\n",
      "42700 tensor(5.5296, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "423000\n",
      "42800 tensor(5.4544, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "424000\n",
      "42900 tensor(5.3009, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "425000\n",
      "43000 tensor(5.4403, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "426000\n",
      "43100 tensor(5.4384, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "427000\n",
      "43200 tensor(5.2520, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "428000\n",
      "43300 tensor(5.2945, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "429000\n",
      "43400 tensor(5.4455, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "430000\n",
      "43500 tensor(5.1633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "431000\n",
      "43600 tensor(5.3649, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "432000\n",
      "epoch: = 3\n",
      "0 tensor(5.3427, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "1000\n",
      "100 tensor(5.4180, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200 tensor(5.2939, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "2000\n",
      "300 tensor(5.3083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "3000\n",
      "400 tensor(5.3086, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "4000\n",
      "500 tensor(5.4733, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "5000\n",
      "600 tensor(5.2627, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "6000\n",
      "700 tensor(5.5664, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "7000\n",
      "800 tensor(5.1641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "8000\n",
      "900 tensor(5.4272, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9000\n",
      "1000 tensor(5.2926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "10000\n",
      "1100 tensor(5.4848, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "11000\n",
      "1200 tensor(5.5283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "12000\n",
      "1300 tensor(5.4635, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "13000\n",
      "1400 tensor(5.4590, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "14000\n",
      "1500 tensor(5.5386, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "15000\n",
      "1600 tensor(5.2150, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "16000\n",
      "1700 tensor(5.3116, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "17000\n",
      "1800 tensor(5.3130, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "18000\n",
      "1900 tensor(5.2889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "19000\n",
      "2000 tensor(5.3574, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20000\n",
      "2100 tensor(5.4860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "21000\n",
      "2200 tensor(5.3206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "22000\n",
      "2300 tensor(5.3447, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "23000\n",
      "2400 tensor(5.3333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "24000\n",
      "2500 tensor(5.3822, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "25000\n",
      "2600 tensor(5.4039, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "26000\n",
      "2700 tensor(5.4280, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "27000\n",
      "2800 tensor(5.4575, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "28000\n",
      "2900 tensor(5.5878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "29000\n",
      "3000 tensor(5.3311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30000\n",
      "3100 tensor(5.4103, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "31000\n",
      "3200 tensor(5.4323, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "32000\n",
      "3300 tensor(5.3521, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "33000\n",
      "3400 tensor(5.2512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "34000\n",
      "3500 tensor(5.3813, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35000\n",
      "3600 tensor(5.4000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "36000\n",
      "3700 tensor(5.3312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "37000\n",
      "3800 tensor(5.3553, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "38000\n",
      "3900 tensor(5.2275, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "39000\n",
      "4000 tensor(5.2883, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40000\n",
      "4100 tensor(5.4294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "41000\n",
      "4200 tensor(5.4801, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "42000\n",
      "4300 tensor(5.3863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "43000\n",
      "4400 tensor(5.4470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "44000\n",
      "4500 tensor(5.2610, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "45000\n",
      "4600 tensor(5.5962, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "46000\n",
      "4700 tensor(5.3029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "47000\n",
      "4800 tensor(5.4265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "48000\n",
      "4900 tensor(5.4823, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "49000\n",
      "5000 tensor(5.4749, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "50000\n",
      "5100 tensor(5.5356, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "51000\n",
      "5200 tensor(5.5513, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "52000\n",
      "5300 tensor(5.5476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "53000\n",
      "5400 tensor(5.4039, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "54000\n",
      "5500 tensor(5.5156, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "55000\n",
      "5600 tensor(5.2975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "56000\n",
      "5700 tensor(5.5492, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "57000\n",
      "5800 tensor(5.5379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "58000\n",
      "5900 tensor(5.4874, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "59000\n",
      "6000 tensor(5.3808, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "60000\n",
      "6100 tensor(5.3932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "61000\n",
      "6200 tensor(5.5657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "62000\n",
      "6300 tensor(5.4233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "63000\n",
      "6400 tensor(5.3438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "64000\n",
      "6500 tensor(5.2002, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "65000\n",
      "6600 tensor(5.3774, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "66000\n",
      "6700 tensor(5.3193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "67000\n",
      "6800 tensor(5.5394, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "68000\n",
      "6900 tensor(5.5196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "69000\n",
      "7000 tensor(5.4282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "70000\n",
      "7100 tensor(5.2296, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "71000\n",
      "7200 tensor(5.3175, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "72000\n",
      "7300 tensor(5.5642, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "73000\n",
      "7400 tensor(5.3784, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "74000\n",
      "7500 tensor(5.2475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "75000\n",
      "7600 tensor(5.3194, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "76000\n",
      "7700 tensor(5.3934, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "77000\n",
      "7800 tensor(5.5041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "78000\n",
      "7900 tensor(5.1814, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "79000\n",
      "8000 tensor(5.2426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "80000\n",
      "8100 tensor(5.4104, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "81000\n",
      "8200 tensor(5.4198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "82000\n",
      "8300 tensor(5.3854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "83000\n",
      "8400 tensor(5.5128, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "84000\n",
      "8500 tensor(5.4898, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "85000\n",
      "8600 tensor(5.4943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "86000\n",
      "8700 tensor(5.6012, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "87000\n",
      "8800 tensor(5.4790, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "88000\n",
      "8900 tensor(5.3312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "89000\n",
      "9000 tensor(5.4456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "90000\n",
      "9100 tensor(5.4537, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "91000\n",
      "9200 tensor(5.3643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "92000\n",
      "9300 tensor(5.4085, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "93000\n",
      "9400 tensor(5.2527, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "94000\n",
      "9500 tensor(5.3289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "95000\n",
      "9600 tensor(5.4516, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "96000\n",
      "9700 tensor(5.3881, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "97000\n",
      "9800 tensor(5.4321, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9900 tensor(5.2532, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "98000\n",
      "10000 tensor(5.4727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "99000\n",
      "10100 tensor(5.3607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "100000\n",
      "10200 tensor(5.2989, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "101000\n",
      "10300 tensor(5.4168, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "102000\n",
      "10400 tensor(5.4272, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "103000\n",
      "10500 tensor(5.4838, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "104000\n",
      "10600 tensor(5.5675, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "105000\n",
      "10700 tensor(5.4027, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "106000\n",
      "10800 tensor(5.4252, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "107000\n",
      "10900 tensor(5.3408, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "108000\n",
      "11000 tensor(5.5754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "109000\n",
      "11100 tensor(5.1920, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "110000\n",
      "11200 tensor(5.3604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "111000\n",
      "11300 tensor(5.3836, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "112000\n",
      "11400 tensor(5.3330, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "113000\n",
      "11500 tensor(5.4023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "114000\n",
      "11600 tensor(5.3923, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "115000\n",
      "11700 tensor(5.3145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "116000\n",
      "11800 tensor(5.5174, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "117000\n",
      "11900 tensor(5.3522, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "118000\n",
      "12000 tensor(5.4232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "119000\n",
      "12100 tensor(5.4382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "120000\n",
      "12200 tensor(5.4488, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "121000\n",
      "12300 tensor(5.5409, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "122000\n",
      "12400 tensor(5.4200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "123000\n",
      "12500 tensor(5.3292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "124000\n",
      "12600 tensor(5.3788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "125000\n",
      "12700 tensor(5.3116, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "126000\n",
      "12800 tensor(5.4948, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "127000\n",
      "12900 tensor(5.3557, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "128000\n",
      "13000 tensor(5.1732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "129000\n",
      "13100 tensor(5.3782, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "130000\n",
      "13200 tensor(5.4178, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "131000\n",
      "13300 tensor(5.2929, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "132000\n",
      "13400 tensor(5.3806, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "133000\n",
      "13500 tensor(5.3394, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "134000\n",
      "13600 tensor(5.4191, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "135000\n",
      "13700 tensor(5.3856, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "136000\n",
      "13800 tensor(5.3839, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "137000\n",
      "13900 tensor(5.2391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "138000\n",
      "14000 tensor(5.4865, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "139000\n",
      "14100 tensor(5.1952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "140000\n",
      "14200 tensor(5.4670, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "141000\n",
      "14300 tensor(5.4385, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "142000\n",
      "14400 tensor(5.3347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "143000\n",
      "14500 tensor(5.4370, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "144000\n",
      "14600 tensor(5.4695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "145000\n",
      "14700 tensor(5.3453, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "146000\n",
      "14800 tensor(5.7928, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "147000\n",
      "14900 tensor(5.4451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "148000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15000 tensor(5.3087, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "149000\n",
      "15100 tensor(5.5241, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "150000\n",
      "15200 tensor(5.3894, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "151000\n",
      "15300 tensor(5.3809, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "152000\n",
      "15400 tensor(5.2696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "153000\n",
      "15500 tensor(5.4343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "154000\n",
      "15600 tensor(5.4322, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "155000\n",
      "15700 tensor(5.3296, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "156000\n",
      "15800 tensor(5.2456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "157000\n",
      "15900 tensor(5.3806, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "158000\n",
      "16000 tensor(5.2008, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "159000\n",
      "16100 tensor(5.2489, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "160000\n",
      "16200 tensor(5.5902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "161000\n",
      "16300 tensor(5.4159, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "162000\n",
      "16400 tensor(5.3966, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "163000\n",
      "16500 tensor(5.5113, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "164000\n",
      "16600 tensor(5.3599, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "165000\n",
      "16700 tensor(5.3372, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "166000\n",
      "16800 tensor(5.4158, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "167000\n",
      "16900 tensor(5.1788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "168000\n",
      "17000 tensor(5.4497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "169000\n",
      "17100 tensor(5.2981, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "170000\n",
      "17200 tensor(5.4330, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "171000\n",
      "17300 tensor(5.4495, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "172000\n",
      "17400 tensor(5.2431, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "173000\n",
      "17500 tensor(5.2652, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "174000\n",
      "17600 tensor(5.3007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "175000\n",
      "17700 tensor(5.2852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "176000\n",
      "17800 tensor(5.3431, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "177000\n",
      "17900 tensor(5.2395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "178000\n",
      "18000 tensor(5.4841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "179000\n",
      "18100 tensor(5.4218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "180000\n",
      "18200 tensor(5.3397, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "181000\n",
      "18300 tensor(5.3426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "182000\n",
      "18400 tensor(5.3654, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "183000\n",
      "18500 tensor(5.3484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "184000\n",
      "18600 tensor(5.5509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "185000\n",
      "18700 tensor(5.3702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "186000\n",
      "18800 tensor(5.5361, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "187000\n",
      "18900 tensor(5.4132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "188000\n",
      "19000 tensor(5.4235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "189000\n",
      "19100 tensor(5.5318, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "190000\n",
      "19200 tensor(5.4136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "191000\n",
      "19300 tensor(5.5053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "192000\n",
      "19400 tensor(5.3472, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "193000\n",
      "19500 tensor(5.3511, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "194000\n",
      "19600 tensor(5.3861, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "195000\n",
      "19700 tensor(5.4345, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "196000\n",
      "19800 tensor(5.3067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "197000\n",
      "19900 tensor(5.3079, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "198000\n",
      "20000 tensor(5.3268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "199000\n",
      "20100 tensor(5.2668, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200000\n",
      "20200 tensor(5.1998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "201000\n",
      "20300 tensor(5.3105, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20400 tensor(5.3584, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "202000\n",
      "20500 tensor(5.3580, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "203000\n",
      "20600 tensor(5.5528, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "204000\n",
      "20700 tensor(5.3871, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "205000\n",
      "20800 tensor(5.2208, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "206000\n",
      "20900 tensor(5.5007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "207000\n",
      "21000 tensor(5.3396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "208000\n",
      "21100 tensor(5.3407, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "209000\n",
      "21200 tensor(5.2243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "210000\n",
      "21300 tensor(5.4206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "211000\n",
      "21400 tensor(5.4574, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "212000\n",
      "21500 tensor(5.2328, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "213000\n",
      "21600 tensor(5.2233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "214000\n",
      "21700 tensor(5.2152, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "215000\n",
      "21800 tensor(5.3497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "216000\n",
      "21900 tensor(5.3425, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "217000\n",
      "22000 tensor(5.3277, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "218000\n",
      "22100 tensor(5.2012, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "219000\n",
      "22200 tensor(5.0736, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "220000\n",
      "22300 tensor(5.5070, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "221000\n",
      "22400 tensor(5.2190, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "222000\n",
      "22500 tensor(5.2434, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "223000\n",
      "22600 tensor(5.4325, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "224000\n",
      "22700 tensor(5.1909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "225000\n",
      "22800 tensor(5.4576, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "226000\n",
      "22900 tensor(5.5069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "227000\n",
      "23000 tensor(5.4041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "228000\n",
      "23100 tensor(5.3908, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "229000\n",
      "23200 tensor(5.3866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "230000\n",
      "23300 tensor(5.4714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "231000\n",
      "23400 tensor(5.4781, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "232000\n",
      "23500 tensor(5.3154, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "233000\n",
      "23600 tensor(5.2854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "234000\n",
      "23700 tensor(5.3050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "235000\n",
      "23800 tensor(5.1721, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "236000\n",
      "23900 tensor(5.2637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "237000\n",
      "24000 tensor(5.2519, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "238000\n",
      "24100 tensor(5.3407, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "239000\n",
      "24200 tensor(5.5137, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "240000\n",
      "24300 tensor(5.4080, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "241000\n",
      "24400 tensor(5.5379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "242000\n",
      "24500 tensor(5.3255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "243000\n",
      "24600 tensor(5.4515, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "244000\n",
      "24700 tensor(5.3535, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "245000\n",
      "24800 tensor(5.3935, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "246000\n",
      "24900 tensor(5.4553, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "247000\n",
      "25000 tensor(5.4708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "248000\n",
      "25100 tensor(5.3920, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "249000\n",
      "25200 tensor(5.4083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "250000\n",
      "25300 tensor(5.4332, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "251000\n",
      "25400 tensor(5.4136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "252000\n",
      "25500 tensor(5.3147, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "253000\n",
      "25600 tensor(5.5860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "254000\n",
      "25700 tensor(5.3490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "255000\n",
      "25800 tensor(5.4464, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "256000\n",
      "25900 tensor(5.3857, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "257000\n",
      "26000 tensor(5.3893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "258000\n",
      "26100 tensor(5.3041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "259000\n",
      "26200 tensor(5.2321, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "260000\n",
      "26300 tensor(5.4289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "261000\n",
      "26400 tensor(5.4663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "262000\n",
      "26500 tensor(5.1922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "263000\n",
      "26600 tensor(5.5283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "264000\n",
      "26700 tensor(5.3933, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "265000\n",
      "26800 tensor(5.5680, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "266000\n",
      "26900 tensor(5.3281, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "267000\n",
      "27000 tensor(5.2408, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "268000\n",
      "27100 tensor(5.2671, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "269000\n",
      "27200 tensor(5.3099, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "270000\n",
      "27300 tensor(5.5049, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "271000\n",
      "27400 tensor(5.3850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "272000\n",
      "27500 tensor(5.2843, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "273000\n",
      "27600 tensor(5.5777, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "274000\n",
      "27700 tensor(5.4017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "275000\n",
      "27800 tensor(5.3994, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "276000\n",
      "27900 tensor(5.5128, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "277000\n",
      "28000 tensor(5.3708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "278000\n",
      "28100 tensor(5.3382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "279000\n",
      "28200 tensor(5.4996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "280000\n",
      "28300 tensor(5.1214, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "281000\n",
      "28400 tensor(5.5647, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "282000\n",
      "28500 tensor(5.3959, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "283000\n",
      "28600 tensor(5.3312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "284000\n",
      "28700 tensor(5.4663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "285000\n",
      "28800 tensor(5.5155, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "286000\n",
      "28900 tensor(5.3872, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "287000\n",
      "29000 tensor(5.3017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "288000\n",
      "29100 tensor(5.0583, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "289000\n",
      "29200 tensor(5.2099, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "290000\n",
      "29300 tensor(5.4934, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "291000\n",
      "29400 tensor(5.6202, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "292000\n",
      "29500 tensor(5.4016, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "293000\n",
      "29600 tensor(5.2601, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "294000\n",
      "29700 tensor(5.4038, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "295000\n",
      "29800 tensor(5.2475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "296000\n",
      "29900 tensor(5.4960, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "297000\n",
      "30000 tensor(5.2438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "298000\n",
      "30100 tensor(5.3221, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "299000\n",
      "30200 tensor(5.2686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "300000\n",
      "30300 tensor(5.3735, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "301000\n",
      "30400 tensor(5.2057, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "302000\n",
      "30500 tensor(5.3767, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30600 tensor(5.3515, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "303000\n",
      "30700 tensor(5.3841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "304000\n",
      "30800 tensor(5.3889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "305000\n",
      "30900 tensor(5.4117, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "306000\n",
      "31000 tensor(5.5205, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "307000\n",
      "31100 tensor(5.1742, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "308000\n",
      "31200 tensor(5.2173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "309000\n",
      "31300 tensor(5.4785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "310000\n",
      "31400 tensor(5.2577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "311000\n",
      "31500 tensor(5.4429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "312000\n",
      "31600 tensor(5.4289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "313000\n",
      "31700 tensor(5.3961, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "314000\n",
      "31800 tensor(5.4999, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "315000\n",
      "31900 tensor(5.1248, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "316000\n",
      "32000 tensor(5.3122, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "317000\n",
      "32100 tensor(5.1931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "318000\n",
      "32200 tensor(5.5096, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "319000\n",
      "32300 tensor(5.4973, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "320000\n",
      "32400 tensor(5.4742, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "321000\n",
      "32500 tensor(5.2964, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "322000\n",
      "32600 tensor(5.4063, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "323000\n",
      "32700 tensor(5.3369, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "324000\n",
      "32800 tensor(5.5636, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "325000\n",
      "32900 tensor(5.4245, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "326000\n",
      "33000 tensor(5.2032, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "327000\n",
      "33100 tensor(5.4095, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "328000\n",
      "33200 tensor(5.5071, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "329000\n",
      "33300 tensor(5.2729, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "330000\n",
      "33400 tensor(5.5492, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "331000\n",
      "33500 tensor(5.3701, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "332000\n",
      "33600 tensor(5.3223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "333000\n",
      "33700 tensor(5.3725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "334000\n",
      "33800 tensor(5.4572, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "335000\n",
      "33900 tensor(5.1889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "336000\n",
      "34000 tensor(5.4090, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "337000\n",
      "34100 tensor(5.3798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "338000\n",
      "34200 tensor(5.4259, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "339000\n",
      "34300 tensor(5.2132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "340000\n",
      "34400 tensor(5.6692, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "341000\n",
      "34500 tensor(5.5324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "342000\n",
      "34600 tensor(5.4271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "343000\n",
      "34700 tensor(5.4978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "344000\n",
      "34800 tensor(5.5230, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "345000\n",
      "34900 tensor(5.5652, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "346000\n",
      "35000 tensor(5.5478, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "347000\n",
      "35100 tensor(5.3700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "348000\n",
      "35200 tensor(5.2958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "349000\n",
      "35300 tensor(5.5219, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "350000\n",
      "35400 tensor(5.1702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "351000\n",
      "35500 tensor(5.2604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "352000\n",
      "35600 tensor(5.3821, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "353000\n",
      "35700 tensor(5.2551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "354000\n",
      "35800 tensor(5.3840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "355000\n",
      "35900 tensor(5.3635, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "356000\n",
      "36000 tensor(5.1400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "357000\n",
      "36100 tensor(5.5134, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "358000\n",
      "36200 tensor(5.3632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "359000\n",
      "36300 tensor(5.6461, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "360000\n",
      "36400 tensor(5.3415, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "361000\n",
      "36500 tensor(5.3659, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "362000\n",
      "36600 tensor(5.3874, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "363000\n",
      "36700 tensor(5.1886, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "364000\n",
      "36800 tensor(5.2958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "365000\n",
      "36900 tensor(5.4094, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "366000\n",
      "37000 tensor(5.3023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "367000\n",
      "37100 tensor(5.3287, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "368000\n",
      "37200 tensor(5.3996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "369000\n",
      "37300 tensor(5.3001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "370000\n",
      "37400 tensor(5.6516, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "371000\n",
      "37500 tensor(5.3366, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "372000\n",
      "37600 tensor(5.3282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "373000\n",
      "37700 tensor(5.5061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "374000\n",
      "37800 tensor(5.3408, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "375000\n",
      "37900 tensor(5.3203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "376000\n",
      "38000 tensor(5.3996, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "377000\n",
      "38100 tensor(5.4133, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "378000\n",
      "38200 tensor(5.4262, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "379000\n",
      "38300 tensor(5.3305, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "380000\n",
      "38400 tensor(5.3983, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "381000\n",
      "38500 tensor(5.4246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "382000\n",
      "38600 tensor(5.3713, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "383000\n",
      "38700 tensor(5.3634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "384000\n",
      "38800 tensor(5.4504, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "385000\n",
      "38900 tensor(5.5273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "386000\n",
      "39000 tensor(5.2229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "387000\n",
      "39100 tensor(5.4503, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "388000\n",
      "39200 tensor(5.5406, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "389000\n",
      "39300 tensor(5.3640, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "390000\n",
      "39400 tensor(5.4311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "391000\n",
      "39500 tensor(5.5292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "392000\n",
      "39600 tensor(5.2217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "393000\n",
      "39700 tensor(5.2121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "394000\n",
      "39800 tensor(5.3415, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "395000\n",
      "39900 tensor(5.1605, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "396000\n",
      "40000 tensor(5.2472, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "397000\n",
      "40100 tensor(5.3351, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "398000\n",
      "40200 tensor(5.3198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "399000\n",
      "40300 tensor(5.3862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40400 tensor(5.3946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "400000\n",
      "40500 tensor(5.3120, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "401000\n",
      "40600 tensor(5.3741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "402000\n",
      "40700 tensor(5.4199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "403000\n",
      "40800 tensor(5.3702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "404000\n",
      "40900 tensor(5.3212, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "405000\n",
      "41000 tensor(5.3683, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "406000\n",
      "41100 tensor(5.3491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "407000\n",
      "41200 tensor(5.2400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "408000\n",
      "41300 tensor(5.3728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "409000\n",
      "41400 tensor(5.2643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "410000\n",
      "41500 tensor(5.4064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "411000\n",
      "41600 tensor(5.3238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "412000\n",
      "41700 tensor(5.3469, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "413000\n",
      "41800 tensor(5.3432, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "414000\n",
      "41900 tensor(5.3521, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "415000\n",
      "42000 tensor(5.5087, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "416000\n",
      "42100 tensor(5.2556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "417000\n",
      "42200 tensor(5.3407, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "418000\n",
      "42300 tensor(5.4058, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "419000\n",
      "42400 tensor(5.2231, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "420000\n",
      "42500 tensor(5.3912, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "421000\n",
      "42600 tensor(5.1878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "422000\n",
      "42700 tensor(5.4955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "423000\n",
      "42800 tensor(5.4193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "424000\n",
      "42900 tensor(5.2662, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "425000\n",
      "43000 tensor(5.4093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "426000\n",
      "43100 tensor(5.4089, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "427000\n",
      "43200 tensor(5.2223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "428000\n",
      "43300 tensor(5.2456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "429000\n",
      "43400 tensor(5.4129, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "430000\n",
      "43500 tensor(5.1283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "431000\n",
      "43600 tensor(5.3275, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "432000\n",
      "epoch: = 4\n",
      "0 tensor(5.3172, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "1000\n",
      "100 tensor(5.3864, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200 tensor(5.2618, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "2000\n",
      "300 tensor(5.2652, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "3000\n",
      "400 tensor(5.2749, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "4000\n",
      "500 tensor(5.4347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "5000\n",
      "600 tensor(5.2271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "6000\n",
      "700 tensor(5.5396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "7000\n",
      "800 tensor(5.1379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "8000\n",
      "900 tensor(5.3861, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9000\n",
      "1000 tensor(5.2629, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "10000\n",
      "1100 tensor(5.4575, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "11000\n",
      "1200 tensor(5.4936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "12000\n",
      "1300 tensor(5.4281, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "13000\n",
      "1400 tensor(5.4186, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "14000\n",
      "1500 tensor(5.5070, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "15000\n",
      "1600 tensor(5.1769, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "16000\n",
      "1700 tensor(5.2856, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "17000\n",
      "1800 tensor(5.2827, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "18000\n",
      "1900 tensor(5.2544, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "19000\n",
      "2000 tensor(5.3218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20000\n",
      "2100 tensor(5.4549, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "21000\n",
      "2200 tensor(5.2864, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "22000\n",
      "2300 tensor(5.3145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "23000\n",
      "2400 tensor(5.2987, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "24000\n",
      "2500 tensor(5.3498, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "25000\n",
      "2600 tensor(5.3730, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "26000\n",
      "2700 tensor(5.4017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "27000\n",
      "2800 tensor(5.4255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "28000\n",
      "2900 tensor(5.5475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "29000\n",
      "3000 tensor(5.2988, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30000\n",
      "3100 tensor(5.3753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "31000\n",
      "3200 tensor(5.4049, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "32000\n",
      "3300 tensor(5.3206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "33000\n",
      "3400 tensor(5.2159, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "34000\n",
      "3500 tensor(5.3423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "35000\n",
      "3600 tensor(5.3717, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "36000\n",
      "3700 tensor(5.3042, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "37000\n",
      "3800 tensor(5.3258, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "38000\n",
      "3900 tensor(5.1989, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "39000\n",
      "4000 tensor(5.2650, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40000\n",
      "4100 tensor(5.3953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "41000\n",
      "4200 tensor(5.4542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "42000\n",
      "4300 tensor(5.3466, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "43000\n",
      "4400 tensor(5.4222, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "44000\n",
      "4500 tensor(5.2254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "45000\n",
      "4600 tensor(5.5610, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "46000\n",
      "4700 tensor(5.2753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "47000\n",
      "4800 tensor(5.4028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "48000\n",
      "4900 tensor(5.4516, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "49000\n",
      "5000 tensor(5.4464, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "50000\n",
      "5100 tensor(5.5018, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "51000\n",
      "5200 tensor(5.5194, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "52000\n",
      "5300 tensor(5.5077, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "53000\n",
      "5400 tensor(5.3746, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "54000\n",
      "5500 tensor(5.4847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "55000\n",
      "5600 tensor(5.2664, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "56000\n",
      "5700 tensor(5.5265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "57000\n",
      "5800 tensor(5.5101, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "58000\n",
      "5900 tensor(5.4513, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "59000\n",
      "6000 tensor(5.3554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "60000\n",
      "6100 tensor(5.3616, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "61000\n",
      "6200 tensor(5.5360, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "62000\n",
      "6300 tensor(5.3952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "63000\n",
      "6400 tensor(5.3132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "64000\n",
      "6500 tensor(5.1732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "65000\n",
      "6600 tensor(5.3505, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "66000\n",
      "6700 tensor(5.2919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "67000\n",
      "6800 tensor(5.5064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "68000\n",
      "6900 tensor(5.4881, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "69000\n",
      "7000 tensor(5.3978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "70000\n",
      "7100 tensor(5.2030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "71000\n",
      "7200 tensor(5.2738, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "72000\n",
      "7300 tensor(5.5317, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "73000\n",
      "7400 tensor(5.3487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "74000\n",
      "7500 tensor(5.2133, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "75000\n",
      "7600 tensor(5.2878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "76000\n",
      "7700 tensor(5.3644, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "77000\n",
      "7800 tensor(5.4711, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "78000\n",
      "7900 tensor(5.1445, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "79000\n",
      "8000 tensor(5.2138, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "80000\n",
      "8100 tensor(5.3741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "81000\n",
      "8200 tensor(5.3893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "82000\n",
      "8300 tensor(5.3492, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "83000\n",
      "8400 tensor(5.4797, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "84000\n",
      "8500 tensor(5.4501, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "85000\n",
      "8600 tensor(5.4600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "86000\n",
      "8700 tensor(5.5758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "87000\n",
      "8800 tensor(5.4493, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "88000\n",
      "8900 tensor(5.3035, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "89000\n",
      "9000 tensor(5.4164, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "90000\n",
      "9100 tensor(5.4273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "91000\n",
      "9200 tensor(5.3343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "92000\n",
      "9300 tensor(5.3797, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "93000\n",
      "9400 tensor(5.2260, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "94000\n",
      "9500 tensor(5.3006, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "95000\n",
      "9600 tensor(5.4211, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "96000\n",
      "9700 tensor(5.3615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "97000\n",
      "9800 tensor(5.4089, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "9900 tensor(5.2200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "98000\n",
      "10000 tensor(5.4428, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "99000\n",
      "10100 tensor(5.3219, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "100000\n",
      "10200 tensor(5.2692, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "101000\n",
      "10300 tensor(5.3854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "102000\n",
      "10400 tensor(5.3984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "103000\n",
      "10500 tensor(5.4516, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "104000\n",
      "10600 tensor(5.5380, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "105000\n",
      "10700 tensor(5.3724, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "106000\n",
      "10800 tensor(5.3862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "107000\n",
      "10900 tensor(5.3102, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "108000\n",
      "11000 tensor(5.5487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "109000\n",
      "11100 tensor(5.1684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "110000\n",
      "11200 tensor(5.3303, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "111000\n",
      "11300 tensor(5.3537, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "112000\n",
      "11400 tensor(5.3064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "113000\n",
      "11500 tensor(5.3775, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "114000\n",
      "11600 tensor(5.3649, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "115000\n",
      "11700 tensor(5.2920, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "116000\n",
      "11800 tensor(5.4908, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "117000\n",
      "11900 tensor(5.3293, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "118000\n",
      "12000 tensor(5.3926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "119000\n",
      "12100 tensor(5.4045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "120000\n",
      "12200 tensor(5.4246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "121000\n",
      "12300 tensor(5.5096, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "122000\n",
      "12400 tensor(5.3884, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "123000\n",
      "12500 tensor(5.3057, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "124000\n",
      "12600 tensor(5.3466, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "125000\n",
      "12700 tensor(5.2898, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "126000\n",
      "12800 tensor(5.4714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "127000\n",
      "12900 tensor(5.3255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "128000\n",
      "13000 tensor(5.1438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "129000\n",
      "13100 tensor(5.3498, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "130000\n",
      "13200 tensor(5.3890, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "131000\n",
      "13300 tensor(5.2710, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "132000\n",
      "13400 tensor(5.3541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "133000\n",
      "13500 tensor(5.3156, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "134000\n",
      "13600 tensor(5.3957, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "135000\n",
      "13700 tensor(5.3548, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "136000\n",
      "13800 tensor(5.3577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "137000\n",
      "13900 tensor(5.2122, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "138000\n",
      "14000 tensor(5.4587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "139000\n",
      "14100 tensor(5.1704, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "140000\n",
      "14200 tensor(5.4419, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "141000\n",
      "14300 tensor(5.4142, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "142000\n",
      "14400 tensor(5.3058, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "143000\n",
      "14500 tensor(5.4082, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "144000\n",
      "14600 tensor(5.4414, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "145000\n",
      "14700 tensor(5.3177, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "146000\n",
      "14800 tensor(5.7665, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "147000\n",
      "14900 tensor(5.4171, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "148000\n",
      "15000 tensor(5.2698, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "149000\n",
      "15100 tensor(5.4915, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "150000\n",
      "15200 tensor(5.3576, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "151000\n",
      "15300 tensor(5.3567, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "152000\n",
      "15400 tensor(5.2379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "153000\n",
      "15500 tensor(5.4092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "154000\n",
      "15600 tensor(5.4042, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "155000\n",
      "15700 tensor(5.3017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "156000\n",
      "15800 tensor(5.2188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "157000\n",
      "15900 tensor(5.3497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "158000\n",
      "16000 tensor(5.1718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "159000\n",
      "16100 tensor(5.2145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "160000\n",
      "16200 tensor(5.5591, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "161000\n",
      "16300 tensor(5.3864, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "162000\n",
      "16400 tensor(5.3719, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "163000\n",
      "16500 tensor(5.4842, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "164000\n",
      "16600 tensor(5.3329, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "165000\n",
      "16700 tensor(5.3130, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "166000\n",
      "16800 tensor(5.3903, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "167000\n",
      "16900 tensor(5.1551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "168000\n",
      "17000 tensor(5.4229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "169000\n",
      "17100 tensor(5.2686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "170000\n",
      "17200 tensor(5.4099, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "171000\n",
      "17300 tensor(5.4198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "172000\n",
      "17400 tensor(5.2162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "173000\n",
      "17500 tensor(5.2385, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "174000\n",
      "17600 tensor(5.2786, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "175000\n",
      "17700 tensor(5.2576, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "176000\n",
      "17800 tensor(5.3158, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "177000\n",
      "17900 tensor(5.2105, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "178000\n",
      "18000 tensor(5.4627, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "179000\n",
      "18100 tensor(5.3966, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "180000\n",
      "18200 tensor(5.3108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "181000\n",
      "18300 tensor(5.3148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "182000\n",
      "18400 tensor(5.3321, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "183000\n",
      "18500 tensor(5.3171, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "184000\n",
      "18600 tensor(5.5247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "185000\n",
      "18700 tensor(5.3469, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "186000\n",
      "18800 tensor(5.5092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "187000\n",
      "18900 tensor(5.3902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "188000\n",
      "19000 tensor(5.3904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "189000\n",
      "19100 tensor(5.5019, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "190000\n",
      "19200 tensor(5.3838, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "191000\n",
      "19300 tensor(5.4674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "192000\n",
      "19400 tensor(5.3223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "193000\n",
      "19500 tensor(5.3235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "194000\n",
      "19600 tensor(5.3589, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "195000\n",
      "19700 tensor(5.4063, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "196000\n",
      "19800 tensor(5.2838, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "197000\n",
      "19900 tensor(5.2807, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "198000\n",
      "20000 tensor(5.3038, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "199000\n",
      "20100 tensor(5.2397, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "200000\n",
      "20200 tensor(5.1723, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "201000\n",
      "20300 tensor(5.2827, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "20400 tensor(5.3245, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "202000\n",
      "20500 tensor(5.3303, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "203000\n",
      "20600 tensor(5.5211, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "204000\n",
      "20700 tensor(5.3629, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "205000\n",
      "20800 tensor(5.1882, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "206000\n",
      "20900 tensor(5.4671, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "207000\n",
      "21000 tensor(5.3110, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "208000\n",
      "21100 tensor(5.3181, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "209000\n",
      "21200 tensor(5.1968, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "210000\n",
      "21300 tensor(5.3940, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "211000\n",
      "21400 tensor(5.4308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "212000\n",
      "21500 tensor(5.2127, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "213000\n",
      "21600 tensor(5.2003, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "214000\n",
      "21700 tensor(5.1881, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "215000\n",
      "21800 tensor(5.3180, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "216000\n",
      "21900 tensor(5.3197, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "217000\n",
      "22000 tensor(5.3005, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "218000\n",
      "22100 tensor(5.1776, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "219000\n",
      "22200 tensor(5.0509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "220000\n",
      "22300 tensor(5.4807, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "221000\n",
      "22400 tensor(5.2040, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "222000\n",
      "22500 tensor(5.2161, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "223000\n",
      "22600 tensor(5.4083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "224000\n",
      "22700 tensor(5.1619, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "225000\n",
      "22800 tensor(5.4301, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "226000\n",
      "22900 tensor(5.4791, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "227000\n",
      "23000 tensor(5.3785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "228000\n",
      "23100 tensor(5.3705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "229000\n",
      "23200 tensor(5.3633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "230000\n",
      "23300 tensor(5.4443, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "231000\n",
      "23400 tensor(5.4496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "232000\n",
      "23500 tensor(5.2961, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "233000\n",
      "23600 tensor(5.2603, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "234000\n",
      "23700 tensor(5.2793, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "235000\n",
      "23800 tensor(5.1461, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "236000\n",
      "23900 tensor(5.2376, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "237000\n",
      "24000 tensor(5.2269, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "238000\n",
      "24100 tensor(5.3154, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "239000\n",
      "24200 tensor(5.4852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "240000\n",
      "24300 tensor(5.3785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "241000\n",
      "24400 tensor(5.5053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "242000\n",
      "24500 tensor(5.2987, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "243000\n",
      "24600 tensor(5.4275, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "244000\n",
      "24700 tensor(5.3283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "245000\n",
      "24800 tensor(5.3707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "246000\n",
      "24900 tensor(5.4294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "247000\n",
      "25000 tensor(5.4479, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "248000\n",
      "25100 tensor(5.3629, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "249000\n",
      "25200 tensor(5.3849, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "250000\n",
      "25300 tensor(5.4124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "251000\n",
      "25400 tensor(5.3932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "252000\n",
      "25500 tensor(5.2893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "253000\n",
      "25600 tensor(5.5512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "254000\n",
      "25700 tensor(5.3227, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "255000\n",
      "25800 tensor(5.4217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "256000\n",
      "25900 tensor(5.3637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "257000\n",
      "26000 tensor(5.3632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "258000\n",
      "26100 tensor(5.2841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "259000\n",
      "26200 tensor(5.2107, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "260000\n",
      "26300 tensor(5.4024, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "261000\n",
      "26400 tensor(5.4410, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "262000\n",
      "26500 tensor(5.1685, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "263000\n",
      "26600 tensor(5.5023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "264000\n",
      "26700 tensor(5.3654, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "265000\n",
      "26800 tensor(5.5407, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "266000\n",
      "26900 tensor(5.3000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "267000\n",
      "27000 tensor(5.2141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "268000\n",
      "27100 tensor(5.2490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "269000\n",
      "27200 tensor(5.2850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "270000\n",
      "27300 tensor(5.4811, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "271000\n",
      "27400 tensor(5.3561, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "272000\n",
      "27500 tensor(5.2602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "273000\n",
      "27600 tensor(5.5429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "274000\n",
      "27700 tensor(5.3794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "275000\n",
      "27800 tensor(5.3792, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "276000\n",
      "27900 tensor(5.4873, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "277000\n",
      "28000 tensor(5.3454, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "278000\n",
      "28100 tensor(5.3113, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "279000\n",
      "28200 tensor(5.4785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "280000\n",
      "28300 tensor(5.1013, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "281000\n",
      "28400 tensor(5.5403, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "282000\n",
      "28500 tensor(5.3676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "283000\n",
      "28600 tensor(5.3108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "284000\n",
      "28700 tensor(5.4403, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "285000\n",
      "28800 tensor(5.4926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "286000\n",
      "28900 tensor(5.3638, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "287000\n",
      "29000 tensor(5.2819, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "288000\n",
      "29100 tensor(5.0362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "289000\n",
      "29200 tensor(5.1871, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "290000\n",
      "29300 tensor(5.4697, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "291000\n",
      "29400 tensor(5.5909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "292000\n",
      "29500 tensor(5.3807, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "293000\n",
      "29600 tensor(5.2398, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "294000\n",
      "29700 tensor(5.3690, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "295000\n",
      "29800 tensor(5.2220, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "296000\n",
      "29900 tensor(5.4597, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "297000\n",
      "30000 tensor(5.2205, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "298000\n",
      "30100 tensor(5.3061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "299000\n",
      "30200 tensor(5.2432, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "300000\n",
      "30300 tensor(5.3527, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "301000\n",
      "30400 tensor(5.1823, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "302000\n",
      "30500 tensor(5.3526, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "30600 tensor(5.3318, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "303000\n",
      "30700 tensor(5.3634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "304000\n",
      "30800 tensor(5.3571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "305000\n",
      "30900 tensor(5.3875, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "306000\n",
      "31000 tensor(5.4983, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "307000\n",
      "31100 tensor(5.1554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "308000\n",
      "31200 tensor(5.1952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "309000\n",
      "31300 tensor(5.4546, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "310000\n",
      "31400 tensor(5.2307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "311000\n",
      "31500 tensor(5.4188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "312000\n",
      "31600 tensor(5.4085, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "313000\n",
      "31700 tensor(5.3744, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "314000\n",
      "31800 tensor(5.4766, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "315000\n",
      "31900 tensor(5.1062, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "316000\n",
      "32000 tensor(5.2924, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "317000\n",
      "32100 tensor(5.1728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "318000\n",
      "32200 tensor(5.4863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "319000\n",
      "32300 tensor(5.4748, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "320000\n",
      "32400 tensor(5.4518, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "321000\n",
      "32500 tensor(5.2752, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "322000\n",
      "32600 tensor(5.3822, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "323000\n",
      "32700 tensor(5.3088, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "324000\n",
      "32800 tensor(5.5403, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "325000\n",
      "32900 tensor(5.4000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "326000\n",
      "33000 tensor(5.1837, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "327000\n",
      "33100 tensor(5.3888, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "328000\n",
      "33200 tensor(5.4849, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "329000\n",
      "33300 tensor(5.2471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "330000\n",
      "33400 tensor(5.5246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "331000\n",
      "33500 tensor(5.3479, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "332000\n",
      "33600 tensor(5.3043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "333000\n",
      "33700 tensor(5.3487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "334000\n",
      "33800 tensor(5.4368, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "335000\n",
      "33900 tensor(5.1620, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "336000\n",
      "34000 tensor(5.3873, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "337000\n",
      "34100 tensor(5.3545, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "338000\n",
      "34200 tensor(5.4001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "339000\n",
      "34300 tensor(5.1902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "340000\n",
      "34400 tensor(5.6453, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "341000\n",
      "34500 tensor(5.5124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "342000\n",
      "34600 tensor(5.4069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "343000\n",
      "34700 tensor(5.4734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "344000\n",
      "34800 tensor(5.5014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "345000\n",
      "34900 tensor(5.5412, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "346000\n",
      "35000 tensor(5.5132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "347000\n",
      "35100 tensor(5.3455, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "348000\n",
      "35200 tensor(5.2694, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "349000\n",
      "35300 tensor(5.4988, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "350000\n",
      "35400 tensor(5.1485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "351000\n",
      "35500 tensor(5.2299, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "352000\n",
      "35600 tensor(5.3643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "353000\n",
      "35700 tensor(5.2247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "354000\n",
      "35800 tensor(5.3615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "355000\n",
      "35900 tensor(5.3453, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "356000\n",
      "36000 tensor(5.1217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "357000\n",
      "36100 tensor(5.4909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "358000\n",
      "36200 tensor(5.3382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "359000\n",
      "36300 tensor(5.6225, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "360000\n",
      "36400 tensor(5.3167, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "361000\n",
      "36500 tensor(5.3458, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "362000\n",
      "36600 tensor(5.3608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "363000\n",
      "36700 tensor(5.1660, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "364000\n",
      "36800 tensor(5.2737, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "365000\n",
      "36900 tensor(5.3883, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "366000\n",
      "37000 tensor(5.2783, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "367000\n",
      "37100 tensor(5.3110, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "368000\n",
      "37200 tensor(5.3794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "369000\n",
      "37300 tensor(5.2802, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "370000\n",
      "37400 tensor(5.6133, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "371000\n",
      "37500 tensor(5.3138, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "372000\n",
      "37600 tensor(5.3083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "373000\n",
      "37700 tensor(5.4860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "374000\n",
      "37800 tensor(5.3216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "375000\n",
      "37900 tensor(5.2969, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "376000\n",
      "38000 tensor(5.3759, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "377000\n",
      "38100 tensor(5.3914, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "378000\n",
      "38200 tensor(5.4089, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "379000\n",
      "38300 tensor(5.3068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "380000\n",
      "38400 tensor(5.3798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "381000\n",
      "38500 tensor(5.4051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "382000\n",
      "38600 tensor(5.3471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "383000\n",
      "38700 tensor(5.3415, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "384000\n",
      "38800 tensor(5.4310, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "385000\n",
      "38900 tensor(5.5029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "386000\n",
      "39000 tensor(5.2021, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "387000\n",
      "39100 tensor(5.4283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "388000\n",
      "39200 tensor(5.5158, device='cuda:0', grad_fn=<NllLossBackward0>)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "389000\n",
      "39300 tensor(5.3452, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "390000\n",
      "39400 tensor(5.4111, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "391000\n",
      "39500 tensor(5.4969, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "392000\n",
      "39600 tensor(5.1952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "393000\n",
      "39700 tensor(5.1946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "394000\n",
      "39800 tensor(5.3234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "395000\n",
      "39900 tensor(5.1354, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "396000\n",
      "40000 tensor(5.2210, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "397000\n",
      "40100 tensor(5.3133, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "398000\n",
      "40200 tensor(5.2990, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "399000\n",
      "40300 tensor(5.3684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "40400 tensor(5.3700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "400000\n",
      "40500 tensor(5.2911, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "401000\n",
      "40600 tensor(5.3497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "402000\n",
      "40700 tensor(5.3981, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "403000\n",
      "40800 tensor(5.3436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "404000\n",
      "40900 tensor(5.2978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "405000\n",
      "41000 tensor(5.3420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "406000\n",
      "41100 tensor(5.3342, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "407000\n",
      "41200 tensor(5.2226, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "408000\n",
      "41300 tensor(5.3573, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "409000\n",
      "41400 tensor(5.2448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "410000\n",
      "41500 tensor(5.3863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "411000\n",
      "41600 tensor(5.3051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "412000\n",
      "41700 tensor(5.3294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "413000\n",
      "41800 tensor(5.3191, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "414000\n",
      "41900 tensor(5.3289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "415000\n",
      "42000 tensor(5.4860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "416000\n",
      "42100 tensor(5.2358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "417000\n",
      "42200 tensor(5.3253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "418000\n",
      "42300 tensor(5.3869, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "419000\n",
      "42400 tensor(5.2062, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "420000\n",
      "42500 tensor(5.3712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "421000\n",
      "42600 tensor(5.1718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "422000\n",
      "42700 tensor(5.4735, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "423000\n",
      "42800 tensor(5.3973, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "424000\n",
      "42900 tensor(5.2447, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "425000\n",
      "43000 tensor(5.3896, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "426000\n",
      "43100 tensor(5.3916, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "427000\n",
      "43200 tensor(5.2044, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "428000\n",
      "43300 tensor(5.2167, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "429000\n",
      "43400 tensor(5.3933, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "430000\n",
      "43500 tensor(5.1078, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "431000\n",
      "43600 tensor(5.3045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
      "432000\n"
     ]
    }
   ],
   "source": [
    "data = DataLoader(train_dataset, batch_size=batch_s)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)\n",
    "criterion = torch.nn.NLLLoss()\n",
    "torch.cuda.empty_cache()\n",
    "gc.collect()\n",
    "\n",
    "model.load_state_dict(torch.load('model-bigram_final.bin'))\n",
    "for i in range(1, epochs+1):\n",
    "    print('epoch: =', i)\n",
    "    model.train()\n",
    "    step = 0\n",
    "    for x, y in data: # prev, predicting, following words\n",
    "       x = x.to(device)\n",
    "       y = y.to(device)\n",
    "       optimizer.zero_grad()\n",
    "       ypredicted = model(x) #previous, following word\n",
    "       loss = criterion(torch.log(ypredicted), y)\n",
    "       if step % 100 == 0:\n",
    "          print(step, loss)\n",
    "       step += 1\n",
    "       loss.backward()\n",
    "       optimizer.step()\n",
    "    torch.save(model.state_dict(), f'model-bigram_2nd-run{i}.bin')    \n",
    "torch.save(model.state_dict(), f'model-bigram_final.bin')    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('be', 11, 0.2570849657058716),\n",
       " ('<unk>', 0, 0.07411641627550125),\n",
       " ('not', 22, 0.05940083786845207),\n",
       " ('have', 28, 0.02751326560974121),\n",
       " ('bo', 167, 0.014936885796487331),\n",
       " ('make', 116, 0.013943656347692013),\n",
       " ('give', 193, 0.011286991648375988),\n",
       " ('take', 153, 0.011171611957252026),\n",
       " ('do', 86, 0.010088067501783371),\n",
       " ('he', 20, 0.009703895077109337)]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "device = 'cuda'\n",
    "torch.cuda.empty_cache()\n",
    "model = SimpleBigramNeuralLanguageModel(vocab_size, embed_size).to(device)\n",
    "model.load_state_dict(torch.load(f'model-bigram_final.bin'))\n",
    "model.eval()\n",
    "\n",
    "ixs = torch.tensor(vocab.forward(['will'])).to(device)\n",
    "\n",
    "out = model(ixs)\n",
    "top = torch.topk(out[0], 10)\n",
    "top_indices = top.indices.tolist()\n",
    "top_probs = top.values.tolist()\n",
    "top_words = vocab.lookup_tokens(top_indices)\n",
    "list(zip(top_words, top_indices, top_probs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('<unk>', 0, 0.19996878504753113),\n",
       " ('and', 3, 0.05288130044937134),\n",
       " ('of', 2, 0.042051784694194794),\n",
       " ('the', 1, 0.026572922244668007),\n",
       " ('to', 4, 0.022689413279294968),\n",
       " ('in', 6, 0.015904497355222702),\n",
       " ('The', 17, 0.012827681377530098),\n",
       " ('a', 5, 0.00961760152131319),\n",
       " ('for', 8, 0.008938422426581383),\n",
       " ('</s>', 32, 0.00840282253921032)]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab = train_dataset.vocab\n",
    "ixs = torch.tensor(vocab.forward(['cerned.'])).to(device)\n",
    "\n",
    "out = model(ixs)\n",
    "top = torch.topk(out[0], 10)\n",
    "top_indices = top.indices.tolist()\n",
    "top_probs = top.values.tolist()\n",
    "top_words = vocab.lookup_tokens(top_indices)\n",
    "list(zip(top_words, top_indices, top_probs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('<unk>', 0, 1.0),\n",
       " ('particular,', 14538, 0.24527804553508759),\n",
       " ('revolution.', 20446, 0.23776617646217346),\n",
       " ('Territory.', 14189, 0.23417341709136963),\n",
       " ('or-', 2261, 0.22888363897800446),\n",
       " ('3', 479, 0.2288265973329544),\n",
       " ('speak.', 13722, 0.2252315878868103),\n",
       " ('attend.', 19397, 0.22110989689826965),\n",
       " ('say,', 1455, 0.22106117010116577),\n",
       " ('Lee.', 15326, 0.21764159202575684)]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cos = nn.CosineSimilarity(dim=1, eps=1e-6)\n",
    "\n",
    "embeddings = model.model[0].weight\n",
    "\n",
    "vec = embeddings[vocab['cerned.']]\n",
    "\n",
    "similarities = cos(vec, embeddings)\n",
    "\n",
    "top = torch.topk(similarities, 10)\n",
    "\n",
    "top_indices = top.indices.tolist()\n",
    "top_probs = top.values.tolist()\n",
    "top_words = vocab.lookup_tokens(top_indices)\n",
    "list(zip(top_words, top_indices, top_probs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_values_from_model(presc_word, model, vocab, k):\n",
    "    ixs =  torch.tensor(vocab.forward([presc_word])).to(device)\n",
    "    out = model(ixs)\n",
    "    top = torch.topk(out[0], k)\n",
    "    top_indices = top.indices.tolist()\n",
    "    top_probs = top.values.tolist()\n",
    "    top_words = vocab.lookup_tokens(top_indices)\n",
    "    return list(zip(top_words, top_probs))\n",
    "\n",
    "def gonito_format(dic):\n",
    "    tab = summarize_probs_unk(dic)\n",
    "    result = ''\n",
    "    for element in tab[:-1]:\n",
    "        result+=str(element[0])+':'+str(element[1])+'\\t'\n",
    "    result+=':'+ str(tab[-1][1])+'\\n'\n",
    "    return result\n",
    "\n",
    "def summarize_probs_unk(dic):\n",
    "    if '<unk>' in dic.keys():\n",
    "        probsum = sum(float(val) for key, val in dic.items())\n",
    "        for key in dic:\n",
    "            dic[key] = dic[key]/probsum ###leave some space for wildcard\n",
    "        wildcard = dic['<unk>']\n",
    "        del dic['<unk>']\n",
    "        tab = [(key, val) for key, val in dic.items()]\n",
    "        tab.append(('<unk>', wildcard))\n",
    "    else:\n",
    "        probsum = sum(float(val) for key, val in dic.items())\n",
    "        for key in dic:\n",
    "            dic[key] = dic[key]/(probsum*(1+wildcard_minweight)) #plus, becouse it's denominator\n",
    "        tab = [(key, val) for key, val in dic.items()]\n",
    "        tab.append(('<unk>', 1-sum([val for val in dic.values()])))\n",
    "    return tab\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.load_state_dict(torch.load('model-bigram_final.bin'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "with lzma.open(test_file, 'rt') as file:\n",
    "    predict_words = []\n",
    "    results = []\n",
    "    for line in file:\n",
    "#         print(line)\n",
    "        line = preprocess(line) #get only relevant\n",
    "        split = line.split('\\t')\n",
    "        predict_words.append(get_last_word(split[0])) #get_first_word(split[1])\n",
    "    vocab = train_dataset.vocab\n",
    "    for presc_word in predict_words:\n",
    "        results.append(dict(get_values_from_model(presc_word, model, vocab, k=k)))\n",
    "    with open(out_file, 'w') as outfile:\n",
    "        for elem in results:\n",
    "            outfile.write(gonito_format(elem))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.6"
  },
  "org": null
 },
 "nbformat": 4,
 "nbformat_minor": 1
}