challenging-america-word-ga.../Zad_7.ipynb

4995 lines
260 KiB
Plaintext
Raw Normal View History

2023-04-28 09:29:54 +02:00
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"provenance": [],
"toc_visible": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"language_info": {
"name": "python"
},
"accelerator": "GPU",
"gpuClass": "standard"
},
"cells": [
{
"cell_type": "markdown",
"source": [
"## connect to google drive (working on colab)"
],
"metadata": {
"id": "G0ujnpy2tuBE"
}
},
{
"cell_type": "code",
"source": [
"from google.colab import drive\n",
"drive.mount('/content/drive')"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "Lwuh_S5pWY1j",
"outputId": "27838dab-7be0-4447-883a-95559887c7c8"
},
"execution_count": 1,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Mounted at /content/drive\n"
]
}
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "QNiUKMiqWLd0"
},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"source": [
"!mkdir moj7\n"
],
"metadata": {
"id": "vlnrhRaEWNJF"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"%cd drive"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "539oWG3pXOAX",
"outputId": "a9ae634d-d4a2-47dd-97d2-10c245c7c5d2"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"/content/drive\n"
]
}
]
},
{
"cell_type": "code",
"source": [
"%cd MyDrive"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "wgmiKs4BiAiT",
"outputId": "fbbd0bc7-76bd-47bf-e38b-e051239e5ba7"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"/content/drive/MyDrive\n"
]
}
]
},
{
"cell_type": "code",
"source": [
"%cd moj7"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "jNdkji_hiAlt",
"outputId": "962b875a-8d3f-433d-8d7b-dcd664ee1674"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"/content/drive/MyDrive/moj7\n"
]
}
]
},
{
"cell_type": "code",
"source": [],
"metadata": {
"id": "P249ENeSiAqn"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"!pwd"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "IOHV3Iz4WNLc",
"outputId": "f56a6ab7-73e6-4b03-824b-b5e749c8a82e"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"/content/drive/MyDrive/moj7\n"
]
}
]
},
{
"cell_type": "code",
"source": [],
"metadata": {
"id": "2pLkeHY5Z9oT"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## Preprocess"
],
"metadata": {
"id": "D7jhQfbttn9D"
}
},
{
"cell_type": "code",
"source": [
"import re"
],
"metadata": {
"id": "_IPWOt2BZ_-q"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"train_file ='train/in.tsv.xz'\n",
"test_file = 'test-A/in.tsv.xz'\n",
"out_file = 'test-A/out.tsv'\n",
"\n",
"def preprocess(line):\n",
" line = replace_endline(line)\n",
" line = get_rid_of_header(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 = re.sub(\"\\\\n|\\\\+\", \" \", line)\n",
" return line"
],
"metadata": {
"id": "qDnZdPblWNNr"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"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",
"\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'))\n",
"#vocab_size = 20000\n",
"vocab_size = 20000\n",
"\n",
"vocab = build_vocab_from_iterator(\n",
" get_word_lines_from_file(train_file),\n",
" max_tokens = vocab_size,\n",
" specials = ['<unk>'])\n",
"\n"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "RVN0lKVZfwMe",
"outputId": "305b03e4-f626-4560-a371-41bc5a0ea9c7"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"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"
]
}
]
},
{
"cell_type": "code",
"source": [
"vocab['no']"
],
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "b9vMOTlZxISl",
"outputId": "a3a71b17-3fb2-4794-ae5d-a43d9af49b69"
},
"execution_count": null,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"50"
]
},
"metadata": {},
"execution_count": 19
}
]
},
{
"cell_type": "code",
"source": [
"with open('filename.pickle', 'wb') as handle:\n",
" pickle.dump(vocab, handle, protocol=pickle.HIGHEST_PROTOCOL)"
],
"metadata": {
"id": "6R9l6tuPxB_B"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "markdown",
"source": [
"## Create NN"
],
"metadata": {
"id": "Be25rS6Uvl4V"
}
},
{
"cell_type": "code",
"source": [
"from torch import nn\n",
"import torch\n",
"import pickle\n",
"# embed_size = 150\n",
"embed_size = 150\n",
"\n",
"class Bigram(nn.Module):\n",
" def __init__(self, vocabulary_size, embedding_size):\n",
" super(Bigram, 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",
" def forward(self, x):\n",
" return self.model(x)\n",
"\n",
"model = Bigram(vocab_size, embed_size)\n",
"\n",
"vocab.set_default_index(vocab['<unk>'])\n",
"res = torch.tensor(vocab.forward(['order']))\n",
"print(res)\n"
],
"metadata": {
"id": "dGTOmcHwWNSi",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "88c30492-7a9a-4b96-9119-ecdf5865bb51"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"tensor([215])\n"
]
}
]
},
{
"cell_type": "code",
"source": [],
"metadata": {
"id": "mWZ_jw-hxNXk"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"from torch.utils.data import IterableDataset\n",
"import itertools\n",
"\n",
"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",
"\n",
"train_dataset = Bigrams(train_file, vocab_size)"
],
"metadata": {
"id": "5CSigeomWNVT",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "4fb3b1ff-f91b-4799-fc5f-17bae10b94ec"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"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"
]
}
]
},
{
"cell_type": "code",
"source": [
"from torch.utils.data import DataLoader\n",
"\n",
"next(iter(DataLoader(train_dataset, batch_size=5)))"
],
"metadata": {
"id": "oYAZ772rWNX7",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "da3a811a-36cf-4d34-82d8-b1ce2447c232"
},
"execution_count": null,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"[tensor([ 23, 191, 5791, 1, 112]),\n",
" tensor([ 191, 5791, 1, 112, 159])]"
]
},
"metadata": {},
"execution_count": 23
}
]
},
{
"cell_type": "markdown",
"source": [
"## Train"
],
"metadata": {
"id": "1H_dI372vrNh"
}
},
{
"cell_type": "code",
"source": [
"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
],
"metadata": {
"id": "N2u4Qmadgtdn"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"\n",
"model = Bigram(vocab_size, embed_size).to(device)\n",
"data = DataLoader(train_dataset, batch_size=1000)\n",
"optimizer = torch.optim.Adam(model.parameters())\n",
"criterion = torch.nn.NLLLoss()\n",
"## epochs=2\n",
"for i in range(2):\n",
" print('epoch: =', i)\n",
" model.train()\n",
" step = 0\n",
" for x, y in data:\n",
" x = x.to(device)\n",
" y = y.to(device)\n",
" optimizer.zero_grad()\n",
" ypredicted = model(x)\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(), 'model.bin') \n"
],
"metadata": {
"id": "OGk2tjbvWNag",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "868503a1-2849-40ff-e703-886fba094927"
},
"execution_count": null,
"outputs": [
{
"metadata": {
"tags": null
},
"name": "stdout",
"output_type": "stream",
"text": [
"epoch: = 0\n"
]
},
{
"metadata": {
"tags": null
},
"name": "stderr",
"output_type": "stream",
"text": [
"/usr/local/lib/python3.10/dist-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"
]
},
{
"output_type": "stream",
"name": "stdout",
"text": [
"0 tensor(10.3037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100 tensor(8.7506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"200 tensor(7.8141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1000\n",
"300 tensor(7.4218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"400 tensor(7.1627, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"500 tensor(6.7964, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2000\n",
"600 tensor(6.4704, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"700 tensor(6.3798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"800 tensor(6.2849, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3000\n",
"900 tensor(6.3975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1000 tensor(6.0096, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1100 tensor(5.7434, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4000\n",
"1200 tensor(5.9602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1300 tensor(6.1623, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1400 tensor(6.1647, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5000\n",
"1500 tensor(6.1010, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1600 tensor(6.0634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6000\n",
"1700 tensor(5.9149, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1800 tensor(5.7918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1900 tensor(5.6739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7000\n",
"2000 tensor(5.5298, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2100 tensor(5.8011, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2200 tensor(5.4338, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8000\n",
"2300 tensor(5.7522, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2400 tensor(5.0313, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2500 tensor(5.7116, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9000\n",
"2600 tensor(5.2706, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2700 tensor(5.6324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2800 tensor(5.0710, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10000\n",
"2900 tensor(5.5921, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3000 tensor(5.4808, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11000\n",
"3100 tensor(5.3611, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3200 tensor(5.6228, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3300 tensor(5.4286, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12000\n",
"3400 tensor(5.3550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3500 tensor(5.4032, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3600 tensor(5.1070, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13000\n",
"3700 tensor(5.4506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3800 tensor(5.4622, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3900 tensor(5.4984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14000\n",
"4000 tensor(5.1740, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4100 tensor(5.6064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4200 tensor(5.0705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15000\n",
"4300 tensor(5.5181, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4400 tensor(5.2919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16000\n",
"4500 tensor(5.5021, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4600 tensor(5.5308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4700 tensor(5.4699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17000\n",
"4800 tensor(5.2686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4900 tensor(5.4776, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5000 tensor(5.5061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18000\n",
"5100 tensor(5.3180, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5200 tensor(5.5524, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5300 tensor(5.3481, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19000\n",
"5400 tensor(5.2153, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5500 tensor(5.4478, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20000\n",
"5600 tensor(5.3441, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5700 tensor(5.3958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5800 tensor(5.8945, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21000\n",
"5900 tensor(5.5684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6000 tensor(5.5715, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6100 tensor(5.2367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22000\n",
"6200 tensor(5.6976, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6300 tensor(5.5367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6400 tensor(5.3024, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23000\n",
"6500 tensor(5.3010, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6600 tensor(6.0962, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6700 tensor(5.0961, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24000\n",
"6800 tensor(5.1091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6900 tensor(5.4123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25000\n",
"7000 tensor(5.3128, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7100 tensor(5.3416, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7200 tensor(5.4973, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26000\n",
"7300 tensor(5.4418, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7400 tensor(5.2171, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7500 tensor(5.6509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27000\n",
"7600 tensor(5.0550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7700 tensor(5.4937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7800 tensor(5.9218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28000\n",
"7900 tensor(5.2853, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8000 tensor(5.3146, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8100 tensor(4.8552, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29000\n",
"8200 tensor(5.3389, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8300 tensor(5.2421, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30000\n",
"8400 tensor(5.2460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8500 tensor(5.0331, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8600 tensor(5.0050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31000\n",
"8700 tensor(5.3844, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8800 tensor(5.4491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8900 tensor(5.6790, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32000\n",
"9000 tensor(5.1118, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9100 tensor(5.3567, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9200 tensor(5.4141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33000\n",
"9300 tensor(5.3085, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9400 tensor(5.2808, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34000\n",
"9500 tensor(5.0931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9600 tensor(5.1090, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9700 tensor(5.2519, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35000\n",
"9800 tensor(5.3852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9900 tensor(5.0943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10000 tensor(5.4690, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36000\n",
"10100 tensor(5.4348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10200 tensor(5.3262, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10300 tensor(5.4878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37000\n",
"10400 tensor(5.2384, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10500 tensor(5.2151, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10600 tensor(4.8722, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38000\n",
"10700 tensor(5.4325, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10800 tensor(4.8699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39000\n",
"10900 tensor(5.3448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11000 tensor(5.1358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11100 tensor(5.0432, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40000\n",
"11200 tensor(5.4062, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11300 tensor(5.4040, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11400 tensor(5.5312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41000\n",
"11500 tensor(5.4374, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11600 tensor(5.0998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11700 tensor(5.4217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42000\n",
"11800 tensor(5.5747, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11900 tensor(5.0467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12000 tensor(5.4270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43000\n",
"12100 tensor(5.2043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12200 tensor(5.2369, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44000\n",
"12300 tensor(5.4465, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12400 tensor(4.9839, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12500 tensor(5.3214, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45000\n",
"12600 tensor(5.1928, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12700 tensor(4.9646, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12800 tensor(5.3325, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46000\n",
"12900 tensor(5.4429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13000 tensor(5.0652, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13100 tensor(5.3126, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47000\n",
"13200 tensor(5.4124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13300 tensor(5.5385, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13400 tensor(5.0986, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48000\n",
"13500 tensor(5.2693, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13600 tensor(5.2136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49000\n",
"13700 tensor(5.5169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13800 tensor(5.1840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13900 tensor(5.2700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50000\n",
"14000 tensor(5.2077, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14100 tensor(5.3791, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14200 tensor(5.4008, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51000\n",
"14300 tensor(5.3506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14400 tensor(4.7662, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14500 tensor(4.9474, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52000\n",
"14600 tensor(5.0245, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14700 tensor(5.3977, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14800 tensor(4.9653, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53000\n",
"14900 tensor(4.8947, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15000 tensor(5.3548, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54000\n",
"15100 tensor(4.7244, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15200 tensor(4.9752, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15300 tensor(5.3929, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55000\n",
"15400 tensor(5.3096, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15500 tensor(5.1247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15600 tensor(5.2753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56000\n",
"15700 tensor(5.2373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15800 tensor(4.9997, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15900 tensor(5.1718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57000\n",
"16000 tensor(5.5952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16100 tensor(5.3699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16200 tensor(5.0923, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58000\n",
"16300 tensor(4.9985, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16400 tensor(5.3076, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59000\n",
"16500 tensor(5.1994, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16600 tensor(5.3672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16700 tensor(5.2054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60000\n",
"16800 tensor(5.3379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16900 tensor(5.2785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17000 tensor(5.2590, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61000\n",
"17100 tensor(5.3564, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17200 tensor(5.3598, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17300 tensor(4.7786, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62000\n",
"17400 tensor(5.2639, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17500 tensor(5.2037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17600 tensor(5.1158, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63000\n",
"17700 tensor(4.9831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17800 tensor(4.8950, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64000\n",
"17900 tensor(5.0928, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18000 tensor(5.3423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18100 tensor(5.1760, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65000\n",
"18200 tensor(5.2021, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18300 tensor(5.1306, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18400 tensor(5.1199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66000\n",
"18500 tensor(5.2082, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18600 tensor(5.3290, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18700 tensor(5.2257, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67000\n",
"18800 tensor(4.9107, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18900 tensor(5.3400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68000\n",
"19000 tensor(5.1366, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19100 tensor(5.1199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19200 tensor(5.2202, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69000\n",
"19300 tensor(5.2236, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19400 tensor(5.2953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19500 tensor(5.1308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70000\n",
"19600 tensor(5.3578, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19700 tensor(5.1600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19800 tensor(4.6220, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71000\n",
"19900 tensor(5.3731, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20000 tensor(4.9936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20100 tensor(5.0817, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72000\n",
"20200 tensor(5.1613, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20300 tensor(5.3877, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73000\n",
"20400 tensor(5.4114, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20500 tensor(5.2609, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20600 tensor(5.1378, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74000\n",
"20700 tensor(5.0799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20800 tensor(5.3615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20900 tensor(5.3365, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75000\n",
"21000 tensor(4.9244, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21100 tensor(5.5084, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21200 tensor(4.8769, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76000\n",
"21300 tensor(5.3414, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21400 tensor(5.0648, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21500 tensor(5.0594, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77000\n",
"21600 tensor(5.2537, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21700 tensor(5.1834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21800 tensor(4.8151, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78000\n",
"21900 tensor(5.3335, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22000 tensor(4.9580, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79000\n",
"22100 tensor(5.2262, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22200 tensor(5.1946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22300 tensor(5.2404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80000\n",
"22400 tensor(4.9491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22500 tensor(4.6901, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22600 tensor(5.1937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81000\n",
"22700 tensor(4.9937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22800 tensor(5.1401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22900 tensor(5.0599, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82000\n",
"23000 tensor(5.4315, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23100 tensor(5.1854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83000\n",
"23200 tensor(5.1033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23300 tensor(5.2352, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23400 tensor(5.2004, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84000\n",
"23500 tensor(5.0866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23600 tensor(5.2372, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23700 tensor(5.4711, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85000\n",
"23800 tensor(5.4030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23900 tensor(5.3589, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24000 tensor(5.1646, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86000\n",
"24100 tensor(5.4865, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24200 tensor(5.3663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24300 tensor(5.1760, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87000\n",
"24400 tensor(5.2950, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24500 tensor(5.0376, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88000\n",
"24600 tensor(5.1229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24700 tensor(5.3261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24800 tensor(5.3953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89000\n",
"24900 tensor(5.2734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25000 tensor(5.5544, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25100 tensor(5.1847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90000\n",
"25200 tensor(5.2963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25300 tensor(5.2350, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25400 tensor(5.1483, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91000\n",
"25500 tensor(5.0571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25600 tensor(4.8022, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25700 tensor(5.2115, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92000\n",
"25800 tensor(5.1675, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25900 tensor(5.2786, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93000\n",
"26000 tensor(4.8893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26100 tensor(5.3295, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26200 tensor(4.9849, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94000\n",
"26300 tensor(5.2289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26400 tensor(5.1430, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26500 tensor(5.1389, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95000\n",
"26600 tensor(5.3339, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26700 tensor(5.1904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26800 tensor(5.0073, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96000\n",
"26900 tensor(5.1798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27000 tensor(4.9300, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27100 tensor(5.1331, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97000\n",
"27200 tensor(5.1750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27300 tensor(5.1801, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98000\n",
"27400 tensor(5.0430, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27500 tensor(5.3107, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27600 tensor(5.2876, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99000\n",
"27700 tensor(5.2763, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27800 tensor(5.0737, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27900 tensor(5.0593, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100000\n",
"28000 tensor(5.4758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28100 tensor(5.4440, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28200 tensor(5.0862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101000\n",
"28300 tensor(5.2405, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28400 tensor(4.8600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28500 tensor(5.2056, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102000\n",
"28600 tensor(5.6196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28700 tensor(5.2235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103000\n",
"28800 tensor(5.2124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28900 tensor(4.8836, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29000 tensor(4.9913, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104000\n",
"29100 tensor(5.2689, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29200 tensor(4.9352, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29300 tensor(5.3143, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105000\n",
"29400 tensor(5.2134, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29500 tensor(5.0963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29600 tensor(5.4291, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106000\n",
"29700 tensor(4.8093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29800 tensor(4.9271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29900 tensor(5.0479, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107000\n",
"30000 tensor(5.2494, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30100 tensor(5.2027, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108000\n",
"30200 tensor(5.2714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30300 tensor(4.9223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30400 tensor(4.8816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109000\n",
"30500 tensor(4.9169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30600 tensor(5.1237, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30700 tensor(5.1190, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110000\n",
"30800 tensor(5.2271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30900 tensor(5.0194, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31000 tensor(5.2287, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111000\n",
"31100 tensor(5.3456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31200 tensor(5.0911, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31300 tensor(5.3793, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112000\n",
"31400 tensor(5.0410, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31500 tensor(5.0014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113000\n",
"31600 tensor(4.9323, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31700 tensor(5.3869, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31800 tensor(4.8621, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114000\n",
"31900 tensor(5.1184, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32000 tensor(5.2873, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32100 tensor(4.8723, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115000\n",
"32200 tensor(5.4421, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32300 tensor(5.0732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32400 tensor(5.0461, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116000\n",
"32500 tensor(5.3476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32600 tensor(5.1149, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117000\n",
"32700 tensor(5.2261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32800 tensor(5.1622, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32900 tensor(5.0429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118000\n",
"33000 tensor(5.1329, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33100 tensor(5.2284, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33200 tensor(5.1794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119000\n",
"33300 tensor(4.8535, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33400 tensor(4.7420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33500 tensor(5.4625, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120000\n",
"33600 tensor(4.9975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33700 tensor(5.2172, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33800 tensor(5.1135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"121000\n",
"33900 tensor(5.1974, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34000 tensor(5.0866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"122000\n",
"34100 tensor(5.1738, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34200 tensor(5.2338, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34300 tensor(5.3866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"123000\n",
"34400 tensor(5.2937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34500 tensor(5.4265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34600 tensor(5.0077, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"124000\n",
"34700 tensor(5.3335, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34800 tensor(4.9879, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34900 tensor(5.3270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"125000\n",
"35000 tensor(5.2056, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35100 tensor(5.1341, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35200 tensor(5.2629, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"126000\n",
"35300 tensor(5.0193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35400 tensor(5.2414, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"127000\n",
"35500 tensor(5.3116, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35600 tensor(5.3802, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35700 tensor(5.5126, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"128000\n",
"35800 tensor(5.2014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35900 tensor(5.2182, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36000 tensor(5.2556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"129000\n",
"36100 tensor(5.2955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36200 tensor(5.2068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36300 tensor(5.4835, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"130000\n",
"36400 tensor(5.2844, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36500 tensor(5.4568, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"131000\n",
"36600 tensor(4.9546, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36700 tensor(5.5436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36800 tensor(5.4078, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"132000\n",
"36900 tensor(4.9905, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37000 tensor(5.2532, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37100 tensor(4.9246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"133000\n",
"37200 tensor(4.8595, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37300 tensor(4.9436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37400 tensor(4.9957, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"134000\n",
"37500 tensor(4.8686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37600 tensor(4.8965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37700 tensor(5.1955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"135000\n",
"37800 tensor(5.2660, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37900 tensor(5.3396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"136000\n",
"38000 tensor(5.0582, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38100 tensor(5.2351, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38200 tensor(4.9177, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"137000\n",
"38300 tensor(5.3007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38400 tensor(4.9814, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38500 tensor(5.2672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"138000\n",
"38600 tensor(4.8161, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38700 tensor(5.1975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38800 tensor(4.9423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"139000\n",
"38900 tensor(5.0342, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39000 tensor(5.2703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39100 tensor(4.9539, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"140000\n",
"39200 tensor(5.4193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39300 tensor(5.1109, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"141000\n",
"39400 tensor(5.0914, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39500 tensor(5.3126, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39600 tensor(4.9927, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"142000\n",
"39700 tensor(5.3725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39800 tensor(5.2926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39900 tensor(4.8999, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"143000\n",
"40000 tensor(5.0333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40100 tensor(5.1059, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40200 tensor(5.1685, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"144000\n",
"40300 tensor(5.1375, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40400 tensor(5.3952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40500 tensor(4.8467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"145000\n",
"40600 tensor(5.4346, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40700 tensor(5.2852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"146000\n",
"40800 tensor(5.0129, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40900 tensor(5.0657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41000 tensor(5.1874, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"147000\n",
"41100 tensor(5.0094, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41200 tensor(5.2361, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41300 tensor(4.8607, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"148000\n",
"41400 tensor(4.9584, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41500 tensor(5.3754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41600 tensor(4.9372, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"149000\n",
"41700 tensor(5.2398, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41800 tensor(5.1772, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41900 tensor(5.1525, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"150000\n",
"42000 tensor(5.2827, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42100 tensor(4.6855, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"151000\n",
"42200 tensor(5.0215, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42300 tensor(5.1695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42400 tensor(5.0091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"152000\n",
"42500 tensor(5.1274, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42600 tensor(4.6771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42700 tensor(5.3160, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"153000\n",
"42800 tensor(5.2484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42900 tensor(5.0599, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43000 tensor(5.2023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"154000\n",
"43100 tensor(4.9059, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43200 tensor(5.0727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43300 tensor(5.1986, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"155000\n",
"43400 tensor(5.1041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43500 tensor(5.4856, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"156000\n",
"43600 tensor(5.3947, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43700 tensor(5.0493, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43800 tensor(5.1703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"157000\n",
"43900 tensor(4.9899, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44000 tensor(4.8695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44100 tensor(5.3520, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"158000\n",
"44200 tensor(4.9379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44300 tensor(5.4677, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44400 tensor(5.0739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"159000\n",
"44500 tensor(4.9127, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44600 tensor(5.0247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"160000\n",
"44700 tensor(4.9641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44800 tensor(5.2538, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44900 tensor(5.0657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"161000\n",
"45000 tensor(4.9174, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45100 tensor(5.2311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45200 tensor(5.0617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"162000\n",
"45300 tensor(5.2111, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45400 tensor(4.6137, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45500 tensor(4.9275, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"163000\n",
"45600 tensor(5.2501, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45700 tensor(5.2526, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45800 tensor(5.1342, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"164000\n",
"45900 tensor(4.9400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46000 tensor(5.1521, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"165000\n",
"46100 tensor(5.0033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46200 tensor(4.9825, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46300 tensor(5.3210, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"166000\n",
"46400 tensor(5.3635, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46500 tensor(4.9169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46600 tensor(4.9497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"167000\n",
"46700 tensor(5.2150, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46800 tensor(4.8570, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46900 tensor(4.9026, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"168000\n",
"47000 tensor(5.1486, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47100 tensor(5.4026, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47200 tensor(5.1028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"169000\n",
"47300 tensor(5.2411, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47400 tensor(4.9942, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"170000\n",
"47500 tensor(5.1984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47600 tensor(5.1100, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47700 tensor(5.1202, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"171000\n",
"47800 tensor(5.2505, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47900 tensor(5.2392, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48000 tensor(4.9209, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"172000\n",
"48100 tensor(5.3051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48200 tensor(5.1233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48300 tensor(5.2021, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"173000\n",
"48400 tensor(4.8006, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48500 tensor(5.0496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48600 tensor(5.1875, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"174000\n",
"48700 tensor(5.1499, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48800 tensor(4.9676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"175000\n",
"48900 tensor(5.2299, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49000 tensor(5.0078, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49100 tensor(5.1948, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"176000\n",
"49200 tensor(5.2544, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49300 tensor(5.0308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49400 tensor(5.2188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"177000\n",
"49500 tensor(5.3279, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49600 tensor(4.9758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49700 tensor(5.0799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"178000\n",
"49800 tensor(5.0324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49900 tensor(5.2124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50000 tensor(5.1213, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"179000\n",
"50100 tensor(5.2672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50200 tensor(4.9766, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"180000\n",
"50300 tensor(5.2499, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50400 tensor(4.9455, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50500 tensor(5.2477, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"181000\n",
"50600 tensor(5.0518, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50700 tensor(4.9514, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50800 tensor(4.8579, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"182000\n",
"50900 tensor(5.1602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51000 tensor(5.2193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51100 tensor(5.0982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"183000\n",
"51200 tensor(5.2880, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51300 tensor(4.8375, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51400 tensor(5.1995, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"184000\n",
"51500 tensor(5.1672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51600 tensor(5.2276, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"185000\n",
"51700 tensor(5.0882, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51800 tensor(5.2304, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51900 tensor(5.0226, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"186000\n",
"52000 tensor(5.1911, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52100 tensor(5.3534, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52200 tensor(5.1108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"187000\n",
"52300 tensor(5.0774, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52400 tensor(5.1188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52500 tensor(5.1506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"188000\n",
"52600 tensor(5.0832, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52700 tensor(5.2014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"189000\n",
"52800 tensor(5.2696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52900 tensor(5.0779, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53000 tensor(5.0554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"190000\n",
"53100 tensor(5.0068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53200 tensor(4.7833, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53300 tensor(5.2709, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"191000\n",
"53400 tensor(5.1926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53500 tensor(5.0873, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53600 tensor(5.1936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"192000\n",
"53700 tensor(5.1878, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53800 tensor(4.8307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53900 tensor(4.8449, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"193000\n",
"54000 tensor(5.3690, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54100 tensor(4.4637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"194000\n",
"54200 tensor(5.2497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54300 tensor(5.0490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54400 tensor(4.7058, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"195000\n",
"54500 tensor(5.2924, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54600 tensor(5.3235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54700 tensor(5.0440, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"196000\n",
"54800 tensor(5.2241, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54900 tensor(4.9492, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55000 tensor(5.0754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"197000\n",
"55100 tensor(5.1074, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55200 tensor(5.2155, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55300 tensor(5.1454, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"198000\n",
"55400 tensor(5.2931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55500 tensor(5.3043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"199000\n",
"55600 tensor(4.8684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55700 tensor(5.0869, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55800 tensor(4.6445, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"200000\n",
"55900 tensor(5.1047, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56000 tensor(4.9902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56100 tensor(5.2816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"201000\n",
"56200 tensor(5.3301, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56300 tensor(5.1437, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56400 tensor(5.2254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"202000\n",
"56500 tensor(5.2339, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56600 tensor(4.9340, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56700 tensor(5.0758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"203000\n",
"56800 tensor(5.5707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56900 tensor(4.9705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"204000\n",
"57000 tensor(5.3555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57100 tensor(5.1206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57200 tensor(4.9509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"205000\n",
"57300 tensor(5.0644, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57400 tensor(5.2678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57500 tensor(4.9110, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"206000\n",
"57600 tensor(4.7788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57700 tensor(4.9975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57800 tensor(5.3311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"207000\n",
"57900 tensor(5.2600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58000 tensor(5.0070, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"208000\n",
"58100 tensor(5.3885, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58200 tensor(5.4209, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58300 tensor(4.9688, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"209000\n",
"58400 tensor(4.8874, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58500 tensor(4.9976, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58600 tensor(5.4753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"210000\n",
"58700 tensor(4.9572, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58800 tensor(5.1014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58900 tensor(5.2920, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"211000\n",
"59000 tensor(5.2007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59100 tensor(5.0399, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59200 tensor(5.1078, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"212000\n",
"59300 tensor(4.9478, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59400 tensor(5.1439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"213000\n",
"59500 tensor(5.0825, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59600 tensor(5.0066, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59700 tensor(4.8766, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"214000\n",
"59800 tensor(5.0725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59900 tensor(5.1101, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60000 tensor(5.1072, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"215000\n",
"60100 tensor(5.1821, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60200 tensor(5.2384, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60300 tensor(5.0115, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"216000\n",
"60400 tensor(5.1308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60500 tensor(5.1859, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60600 tensor(5.0918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"217000\n",
"60700 tensor(5.0108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60800 tensor(5.1547, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"218000\n",
"60900 tensor(5.2653, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61000 tensor(5.2235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61100 tensor(5.1075, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"219000\n",
"61200 tensor(4.8467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61300 tensor(5.0838, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61400 tensor(5.2590, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"220000\n",
"61500 tensor(5.1743, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61600 tensor(4.9954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61700 tensor(4.9919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"221000\n",
"61800 tensor(5.0254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61900 tensor(5.0908, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62000 tensor(5.2753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"222000\n",
"62100 tensor(4.8669, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62200 tensor(4.9641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"223000\n",
"62300 tensor(5.0582, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62400 tensor(4.9853, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62500 tensor(5.0922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"224000\n",
"62600 tensor(4.9824, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62700 tensor(5.1782, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62800 tensor(4.8954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"225000\n",
"62900 tensor(5.3762, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63000 tensor(5.2436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63100 tensor(5.3535, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"226000\n",
"63200 tensor(5.1772, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63300 tensor(4.9112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63400 tensor(5.0962, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"227000\n",
"63500 tensor(4.9657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63600 tensor(5.0014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"228000\n",
"63700 tensor(5.0686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63800 tensor(5.3235, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63900 tensor(5.3906, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"229000\n",
"64000 tensor(4.9716, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64100 tensor(5.2290, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64200 tensor(5.1423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"230000\n",
"64300 tensor(5.1310, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64400 tensor(4.9695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64500 tensor(4.9604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"231000\n",
"64600 tensor(4.9843, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64700 tensor(5.4459, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64800 tensor(5.1474, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"232000\n",
"64900 tensor(5.2715, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65000 tensor(5.2314, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"233000\n",
"65100 tensor(4.7774, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65200 tensor(5.3146, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65300 tensor(5.1604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"234000\n",
"65400 tensor(5.0736, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65500 tensor(5.1152, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65600 tensor(5.1409, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"235000\n",
"65700 tensor(4.9068, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65800 tensor(4.9353, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65900 tensor(5.1612, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"236000\n",
"66000 tensor(5.1866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66100 tensor(4.8965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"237000\n",
"66200 tensor(4.9215, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66300 tensor(4.8641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66400 tensor(5.1550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"238000\n",
"66500 tensor(4.8722, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66600 tensor(5.0268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66700 tensor(4.9948, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"239000\n",
"66800 tensor(5.0565, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66900 tensor(5.2030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67000 tensor(5.2199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"240000\n",
"67100 tensor(5.0560, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67200 tensor(5.4027, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67300 tensor(5.0631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"241000\n",
"67400 tensor(5.2687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67500 tensor(5.0728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"242000\n",
"67600 tensor(5.1050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67700 tensor(5.0495, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67800 tensor(5.2014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"243000\n",
"67900 tensor(4.9060, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68000 tensor(4.9703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68100 tensor(4.8457, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"244000\n",
"68200 tensor(5.1298, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68300 tensor(5.0272, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68400 tensor(5.1238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"245000\n",
"68500 tensor(5.1778, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68600 tensor(5.0997, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68700 tensor(4.7577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"246000\n",
"68800 tensor(5.1663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68900 tensor(4.8385, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"247000\n",
"69000 tensor(5.0810, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69100 tensor(4.8565, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69200 tensor(4.8088, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"248000\n",
"69300 tensor(5.2732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69400 tensor(5.1549, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69500 tensor(5.1323, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"249000\n",
"69600 tensor(5.0379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69700 tensor(5.3768, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69800 tensor(4.9056, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"250000\n",
"69900 tensor(5.4631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70000 tensor(5.1071, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70100 tensor(5.0137, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"251000\n",
"70200 tensor(5.1769, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70300 tensor(4.8435, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"252000\n",
"70400 tensor(5.2020, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70500 tensor(5.1395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70600 tensor(5.3541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"253000\n",
"70700 tensor(5.3092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70800 tensor(5.4793, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70900 tensor(5.2799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"254000\n",
"71000 tensor(5.0872, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71100 tensor(5.1728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71200 tensor(5.2739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"255000\n",
"71300 tensor(4.8960, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71400 tensor(5.0185, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71500 tensor(5.0367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"256000\n",
"71600 tensor(5.1992, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71700 tensor(5.2508, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"257000\n",
"71800 tensor(5.3864, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71900 tensor(5.0381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72000 tensor(5.4758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"258000\n",
"72100 tensor(5.1863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72200 tensor(5.0258, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72300 tensor(5.1253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"259000\n",
"72400 tensor(5.0412, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72500 tensor(5.1304, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72600 tensor(5.1833, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"260000\n",
"72700 tensor(4.8666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72800 tensor(5.3088, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72900 tensor(5.0482, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"261000\n",
"73000 tensor(5.2591, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73100 tensor(4.9459, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"262000\n",
"73200 tensor(5.0086, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73300 tensor(4.9859, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73400 tensor(5.2442, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"263000\n",
"73500 tensor(5.0674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73600 tensor(5.0003, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73700 tensor(5.1476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"264000\n",
"73800 tensor(5.1845, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73900 tensor(5.1606, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74000 tensor(5.3307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"265000\n",
"74100 tensor(5.0139, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74200 tensor(5.0479, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74300 tensor(5.0041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"266000\n",
"74400 tensor(5.1568, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74500 tensor(4.7452, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"267000\n",
"74600 tensor(5.2662, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74700 tensor(5.3145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74800 tensor(5.0200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"268000\n",
"74900 tensor(4.8437, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75000 tensor(5.1458, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75100 tensor(5.3639, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"269000\n",
"75200 tensor(5.0922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75300 tensor(5.3002, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75400 tensor(5.0810, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"270000\n",
"75500 tensor(5.3591, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75600 tensor(5.0206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"271000\n",
"75700 tensor(5.0759, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75800 tensor(5.0702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75900 tensor(5.1349, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"272000\n",
"76000 tensor(4.8964, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76100 tensor(4.8618, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76200 tensor(4.9370, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"273000\n",
"76300 tensor(5.0666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76400 tensor(4.9261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76500 tensor(5.0910, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"274000\n",
"76600 tensor(5.1985, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76700 tensor(4.8590, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76800 tensor(4.7918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"275000\n",
"76900 tensor(5.1386, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77000 tensor(5.0233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"276000\n",
"77100 tensor(5.3827, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77200 tensor(5.0204, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77300 tensor(4.9805, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"277000\n",
"77400 tensor(5.0943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77500 tensor(4.7828, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77600 tensor(5.1103, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"278000\n",
"77700 tensor(5.2850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77800 tensor(4.8699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77900 tensor(4.7829, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"279000\n",
"78000 tensor(5.2236, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78100 tensor(4.8891, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78200 tensor(5.1773, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"280000\n",
"78300 tensor(5.0470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78400 tensor(5.0570, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"281000\n",
"78500 tensor(5.4172, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78600 tensor(4.8807, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78700 tensor(5.3329, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"282000\n",
"78800 tensor(5.2173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78900 tensor(5.4813, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79000 tensor(4.9973, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"283000\n",
"79100 tensor(4.7734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79200 tensor(4.8357, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79300 tensor(5.3520, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"284000\n",
"79400 tensor(5.1467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79500 tensor(5.0294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79600 tensor(5.6394, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"285000\n",
"79700 tensor(4.7860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79800 tensor(5.0916, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"286000\n",
"79900 tensor(5.0640, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80000 tensor(5.2131, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80100 tensor(5.2826, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"287000\n",
"80200 tensor(5.3061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80300 tensor(5.2656, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80400 tensor(5.3741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"288000\n",
"80500 tensor(5.0930, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80600 tensor(5.2616, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80700 tensor(5.3157, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"289000\n",
"80800 tensor(4.8360, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80900 tensor(5.2270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"290000\n",
"81000 tensor(5.1687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81100 tensor(5.1558, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81200 tensor(4.9090, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"291000\n",
"81300 tensor(4.9953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81400 tensor(5.1353, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81500 tensor(5.1045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"292000\n",
"81600 tensor(5.1020, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81700 tensor(5.2123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81800 tensor(5.0393, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"293000\n",
"81900 tensor(5.2712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82000 tensor(5.0969, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82100 tensor(5.1216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"294000\n",
"82200 tensor(5.0557, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82300 tensor(4.9278, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"295000\n",
"82400 tensor(5.1703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82500 tensor(4.8852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82600 tensor(5.0861, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"296000\n",
"82700 tensor(5.1290, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82800 tensor(5.1089, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82900 tensor(4.7936, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"297000\n",
"83000 tensor(5.1975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83100 tensor(5.0460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83200 tensor(5.2952, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"298000\n",
"83300 tensor(4.9863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83400 tensor(5.0724, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83500 tensor(5.0200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"299000\n",
"83600 tensor(4.9496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83700 tensor(5.1706, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"300000\n",
"83800 tensor(4.8822, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83900 tensor(4.9673, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84000 tensor(5.1188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"301000\n",
"84100 tensor(5.4809, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84200 tensor(4.6303, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84300 tensor(5.0608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"302000\n",
"84400 tensor(4.8835, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84500 tensor(5.0710, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84600 tensor(5.2347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"303000\n",
"84700 tensor(4.9578, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84800 tensor(4.9840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84900 tensor(5.2946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"304000\n",
"85000 tensor(5.1757, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85100 tensor(5.0449, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"305000\n",
"85200 tensor(5.0524, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85300 tensor(5.3156, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85400 tensor(5.2982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"306000\n",
"85500 tensor(4.9904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85600 tensor(5.1111, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85700 tensor(5.1132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"307000\n",
"85800 tensor(5.1134, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85900 tensor(5.3567, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86000 tensor(5.1374, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"308000\n",
"86100 tensor(4.8926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86200 tensor(5.0359, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86300 tensor(5.0061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"309000\n",
"86400 tensor(5.0968, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86500 tensor(5.1935, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"310000\n",
"86600 tensor(5.0132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86700 tensor(4.9140, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86800 tensor(4.9166, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"311000\n",
"86900 tensor(4.8179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87000 tensor(5.3136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87100 tensor(5.1485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"312000\n",
"87200 tensor(5.1635, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87300 tensor(5.0104, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87400 tensor(5.0568, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"313000\n",
"87500 tensor(5.1699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87600 tensor(5.2978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87700 tensor(4.9475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"314000\n",
"87800 tensor(4.8997, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87900 tensor(5.1651, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"315000\n",
"88000 tensor(4.6506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88100 tensor(5.0203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88200 tensor(5.2400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"316000\n",
"88300 tensor(5.1478, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88400 tensor(5.2234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88500 tensor(4.8418, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"317000\n",
"88600 tensor(4.8617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88700 tensor(4.8720, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88800 tensor(4.8572, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"318000\n",
"88900 tensor(4.7781, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89000 tensor(4.5953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89100 tensor(5.1780, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"319000\n",
"89200 tensor(4.9773, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89300 tensor(5.3703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"320000\n",
"89400 tensor(4.7298, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89500 tensor(5.0713, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89600 tensor(5.3035, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"321000\n",
"89700 tensor(4.8603, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89800 tensor(5.0780, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89900 tensor(5.0922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"322000\n",
"90000 tensor(5.0946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90100 tensor(4.9839, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90200 tensor(5.0898, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"323000\n",
"90300 tensor(5.0876, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90400 tensor(4.7618, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90500 tensor(4.9550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"324000\n",
"90600 tensor(5.2409, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90700 tensor(4.8453, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"325000\n",
"90800 tensor(5.0906, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90900 tensor(5.1866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91000 tensor(5.1850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"326000\n",
"91100 tensor(4.7448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91200 tensor(5.2125, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91300 tensor(5.3846, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"327000\n",
"91400 tensor(4.7534, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91500 tensor(4.9951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91600 tensor(5.2443, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"328000\n",
"91700 tensor(5.0912, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91800 tensor(5.2768, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"329000\n",
"91900 tensor(5.0739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92000 tensor(5.1095, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92100 tensor(4.9860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"330000\n",
"92200 tensor(4.6185, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92300 tensor(5.1181, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92400 tensor(5.1625, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"331000\n",
"92500 tensor(5.2200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92600 tensor(5.0955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92700 tensor(4.8214, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"332000\n",
"92800 tensor(4.8953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92900 tensor(5.4247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93000 tensor(5.1056, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"333000\n",
"93100 tensor(5.0311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93200 tensor(5.2195, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"334000\n",
"93300 tensor(5.1673, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93400 tensor(5.2275, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93500 tensor(4.9526, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"335000\n",
"93600 tensor(5.0896, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93700 tensor(5.1565, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93800 tensor(4.9091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"336000\n",
"93900 tensor(5.1251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94000 tensor(5.0433, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94100 tensor(4.9108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"337000\n",
"94200 tensor(5.1725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94300 tensor(5.0600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94400 tensor(5.0994, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"338000\n",
"94500 tensor(4.9514, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94600 tensor(5.2426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"339000\n",
"94700 tensor(4.7794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94800 tensor(5.2669, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94900 tensor(5.2703, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"340000\n",
"95000 tensor(5.3905, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95100 tensor(5.0774, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95200 tensor(4.9328, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"341000\n",
"95300 tensor(5.2937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95400 tensor(4.7542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95500 tensor(4.9919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"342000\n",
"95600 tensor(4.9873, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95700 tensor(5.1582, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95800 tensor(5.0927, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"343000\n",
"95900 tensor(4.8471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96000 tensor(5.1951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"344000\n",
"96100 tensor(4.9727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96200 tensor(4.9915, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96300 tensor(5.0515, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"345000\n",
"96400 tensor(5.1081, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96500 tensor(5.3060, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96600 tensor(5.0907, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"346000\n",
"96700 tensor(5.2215, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96800 tensor(5.1097, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96900 tensor(5.2757, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"347000\n",
"97000 tensor(4.9539, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97100 tensor(5.0553, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97200 tensor(5.1827, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"348000\n",
"97300 tensor(5.0123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97400 tensor(5.0624, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"349000\n",
"97500 tensor(5.2006, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97600 tensor(4.9950, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97700 tensor(5.0549, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"350000\n",
"97800 tensor(5.0475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97900 tensor(4.8108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98000 tensor(4.9221, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"351000\n",
"98100 tensor(5.2524, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98200 tensor(4.9745, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98300 tensor(5.1369, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"352000\n",
"98400 tensor(5.2040, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98500 tensor(5.2696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"353000\n",
"98600 tensor(5.3089, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98700 tensor(5.0973, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98800 tensor(5.0995, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"354000\n",
"98900 tensor(5.1192, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99000 tensor(5.2984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99100 tensor(5.1856, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"355000\n",
"99200 tensor(5.1883, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99300 tensor(4.9773, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99400 tensor(5.2897, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"356000\n",
"99500 tensor(5.3124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99600 tensor(5.0386, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99700 tensor(4.8906, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"357000\n",
"99800 tensor(5.1151, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99900 tensor(5.0333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100000 tensor(4.9526, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"358000\n",
"100100 tensor(5.4919, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100200 tensor(4.7812, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"359000\n",
"100300 tensor(5.0547, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100400 tensor(5.0233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100500 tensor(4.8942, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"360000\n",
"100600 tensor(4.9477, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100700 tensor(4.9277, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100800 tensor(5.2499, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"361000\n",
"100900 tensor(5.2162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101000 tensor(5.0550, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101100 tensor(5.0957, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"362000\n",
"101200 tensor(5.0617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101300 tensor(5.3101, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"363000\n",
"101400 tensor(4.8387, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101500 tensor(5.0914, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101600 tensor(5.3198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"364000\n",
"101700 tensor(4.9333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101800 tensor(4.7367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101900 tensor(4.9608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"365000\n",
"102000 tensor(5.2333, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102100 tensor(5.1444, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102200 tensor(5.4328, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"366000\n",
"102300 tensor(5.2760, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102400 tensor(5.2229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102500 tensor(5.2043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"367000\n",
"102600 tensor(5.1444, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102700 tensor(5.2240, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"368000\n",
"102800 tensor(5.1234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102900 tensor(4.8963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103000 tensor(5.0684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"369000\n",
"103100 tensor(5.3707, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103200 tensor(4.9859, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103300 tensor(5.0649, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"370000\n",
"103400 tensor(5.0871, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103500 tensor(4.9837, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103600 tensor(5.1120, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"371000\n",
"103700 tensor(5.2086, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103800 tensor(5.0741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103900 tensor(4.9891, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"372000\n",
"104000 tensor(4.9637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104100 tensor(5.0033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"373000\n",
"104200 tensor(5.1105, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104300 tensor(4.9057, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104400 tensor(5.2394, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"374000\n",
"104500 tensor(5.1488, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104600 tensor(5.0034, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104700 tensor(5.0169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"375000\n",
"104800 tensor(5.5932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104900 tensor(5.0594, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105000 tensor(5.1818, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"376000\n",
"105100 tensor(5.1265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105200 tensor(5.2194, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105300 tensor(5.1688, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"377000\n",
"105400 tensor(5.3924, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105500 tensor(5.2369, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"378000\n",
"105600 tensor(5.0005, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105700 tensor(4.9490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105800 tensor(4.9947, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"379000\n",
"105900 tensor(5.1400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106000 tensor(5.1712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106100 tensor(5.3889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"380000\n",
"106200 tensor(5.0640, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106300 tensor(5.1624, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106400 tensor(4.9938, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"381000\n",
"106500 tensor(5.2659, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106600 tensor(4.9500, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106700 tensor(5.2001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"382000\n",
"106800 tensor(4.9776, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106900 tensor(5.2684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"383000\n",
"107000 tensor(4.9327, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107100 tensor(5.1739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107200 tensor(5.1641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"384000\n",
"107300 tensor(4.9207, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107400 tensor(5.1112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107500 tensor(4.8839, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"385000\n",
"107600 tensor(4.9063, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107700 tensor(5.1218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107800 tensor(5.1329, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"386000\n",
"107900 tensor(5.3598, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108000 tensor(5.1047, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108100 tensor(5.0709, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"387000\n",
"108200 tensor(5.0716, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108300 tensor(5.0979, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"388000\n",
"108400 tensor(5.2054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108500 tensor(4.8856, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108600 tensor(5.0763, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"389000\n",
"108700 tensor(4.6029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108800 tensor(4.9980, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108900 tensor(4.7433, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"390000\n",
"109000 tensor(5.1747, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109100 tensor(4.7590, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109200 tensor(4.9555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"391000\n",
"109300 tensor(4.6271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109400 tensor(4.9270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109500 tensor(4.8670, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"392000\n",
"109600 tensor(5.0673, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109700 tensor(5.4325, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"393000\n",
"109800 tensor(4.7563, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109900 tensor(4.9688, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110000 tensor(5.0599, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"394000\n",
"110100 tensor(4.9771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110200 tensor(5.3440, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110300 tensor(5.2067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"395000\n",
"110400 tensor(4.6912, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110500 tensor(5.1135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110600 tensor(4.9512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"396000\n",
"110700 tensor(4.9049, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110800 tensor(5.3380, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110900 tensor(4.9982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"397000\n",
"111000 tensor(5.1906, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111100 tensor(4.9450, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"398000\n",
"111200 tensor(4.9843, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111300 tensor(5.0615, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111400 tensor(5.1630, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"399000\n",
"111500 tensor(5.2631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111600 tensor(5.2726, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111700 tensor(4.8465, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"400000\n",
"111800 tensor(5.0638, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111900 tensor(5.1322, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112000 tensor(5.0956, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"401000\n",
"112100 tensor(4.6850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112200 tensor(5.0941, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"402000\n",
"112300 tensor(5.2451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112400 tensor(5.2617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112500 tensor(4.8904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"403000\n",
"112600 tensor(4.9779, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112700 tensor(4.9773, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112800 tensor(4.6058, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"404000\n",
"112900 tensor(4.8841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113000 tensor(5.1203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113100 tensor(5.0650, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"405000\n",
"113200 tensor(4.6302, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113300 tensor(4.9626, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113400 tensor(5.0637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"406000\n",
"113500 tensor(5.3193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113600 tensor(5.2732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"407000\n",
"113700 tensor(4.8750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113800 tensor(5.2875, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113900 tensor(5.0677, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"408000\n",
"114000 tensor(4.7401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114100 tensor(4.8493, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114200 tensor(4.8203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"409000\n",
"114300 tensor(4.6964, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114400 tensor(4.9045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114500 tensor(4.9766, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"410000\n",
"114600 tensor(5.2010, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114700 tensor(5.1521, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114800 tensor(5.0945, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"411000\n",
"114900 tensor(5.0311, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115000 tensor(5.0057, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"412000\n",
"115100 tensor(5.1129, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115200 tensor(5.0684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115300 tensor(5.1183, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"413000\n",
"115400 tensor(5.3141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115500 tensor(5.3232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115600 tensor(5.1170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"414000\n",
"115700 tensor(5.0023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115800 tensor(5.1229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115900 tensor(5.1923, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"415000\n",
"116000 tensor(5.1286, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116100 tensor(4.8629, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116200 tensor(5.2643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"416000\n",
"116300 tensor(4.4037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116400 tensor(5.1548, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"417000\n",
"116500 tensor(5.0993, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116600 tensor(5.1907, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116700 tensor(5.0096, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"418000\n",
"116800 tensor(4.8711, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116900 tensor(5.1060, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117000 tensor(5.0866, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"419000\n",
"117100 tensor(5.1294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117200 tensor(4.9697, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117300 tensor(5.3102, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"420000\n",
"117400 tensor(5.3132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117500 tensor(4.8965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"421000\n",
"117600 tensor(5.2093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117700 tensor(5.1284, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117800 tensor(4.9976, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"422000\n",
"117900 tensor(5.1066, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118000 tensor(5.1549, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118100 tensor(4.9796, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"423000\n",
"118200 tensor(5.0112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118300 tensor(5.0054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118400 tensor(5.1148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"424000\n",
"118500 tensor(5.0216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118600 tensor(5.1426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118700 tensor(5.1632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"425000\n",
"118800 tensor(5.1688, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118900 tensor(5.2083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"426000\n",
"119000 tensor(4.7053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119100 tensor(4.9990, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119200 tensor(5.3487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"427000\n",
"119300 tensor(4.9779, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119400 tensor(5.0351, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119500 tensor(5.0373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"428000\n",
"119600 tensor(5.3674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119700 tensor(4.9741, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119800 tensor(5.0577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"429000\n",
"119900 tensor(4.8425, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120000 tensor(4.9597, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120100 tensor(4.9721, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"430000\n",
"120200 tensor(5.1214, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120300 tensor(5.0225, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"431000\n",
"120400 tensor(5.3908, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120500 tensor(4.9955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120600 tensor(5.2588, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"432000\n",
"epoch: = 1\n",
"0 tensor(5.2093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100 tensor(5.3814, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"200 tensor(5.1460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1000\n",
"300 tensor(5.3199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"400 tensor(5.1499, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"500 tensor(5.0303, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2000\n",
"600 tensor(5.0671, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"700 tensor(5.0000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"800 tensor(4.9863, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3000\n",
"900 tensor(5.0532, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1000 tensor(5.0634, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1100 tensor(4.8666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4000\n",
"1200 tensor(4.8362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1300 tensor(5.1104, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1400 tensor(5.2784, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5000\n",
"1500 tensor(5.3135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1600 tensor(5.1400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6000\n",
"1700 tensor(5.2427, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1800 tensor(5.2361, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"1900 tensor(5.0860, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7000\n",
"2000 tensor(4.7754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2100 tensor(5.2257, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2200 tensor(4.9228, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8000\n",
"2300 tensor(5.2087, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2400 tensor(4.6009, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2500 tensor(5.0485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9000\n",
"2600 tensor(4.8077, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2700 tensor(5.1105, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"2800 tensor(4.6892, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10000\n",
"2900 tensor(4.9577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3000 tensor(5.0162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11000\n",
"3100 tensor(4.9243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3200 tensor(5.0965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3300 tensor(4.8072, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12000\n",
"3400 tensor(5.0026, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3500 tensor(4.8778, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3600 tensor(4.7804, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13000\n",
"3700 tensor(5.0536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3800 tensor(5.0348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"3900 tensor(5.0932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14000\n",
"4000 tensor(4.8247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4100 tensor(5.2288, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4200 tensor(4.7091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15000\n",
"4300 tensor(5.1466, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4400 tensor(4.8608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16000\n",
"4500 tensor(5.0831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4600 tensor(5.1534, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4700 tensor(5.1433, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17000\n",
"4800 tensor(5.0023, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"4900 tensor(5.1541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5000 tensor(5.2215, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18000\n",
"5100 tensor(5.0103, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5200 tensor(5.2232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5300 tensor(5.0542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19000\n",
"5400 tensor(4.9754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5500 tensor(5.1588, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20000\n",
"5600 tensor(4.9750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5700 tensor(5.0814, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"5800 tensor(5.4650, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21000\n",
"5900 tensor(5.2366, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6000 tensor(5.2223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6100 tensor(4.9001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22000\n",
"6200 tensor(5.3510, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6300 tensor(5.2366, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6400 tensor(5.0379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23000\n",
"6500 tensor(5.0265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6600 tensor(5.5000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6700 tensor(4.8587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24000\n",
"6800 tensor(4.8712, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"6900 tensor(5.1405, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25000\n",
"7000 tensor(5.0429, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7100 tensor(5.1420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7200 tensor(5.2794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26000\n",
"7300 tensor(5.2271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7400 tensor(4.9754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7500 tensor(5.3659, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27000\n",
"7600 tensor(4.7902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7700 tensor(5.2327, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"7800 tensor(5.5528, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28000\n",
"7900 tensor(5.0152, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8000 tensor(5.1026, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8100 tensor(4.6152, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29000\n",
"8200 tensor(5.0953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8300 tensor(5.0372, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30000\n",
"8400 tensor(5.0211, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8500 tensor(4.7475, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8600 tensor(4.8443, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31000\n",
"8700 tensor(5.1471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8800 tensor(5.1762, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"8900 tensor(5.4800, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32000\n",
"9000 tensor(4.9010, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9100 tensor(5.1490, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9200 tensor(5.2119, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33000\n",
"9300 tensor(5.0652, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9400 tensor(5.0831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34000\n",
"9500 tensor(4.9114, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9600 tensor(4.9139, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9700 tensor(5.0307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35000\n",
"9800 tensor(5.1592, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"9900 tensor(4.9091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10000 tensor(5.2556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36000\n",
"10100 tensor(5.2265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10200 tensor(5.0764, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10300 tensor(5.2197, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37000\n",
"10400 tensor(4.9872, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10500 tensor(5.0030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10600 tensor(4.7093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38000\n",
"10700 tensor(5.1943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"10800 tensor(4.6028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39000\n",
"10900 tensor(5.1688, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11000 tensor(4.9071, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11100 tensor(4.7852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40000\n",
"11200 tensor(5.1724, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11300 tensor(5.1048, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11400 tensor(5.3444, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41000\n",
"11500 tensor(5.2196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11600 tensor(4.8939, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11700 tensor(5.2069, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42000\n",
"11800 tensor(5.3687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"11900 tensor(4.9033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12000 tensor(5.1747, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43000\n",
"12100 tensor(5.0138, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12200 tensor(5.0538, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44000\n",
"12300 tensor(5.2248, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12400 tensor(4.8343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12500 tensor(5.0820, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45000\n",
"12600 tensor(5.0378, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12700 tensor(4.7855, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"12800 tensor(5.1147, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46000\n",
"12900 tensor(5.2598, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13000 tensor(4.8743, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13100 tensor(5.1137, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47000\n",
"13200 tensor(5.2270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13300 tensor(5.3398, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13400 tensor(4.7988, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48000\n",
"13500 tensor(5.1067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13600 tensor(5.0730, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49000\n",
"13700 tensor(5.1172, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13800 tensor(4.9974, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"13900 tensor(5.0809, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50000\n",
"14000 tensor(5.0515, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14100 tensor(5.2082, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14200 tensor(5.2196, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51000\n",
"14300 tensor(5.1963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14400 tensor(4.6420, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14500 tensor(4.7768, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52000\n",
"14600 tensor(4.8853, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14700 tensor(5.2161, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"14800 tensor(4.8348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53000\n",
"14900 tensor(4.7723, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15000 tensor(5.1951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54000\n",
"15100 tensor(4.5943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15200 tensor(4.8785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15300 tensor(5.2401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55000\n",
"15400 tensor(5.1345, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15500 tensor(4.9845, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15600 tensor(5.0955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56000\n",
"15700 tensor(5.0750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15800 tensor(4.8953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"15900 tensor(5.0173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57000\n",
"16000 tensor(5.3809, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16100 tensor(5.2108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16200 tensor(4.9123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58000\n",
"16300 tensor(4.8491, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16400 tensor(5.1661, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59000\n",
"16500 tensor(5.0995, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16600 tensor(5.1714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16700 tensor(5.0571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60000\n",
"16800 tensor(5.1657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"16900 tensor(5.1391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17000 tensor(5.1067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61000\n",
"17100 tensor(5.1680, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17200 tensor(5.2227, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17300 tensor(4.6891, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62000\n",
"17400 tensor(5.0639, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17500 tensor(5.0507, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17600 tensor(4.9053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63000\n",
"17700 tensor(4.8750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"17800 tensor(4.7771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64000\n",
"17900 tensor(4.9931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18000 tensor(5.1721, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18100 tensor(5.0419, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65000\n",
"18200 tensor(5.0401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18300 tensor(4.9477, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18400 tensor(4.9341, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66000\n",
"18500 tensor(5.0645, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18600 tensor(5.1702, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18700 tensor(5.0680, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67000\n",
"18800 tensor(4.7324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"18900 tensor(5.1745, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68000\n",
"19000 tensor(5.0148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19100 tensor(4.9552, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19200 tensor(5.0249, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69000\n",
"19300 tensor(5.0709, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19400 tensor(5.1439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19500 tensor(4.9734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70000\n",
"19600 tensor(5.2309, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19700 tensor(5.0292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"19800 tensor(4.4443, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71000\n",
"19900 tensor(5.2148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20000 tensor(4.8955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20100 tensor(4.9404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72000\n",
"20200 tensor(5.0525, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20300 tensor(5.2610, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73000\n",
"20400 tensor(5.2233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20500 tensor(5.1364, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20600 tensor(4.9917, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74000\n",
"20700 tensor(4.9608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20800 tensor(5.1992, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"20900 tensor(5.1686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75000\n",
"21000 tensor(4.8483, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21100 tensor(5.4130, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21200 tensor(4.7093, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76000\n",
"21300 tensor(5.2148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21400 tensor(4.9624, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21500 tensor(4.9292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77000\n",
"21600 tensor(5.1661, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21700 tensor(5.0725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"21800 tensor(4.7123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78000\n",
"21900 tensor(5.2135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22000 tensor(4.8017, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79000\n",
"22100 tensor(5.1210, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22200 tensor(5.0554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22300 tensor(5.0997, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80000\n",
"22400 tensor(4.8383, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22500 tensor(4.5534, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22600 tensor(5.1061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81000\n",
"22700 tensor(4.9188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22800 tensor(5.0239, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"22900 tensor(4.9186, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82000\n",
"23000 tensor(5.3097, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23100 tensor(5.0750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83000\n",
"23200 tensor(4.9890, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23300 tensor(5.1193, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23400 tensor(5.0708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84000\n",
"23500 tensor(4.9966, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23600 tensor(5.1132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23700 tensor(5.3187, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85000\n",
"23800 tensor(5.2937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"23900 tensor(5.2049, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24000 tensor(5.0551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86000\n",
"24100 tensor(5.3820, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24200 tensor(5.2243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24300 tensor(5.0407, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87000\n",
"24400 tensor(5.1378, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24500 tensor(4.8982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88000\n",
"24600 tensor(5.0213, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24700 tensor(5.1981, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"24800 tensor(5.2277, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89000\n",
"24900 tensor(5.1798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25000 tensor(5.4307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25100 tensor(5.0697, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90000\n",
"25200 tensor(5.1689, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25300 tensor(5.1092, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25400 tensor(5.0354, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91000\n",
"25500 tensor(4.9247, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25600 tensor(4.6927, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25700 tensor(5.0795, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92000\n",
"25800 tensor(5.0327, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"25900 tensor(5.1922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93000\n",
"26000 tensor(4.7922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26100 tensor(5.2135, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26200 tensor(4.8716, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94000\n",
"26300 tensor(5.1197, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26400 tensor(5.0293, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26500 tensor(5.0354, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95000\n",
"26600 tensor(5.2529, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26700 tensor(5.0883, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"26800 tensor(4.8198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96000\n",
"26900 tensor(5.0282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27000 tensor(4.7931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27100 tensor(5.0120, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97000\n",
"27200 tensor(5.0787, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27300 tensor(5.0908, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98000\n",
"27400 tensor(4.9436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27500 tensor(5.1993, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27600 tensor(5.1542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99000\n",
"27700 tensor(5.1619, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27800 tensor(4.9495, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"27900 tensor(4.9496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100000\n",
"28000 tensor(5.3698, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28100 tensor(5.3364, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28200 tensor(4.9353, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101000\n",
"28300 tensor(5.1149, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28400 tensor(4.7524, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28500 tensor(5.0890, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102000\n",
"28600 tensor(5.5074, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28700 tensor(5.1043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103000\n",
"28800 tensor(5.0942, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"28900 tensor(4.7643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29000 tensor(4.9018, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104000\n",
"29100 tensor(5.1374, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29200 tensor(4.8248, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29300 tensor(5.1806, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105000\n",
"29400 tensor(5.1401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29500 tensor(4.9909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29600 tensor(5.3300, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106000\n",
"29700 tensor(4.7358, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29800 tensor(4.8462, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"29900 tensor(4.9805, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107000\n",
"30000 tensor(5.1522, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30100 tensor(5.0940, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108000\n",
"30200 tensor(5.1651, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30300 tensor(4.8124, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30400 tensor(4.7609, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109000\n",
"30500 tensor(4.8187, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30600 tensor(4.9865, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30700 tensor(5.0208, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110000\n",
"30800 tensor(5.0904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"30900 tensor(4.9448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31000 tensor(5.1119, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111000\n",
"31100 tensor(5.1994, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31200 tensor(5.0280, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31300 tensor(5.2946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112000\n",
"31400 tensor(4.9557, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31500 tensor(4.9296, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113000\n",
"31600 tensor(4.8751, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31700 tensor(5.3086, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"31800 tensor(4.7567, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114000\n",
"31900 tensor(4.9941, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32000 tensor(5.2035, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32100 tensor(4.8145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115000\n",
"32200 tensor(5.3565, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32300 tensor(4.9674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32400 tensor(4.9422, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116000\n",
"32500 tensor(5.2489, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32600 tensor(5.0207, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117000\n",
"32700 tensor(5.1569, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32800 tensor(5.0727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"32900 tensor(4.9706, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118000\n",
"33000 tensor(5.0288, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33100 tensor(5.1344, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33200 tensor(5.0739, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119000\n",
"33300 tensor(4.7502, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33400 tensor(4.6710, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33500 tensor(5.3873, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120000\n",
"33600 tensor(4.9222, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33700 tensor(5.1264, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"33800 tensor(4.9766, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"121000\n",
"33900 tensor(5.0954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34000 tensor(5.0210, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"122000\n",
"34100 tensor(5.0876, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34200 tensor(5.1302, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34300 tensor(5.2983, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"123000\n",
"34400 tensor(5.2121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34500 tensor(5.3244, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34600 tensor(4.9485, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"124000\n",
"34700 tensor(5.2090, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34800 tensor(4.9158, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"34900 tensor(5.2443, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"125000\n",
"35000 tensor(5.1179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35100 tensor(5.0717, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35200 tensor(5.1812, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"126000\n",
"35300 tensor(4.9081, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35400 tensor(5.1653, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"127000\n",
"35500 tensor(5.1505, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35600 tensor(5.2750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35700 tensor(5.4164, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"128000\n",
"35800 tensor(5.1043, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"35900 tensor(5.1016, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36000 tensor(5.1461, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"129000\n",
"36100 tensor(5.2056, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36200 tensor(5.1355, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36300 tensor(5.3760, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"130000\n",
"36400 tensor(5.2054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36500 tensor(5.3691, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"131000\n",
"36600 tensor(4.8645, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36700 tensor(5.4200, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"36800 tensor(5.2951, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"132000\n",
"36900 tensor(4.9001, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37000 tensor(5.1626, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37100 tensor(4.8412, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"133000\n",
"37200 tensor(4.7379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37300 tensor(4.8720, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37400 tensor(4.8967, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"134000\n",
"37500 tensor(4.7772, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37600 tensor(4.8302, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37700 tensor(5.1201, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"135000\n",
"37800 tensor(5.1669, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"37900 tensor(5.2367, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"136000\n",
"38000 tensor(4.9588, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38100 tensor(5.1460, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38200 tensor(4.8495, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"137000\n",
"38300 tensor(5.2402, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38400 tensor(4.9173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38500 tensor(5.1901, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"138000\n",
"38600 tensor(4.7848, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38700 tensor(5.0920, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"38800 tensor(4.8625, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"139000\n",
"38900 tensor(4.9405, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39000 tensor(5.1992, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39100 tensor(4.8887, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"140000\n",
"39200 tensor(5.3122, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39300 tensor(5.0380, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"141000\n",
"39400 tensor(4.9989, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39500 tensor(5.1872, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39600 tensor(4.9061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"142000\n",
"39700 tensor(5.2725, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39800 tensor(5.1865, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"39900 tensor(4.8060, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"143000\n",
"40000 tensor(4.9448, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40100 tensor(5.0130, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40200 tensor(5.0753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"144000\n",
"40300 tensor(5.0743, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40400 tensor(5.3216, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40500 tensor(4.7829, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"145000\n",
"40600 tensor(5.3467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40700 tensor(5.2229, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"146000\n",
"40800 tensor(4.9148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"40900 tensor(4.9961, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41000 tensor(5.1233, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"147000\n",
"41100 tensor(4.9480, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41200 tensor(5.1455, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41300 tensor(4.7624, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"148000\n",
"41400 tensor(4.8317, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41500 tensor(5.3050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41600 tensor(4.8164, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"149000\n",
"41700 tensor(5.1542, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41800 tensor(5.0889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"41900 tensor(5.0844, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"150000\n",
"42000 tensor(5.2015, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42100 tensor(4.6232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"151000\n",
"42200 tensor(4.9591, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42300 tensor(5.0888, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42400 tensor(4.9506, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"152000\n",
"42500 tensor(5.0405, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42600 tensor(4.5948, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42700 tensor(5.2494, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"153000\n",
"42800 tensor(5.1611, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"42900 tensor(4.9517, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43000 tensor(5.1292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"154000\n",
"43100 tensor(4.8461, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43200 tensor(5.0209, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43300 tensor(5.1240, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"155000\n",
"43400 tensor(5.0400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43500 tensor(5.3999, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"156000\n",
"43600 tensor(5.3041, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43700 tensor(4.9734, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"43800 tensor(5.1028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"157000\n",
"43900 tensor(4.9373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44000 tensor(4.7666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44100 tensor(5.2840, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"158000\n",
"44200 tensor(4.8623, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44300 tensor(5.3904, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44400 tensor(4.9841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"159000\n",
"44500 tensor(4.8345, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44600 tensor(4.9179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"160000\n",
"44700 tensor(4.8921, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44800 tensor(5.1909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"44900 tensor(5.0083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"161000\n",
"45000 tensor(4.8715, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45100 tensor(5.1750, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45200 tensor(4.9847, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"162000\n",
"45300 tensor(5.1416, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45400 tensor(4.5706, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45500 tensor(4.8476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"163000\n",
"45600 tensor(5.1753, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45700 tensor(5.1638, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"45800 tensor(5.0620, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"164000\n",
"45900 tensor(4.8623, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46000 tensor(5.0958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"165000\n",
"46100 tensor(4.9271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46200 tensor(4.8955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46300 tensor(5.2268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"166000\n",
"46400 tensor(5.2585, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46500 tensor(4.8509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46600 tensor(4.8656, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"167000\n",
"46700 tensor(5.1203, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46800 tensor(4.7754, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"46900 tensor(4.8504, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"168000\n",
"47000 tensor(5.1042, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47100 tensor(5.3206, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47200 tensor(5.0201, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"169000\n",
"47300 tensor(5.1387, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47400 tensor(4.9378, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"170000\n",
"47500 tensor(5.1120, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47600 tensor(5.0451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47700 tensor(5.0578, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"171000\n",
"47800 tensor(5.1534, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"47900 tensor(5.1646, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48000 tensor(4.8708, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"172000\n",
"48100 tensor(5.2243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48200 tensor(5.0376, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48300 tensor(5.1148, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"173000\n",
"48400 tensor(4.7304, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48500 tensor(4.9864, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48600 tensor(5.1053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"174000\n",
"48700 tensor(5.0854, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"48800 tensor(4.8705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"175000\n",
"48900 tensor(5.1415, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49000 tensor(4.9185, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49100 tensor(5.1249, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"176000\n",
"49200 tensor(5.1852, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49300 tensor(4.9728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49400 tensor(5.1541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"177000\n",
"49500 tensor(5.2388, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49600 tensor(4.8876, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49700 tensor(5.0066, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"178000\n",
"49800 tensor(4.9596, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"49900 tensor(5.1618, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50000 tensor(5.0436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"179000\n",
"50100 tensor(5.1911, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50200 tensor(4.9044, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"180000\n",
"50300 tensor(5.1698, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50400 tensor(4.8785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50500 tensor(5.1889, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"181000\n",
"50600 tensor(4.9799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50700 tensor(4.8507, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"50800 tensor(4.7525, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"182000\n",
"50900 tensor(5.0997, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51000 tensor(5.1681, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51100 tensor(5.0502, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"183000\n",
"51200 tensor(5.2121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51300 tensor(4.7797, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51400 tensor(5.1379, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"184000\n",
"51500 tensor(5.0826, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51600 tensor(5.1736, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"185000\n",
"51700 tensor(5.0336, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51800 tensor(5.1728, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"51900 tensor(4.9617, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"186000\n",
"52000 tensor(5.1423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52100 tensor(5.2641, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52200 tensor(5.0587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"187000\n",
"52300 tensor(4.9935, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52400 tensor(5.0743, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52500 tensor(5.0771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"188000\n",
"52600 tensor(5.0109, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52700 tensor(5.1416, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"189000\n",
"52800 tensor(5.1943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"52900 tensor(5.0125, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53000 tensor(4.9937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"190000\n",
"53100 tensor(4.9477, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53200 tensor(4.7327, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53300 tensor(5.2251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"191000\n",
"53400 tensor(5.1211, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53500 tensor(5.0256, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53600 tensor(5.1351, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"192000\n",
"53700 tensor(5.1121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53800 tensor(4.7721, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"53900 tensor(4.7813, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"193000\n",
"54000 tensor(5.2923, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54100 tensor(4.4067, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"194000\n",
"54200 tensor(5.1834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54300 tensor(4.9698, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54400 tensor(4.6432, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"195000\n",
"54500 tensor(5.2431, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54600 tensor(5.2523, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54700 tensor(4.9880, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"196000\n",
"54800 tensor(5.1705, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"54900 tensor(4.8544, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55000 tensor(5.0054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"197000\n",
"55100 tensor(5.0684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55200 tensor(5.1112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55300 tensor(5.0787, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"198000\n",
"55400 tensor(5.2173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55500 tensor(5.2430, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"199000\n",
"55600 tensor(4.8266, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55700 tensor(5.0219, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"55800 tensor(4.5834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"200000\n",
"55900 tensor(5.0455, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56000 tensor(4.9394, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56100 tensor(5.2400, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"201000\n",
"56200 tensor(5.2727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56300 tensor(5.0803, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56400 tensor(5.1270, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"202000\n",
"56500 tensor(5.1758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56600 tensor(4.8928, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56700 tensor(5.0119, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"203000\n",
"56800 tensor(5.5221, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"56900 tensor(4.8963, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"204000\n",
"57000 tensor(5.2899, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57100 tensor(5.0742, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57200 tensor(4.8940, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"205000\n",
"57300 tensor(5.0177, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57400 tensor(5.2162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57500 tensor(4.8549, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"206000\n",
"57600 tensor(4.7256, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57700 tensor(4.9188, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"57800 tensor(5.2799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"207000\n",
"57900 tensor(5.2005, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58000 tensor(4.9628, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"208000\n",
"58100 tensor(5.3107, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58200 tensor(5.3371, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58300 tensor(4.9034, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"209000\n",
"58400 tensor(4.8145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58500 tensor(4.9395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58600 tensor(5.3965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"210000\n",
"58700 tensor(4.9055, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58800 tensor(5.0192, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"58900 tensor(5.2319, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"211000\n",
"59000 tensor(5.1120, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59100 tensor(4.9841, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59200 tensor(5.0341, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"212000\n",
"59300 tensor(4.8836, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59400 tensor(5.0700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"213000\n",
"59500 tensor(5.0094, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59600 tensor(4.9719, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59700 tensor(4.8263, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"214000\n",
"59800 tensor(5.0422, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"59900 tensor(5.0391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60000 tensor(5.0464, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"215000\n",
"60100 tensor(5.1302, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60200 tensor(5.1667, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60300 tensor(4.9755, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"216000\n",
"60400 tensor(5.0662, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60500 tensor(5.1330, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60600 tensor(5.0362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"217000\n",
"60700 tensor(4.9462, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"60800 tensor(5.1028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"218000\n",
"60900 tensor(5.1816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61000 tensor(5.1451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61100 tensor(5.0389, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"219000\n",
"61200 tensor(4.7783, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61300 tensor(5.0208, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61400 tensor(5.2014, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"220000\n",
"61500 tensor(5.0953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61600 tensor(4.9255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61700 tensor(4.9145, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"221000\n",
"61800 tensor(4.9751, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"61900 tensor(5.0497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62000 tensor(5.2065, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"222000\n",
"62100 tensor(4.8016, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62200 tensor(4.9131, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"223000\n",
"62300 tensor(4.9816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62400 tensor(4.9049, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62500 tensor(5.0251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"224000\n",
"62600 tensor(4.9373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62700 tensor(5.0984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"62800 tensor(4.8395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"225000\n",
"62900 tensor(5.3108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63000 tensor(5.1682, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63100 tensor(5.2454, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"226000\n",
"63200 tensor(5.1091, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63300 tensor(4.8577, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63400 tensor(5.0426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"227000\n",
"63500 tensor(4.9117, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63600 tensor(4.9555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"228000\n",
"63700 tensor(4.9914, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63800 tensor(5.2633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"63900 tensor(5.3451, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"229000\n",
"64000 tensor(4.9019, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64100 tensor(5.1581, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64200 tensor(5.0796, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"230000\n",
"64300 tensor(5.0816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64400 tensor(4.9050, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64500 tensor(4.9127, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"231000\n",
"64600 tensor(4.9467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64700 tensor(5.3998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"64800 tensor(5.0784, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"232000\n",
"64900 tensor(5.2337, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65000 tensor(5.1862, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"233000\n",
"65100 tensor(4.7268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65200 tensor(5.2401, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65300 tensor(5.0779, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"234000\n",
"65400 tensor(5.0225, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65500 tensor(5.0218, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65600 tensor(5.1051, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"235000\n",
"65700 tensor(4.8759, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65800 tensor(4.9007, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"65900 tensor(5.1119, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"236000\n",
"66000 tensor(5.1512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66100 tensor(4.8563, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"237000\n",
"66200 tensor(4.8676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66300 tensor(4.8076, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66400 tensor(5.0870, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"238000\n",
"66500 tensor(4.8053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66600 tensor(4.9573, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66700 tensor(4.9532, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"239000\n",
"66800 tensor(5.0142, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"66900 tensor(5.1463, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67000 tensor(5.1613, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"240000\n",
"67100 tensor(5.0125, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67200 tensor(5.3587, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67300 tensor(4.9968, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"241000\n",
"67400 tensor(5.1922, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67500 tensor(5.0287, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"242000\n",
"67600 tensor(5.0606, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67700 tensor(5.0029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"67800 tensor(5.1391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"243000\n",
"67900 tensor(4.8602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68000 tensor(4.9282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68100 tensor(4.7940, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"244000\n",
"68200 tensor(5.0718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68300 tensor(4.9727, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68400 tensor(5.0790, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"245000\n",
"68500 tensor(5.0921, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68600 tensor(5.0522, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68700 tensor(4.7111, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"246000\n",
"68800 tensor(5.1136, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"68900 tensor(4.7838, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"247000\n",
"69000 tensor(5.0396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69100 tensor(4.8118, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69200 tensor(4.7698, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"248000\n",
"69300 tensor(5.2231, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69400 tensor(5.1113, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69500 tensor(5.0799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"249000\n",
"69600 tensor(5.0015, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69700 tensor(5.3189, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"69800 tensor(4.8597, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"250000\n",
"69900 tensor(5.4064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70000 tensor(5.0562, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70100 tensor(4.9788, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"251000\n",
"70200 tensor(5.1238, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70300 tensor(4.7954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"252000\n",
"70400 tensor(5.1417, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70500 tensor(5.0851, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70600 tensor(5.3102, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"253000\n",
"70700 tensor(5.2187, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70800 tensor(5.4180, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"70900 tensor(5.2464, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"254000\n",
"71000 tensor(5.0232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71100 tensor(5.0971, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71200 tensor(5.2289, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"255000\n",
"71300 tensor(4.8301, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71400 tensor(4.9594, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71500 tensor(4.9527, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"256000\n",
"71600 tensor(5.1395, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71700 tensor(5.2029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"257000\n",
"71800 tensor(5.2958, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"71900 tensor(4.9839, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72000 tensor(5.4251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"258000\n",
"72100 tensor(5.1480, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72200 tensor(4.9666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72300 tensor(5.0831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"259000\n",
"72400 tensor(4.9905, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72500 tensor(5.0817, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72600 tensor(5.1276, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"260000\n",
"72700 tensor(4.8184, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72800 tensor(5.2583, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"72900 tensor(4.9964, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"261000\n",
"73000 tensor(5.1877, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73100 tensor(4.8954, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"262000\n",
"73200 tensor(4.9722, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73300 tensor(4.9351, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73400 tensor(5.2009, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"263000\n",
"73500 tensor(5.0204, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73600 tensor(4.9348, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73700 tensor(5.0774, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"264000\n",
"73800 tensor(5.1100, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"73900 tensor(5.1179, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74000 tensor(5.2795, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"265000\n",
"74100 tensor(4.9642, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74200 tensor(4.9902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74300 tensor(4.9494, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"266000\n",
"74400 tensor(5.0772, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74500 tensor(4.6984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"267000\n",
"74600 tensor(5.2123, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74700 tensor(5.2695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"74800 tensor(4.9678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"268000\n",
"74900 tensor(4.7941, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75000 tensor(5.0980, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75100 tensor(5.2916, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"269000\n",
"75200 tensor(5.0355, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75300 tensor(5.2583, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75400 tensor(5.0347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"270000\n",
"75500 tensor(5.3012, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75600 tensor(4.9714, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"271000\n",
"75700 tensor(5.0178, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75800 tensor(4.9899, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"75900 tensor(5.0695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"272000\n",
"76000 tensor(4.8497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76100 tensor(4.8143, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76200 tensor(4.8724, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"273000\n",
"76300 tensor(5.0268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76400 tensor(4.8836, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76500 tensor(5.0496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"274000\n",
"76600 tensor(5.1337, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76700 tensor(4.8128, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"76800 tensor(4.7389, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"275000\n",
"76900 tensor(5.0926, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77000 tensor(4.9780, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"276000\n",
"77100 tensor(5.3241, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77200 tensor(4.9880, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77300 tensor(4.9250, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"277000\n",
"77400 tensor(5.0480, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77500 tensor(4.7045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77600 tensor(5.0635, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"278000\n",
"77700 tensor(5.2122, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77800 tensor(4.8158, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"77900 tensor(4.7201, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"279000\n",
"78000 tensor(5.1809, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78100 tensor(4.8416, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78200 tensor(5.1227, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"280000\n",
"78300 tensor(5.0000, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78400 tensor(5.0232, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"281000\n",
"78500 tensor(5.3672, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78600 tensor(4.8266, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78700 tensor(5.2883, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"282000\n",
"78800 tensor(5.1556, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"78900 tensor(5.3901, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79000 tensor(4.9368, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"283000\n",
"79100 tensor(4.7262, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79200 tensor(4.8004, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79300 tensor(5.2785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"284000\n",
"79400 tensor(5.0932, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79500 tensor(4.9747, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79600 tensor(5.5676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"285000\n",
"79700 tensor(4.7581, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"79800 tensor(5.0427, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"286000\n",
"79900 tensor(5.0180, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80000 tensor(5.1512, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80100 tensor(5.2409, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"287000\n",
"80200 tensor(5.2435, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80300 tensor(5.2283, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80400 tensor(5.3199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"288000\n",
"80500 tensor(5.0696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80600 tensor(5.2085, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80700 tensor(5.2603, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"289000\n",
"80800 tensor(4.7770, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"80900 tensor(5.1645, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"290000\n",
"81000 tensor(5.1184, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81100 tensor(5.1388, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81200 tensor(4.8794, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"291000\n",
"81300 tensor(4.9514, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81400 tensor(5.0975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81500 tensor(5.0673, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"292000\n",
"81600 tensor(5.0619, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81700 tensor(5.1551, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"81800 tensor(5.0009, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"293000\n",
"81900 tensor(5.2308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82000 tensor(5.0541, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82100 tensor(5.0817, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"294000\n",
"82200 tensor(5.0208, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82300 tensor(4.8939, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"295000\n",
"82400 tensor(5.1285, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82500 tensor(4.8489, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82600 tensor(5.0365, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"296000\n",
"82700 tensor(5.0726, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82800 tensor(5.0563, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"82900 tensor(4.7439, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"297000\n",
"83000 tensor(5.1423, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83100 tensor(5.0052, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83200 tensor(5.2631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"298000\n",
"83300 tensor(4.9471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83400 tensor(5.0234, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83500 tensor(4.9797, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"299000\n",
"83600 tensor(4.9047, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83700 tensor(5.1383, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"300000\n",
"83800 tensor(4.8484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"83900 tensor(4.9279, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84000 tensor(5.0825, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"301000\n",
"84100 tensor(5.4376, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84200 tensor(4.5984, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84300 tensor(5.0169, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"302000\n",
"84400 tensor(4.8503, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84500 tensor(5.0322, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84600 tensor(5.2033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"303000\n",
"84700 tensor(4.9052, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84800 tensor(4.9265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"84900 tensor(5.2419, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"304000\n",
"85000 tensor(5.1251, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85100 tensor(5.0005, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"305000\n",
"85200 tensor(5.0029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85300 tensor(5.2717, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85400 tensor(5.2511, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"306000\n",
"85500 tensor(4.9489, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85600 tensor(5.0751, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85700 tensor(5.0649, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"307000\n",
"85800 tensor(5.0660, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"85900 tensor(5.3243, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86000 tensor(5.0756, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"308000\n",
"86100 tensor(4.8535, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86200 tensor(4.9982, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86300 tensor(4.9693, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"309000\n",
"86400 tensor(5.0700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86500 tensor(5.1470, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"310000\n",
"86600 tensor(4.9735, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86700 tensor(4.8743, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"86800 tensor(4.8816, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"311000\n",
"86900 tensor(4.7666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87000 tensor(5.2600, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87100 tensor(5.0591, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"312000\n",
"87200 tensor(5.1044, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87300 tensor(4.9555, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87400 tensor(5.0028, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"313000\n",
"87500 tensor(5.1317, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87600 tensor(5.2621, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87700 tensor(4.8937, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"314000\n",
"87800 tensor(4.8559, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"87900 tensor(5.1339, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"315000\n",
"88000 tensor(4.6132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88100 tensor(4.9682, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88200 tensor(5.1715, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"316000\n",
"88300 tensor(5.1112, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88400 tensor(5.1811, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88500 tensor(4.8046, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"317000\n",
"88600 tensor(4.8210, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88700 tensor(4.8294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"88800 tensor(4.8198, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"318000\n",
"88900 tensor(4.7496, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89000 tensor(4.5654, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89100 tensor(5.1261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"319000\n",
"89200 tensor(4.9374, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89300 tensor(5.3199, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"320000\n",
"89400 tensor(4.6848, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89500 tensor(5.0381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89600 tensor(5.2632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"321000\n",
"89700 tensor(4.8221, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89800 tensor(5.0413, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"89900 tensor(5.0402, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"322000\n",
"90000 tensor(5.0519, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90100 tensor(4.9362, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90200 tensor(5.0257, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"323000\n",
"90300 tensor(5.0438, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90400 tensor(4.7173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90500 tensor(4.9121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"324000\n",
"90600 tensor(5.1965, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90700 tensor(4.8207, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"325000\n",
"90800 tensor(5.0536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"90900 tensor(5.1080, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91000 tensor(5.1328, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"326000\n",
"91100 tensor(4.7033, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91200 tensor(5.1643, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91300 tensor(5.3472, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"327000\n",
"91400 tensor(4.7267, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91500 tensor(4.9552, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91600 tensor(5.2054, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"328000\n",
"91700 tensor(5.0471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"91800 tensor(5.2265, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"329000\n",
"91900 tensor(5.0523, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92000 tensor(5.0726, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92100 tensor(4.9322, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"330000\n",
"92200 tensor(4.5777, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92300 tensor(5.0606, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92400 tensor(5.1319, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"331000\n",
"92500 tensor(5.1785, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92600 tensor(5.0610, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92700 tensor(4.7718, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"332000\n",
"92800 tensor(4.8492, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"92900 tensor(5.3673, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93000 tensor(5.0656, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"333000\n",
"93100 tensor(4.9975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93200 tensor(5.1829, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"334000\n",
"93300 tensor(5.1131, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93400 tensor(5.1585, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93500 tensor(4.9141, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"335000\n",
"93600 tensor(5.0526, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93700 tensor(5.1195, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"93800 tensor(4.8696, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"336000\n",
"93900 tensor(5.0752, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94000 tensor(4.9943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94100 tensor(4.8657, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"337000\n",
"94200 tensor(5.1230, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94300 tensor(5.0292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94400 tensor(5.0633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"338000\n",
"94500 tensor(4.9025, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94600 tensor(5.1955, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"339000\n",
"94700 tensor(4.7546, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94800 tensor(5.2260, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"94900 tensor(5.2268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"340000\n",
"95000 tensor(5.3426, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95100 tensor(5.0509, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95200 tensor(4.8831, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"341000\n",
"95300 tensor(5.2464, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95400 tensor(4.7030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95500 tensor(4.9484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"342000\n",
"95600 tensor(4.9381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95700 tensor(5.1246, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"95800 tensor(5.0476, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"343000\n",
"95900 tensor(4.8117, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96000 tensor(5.1602, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"344000\n",
"96100 tensor(4.9363, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96200 tensor(4.9611, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96300 tensor(5.0053, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"345000\n",
"96400 tensor(5.0561, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96500 tensor(5.2501, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96600 tensor(5.0376, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"346000\n",
"96700 tensor(5.1798, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96800 tensor(5.0684, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"96900 tensor(5.2261, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"347000\n",
"97000 tensor(4.9142, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97100 tensor(5.0132, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97200 tensor(5.1487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"348000\n",
"97300 tensor(4.9732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97400 tensor(5.0108, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"349000\n",
"97500 tensor(5.1571, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97600 tensor(4.9469, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97700 tensor(5.0162, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"350000\n",
"97800 tensor(4.9770, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"97900 tensor(4.7823, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98000 tensor(4.8846, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"351000\n",
"98100 tensor(5.2296, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98200 tensor(4.9268, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98300 tensor(5.0850, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"352000\n",
"98400 tensor(5.1678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98500 tensor(5.2236, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"353000\n",
"98600 tensor(5.2631, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98700 tensor(5.0668, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"98800 tensor(5.0674, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"354000\n",
"98900 tensor(5.0695, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99000 tensor(5.2623, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99100 tensor(5.1354, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"355000\n",
"99200 tensor(5.1557, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99300 tensor(4.9458, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99400 tensor(5.2565, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"356000\n",
"99500 tensor(5.2732, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99600 tensor(5.0011, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99700 tensor(4.8497, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"357000\n",
"99800 tensor(5.0679, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"99900 tensor(5.0005, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100000 tensor(4.9254, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"358000\n",
"100100 tensor(5.4405, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100200 tensor(4.7585, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"359000\n",
"100300 tensor(5.0083, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100400 tensor(4.9893, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100500 tensor(4.8576, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"360000\n",
"100600 tensor(4.8928, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100700 tensor(4.8947, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"100800 tensor(5.2061, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"361000\n",
"100900 tensor(5.1799, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101000 tensor(5.0292, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101100 tensor(5.0604, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"362000\n",
"101200 tensor(5.0287, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101300 tensor(5.2627, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"363000\n",
"101400 tensor(4.8027, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101500 tensor(5.0308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101600 tensor(5.2625, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"364000\n",
"101700 tensor(4.9086, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101800 tensor(4.7064, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"101900 tensor(4.9304, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"365000\n",
"102000 tensor(5.1991, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102100 tensor(5.1194, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102200 tensor(5.3738, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"366000\n",
"102300 tensor(5.2404, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102400 tensor(5.1821, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102500 tensor(5.1676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"367000\n",
"102600 tensor(5.1029, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102700 tensor(5.1796, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"368000\n",
"102800 tensor(5.0834, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"102900 tensor(4.8583, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103000 tensor(5.0342, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"369000\n",
"103100 tensor(5.3343, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103200 tensor(4.9536, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103300 tensor(5.0340, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"370000\n",
"103400 tensor(5.0580, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103500 tensor(4.9467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103600 tensor(5.0845, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"371000\n",
"103700 tensor(5.1711, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103800 tensor(5.0389, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"103900 tensor(4.9456, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"372000\n",
"104000 tensor(4.9239, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104100 tensor(4.9678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"373000\n",
"104200 tensor(5.0781, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104300 tensor(4.8800, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104400 tensor(5.2081, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"374000\n",
"104500 tensor(5.1255, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104600 tensor(4.9805, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104700 tensor(4.9825, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"375000\n",
"104800 tensor(5.5062, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"104900 tensor(5.0209, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105000 tensor(5.1373, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"376000\n",
"105100 tensor(5.0935, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105200 tensor(5.1819, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105300 tensor(5.1386, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"377000\n",
"105400 tensor(5.3493, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105500 tensor(5.2102, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"378000\n",
"105600 tensor(4.9484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105700 tensor(4.9119, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"105800 tensor(4.9584, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"379000\n",
"105900 tensor(5.1010, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106000 tensor(5.1442, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106100 tensor(5.3454, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"380000\n",
"106200 tensor(5.0330, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106300 tensor(5.1396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106400 tensor(4.9693, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"381000\n",
"106500 tensor(5.2242, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106600 tensor(4.9223, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106700 tensor(5.1467, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"382000\n",
"106800 tensor(4.9459, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"106900 tensor(5.2412, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"383000\n",
"107000 tensor(4.9030, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107100 tensor(5.1487, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107200 tensor(5.1303, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"384000\n",
"107300 tensor(4.8918, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107400 tensor(5.0686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107500 tensor(4.8140, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"385000\n",
"107600 tensor(4.8693, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107700 tensor(5.0879, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"107800 tensor(5.0939, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"386000\n",
"107900 tensor(5.3087, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108000 tensor(5.0659, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108100 tensor(5.0273, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"387000\n",
"108200 tensor(5.0357, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108300 tensor(5.0666, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"388000\n",
"108400 tensor(5.1699, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108500 tensor(4.8493, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108600 tensor(5.0399, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"389000\n",
"108700 tensor(4.5729, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108800 tensor(4.9686, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"108900 tensor(4.7163, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"390000\n",
"109000 tensor(5.1307, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109100 tensor(4.7140, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109200 tensor(4.9282, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"391000\n",
"109300 tensor(4.5945, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109400 tensor(4.9045, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109500 tensor(4.8345, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"392000\n",
"109600 tensor(5.0347, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109700 tensor(5.3972, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"393000\n",
"109800 tensor(4.7386, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"109900 tensor(4.9312, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110000 tensor(5.0107, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"394000\n",
"110100 tensor(4.9484, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110200 tensor(5.3088, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110300 tensor(5.1758, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"395000\n",
"110400 tensor(4.6608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110500 tensor(5.0811, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110600 tensor(4.9037, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"396000\n",
"110700 tensor(4.8701, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110800 tensor(5.3046, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"110900 tensor(4.9689, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"397000\n",
"111000 tensor(5.1637, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111100 tensor(4.9131, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"398000\n",
"111200 tensor(4.9308, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111300 tensor(5.0290, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111400 tensor(5.1249, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"399000\n",
"111500 tensor(5.2130, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111600 tensor(5.2382, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111700 tensor(4.8168, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"400000\n",
"111800 tensor(5.0276, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"111900 tensor(5.1121, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112000 tensor(5.0463, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"401000\n",
"112100 tensor(4.6612, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112200 tensor(5.0632, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"402000\n",
"112300 tensor(5.2173, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112400 tensor(5.2271, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112500 tensor(4.8618, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"403000\n",
"112600 tensor(4.9399, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112700 tensor(4.9514, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"112800 tensor(4.5700, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"404000\n",
"112900 tensor(4.8508, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113000 tensor(5.0784, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113100 tensor(5.0205, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"405000\n",
"113200 tensor(4.5998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113300 tensor(4.9371, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113400 tensor(5.0241, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"406000\n",
"113500 tensor(5.2931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113600 tensor(5.2488, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"407000\n",
"113700 tensor(4.8483, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113800 tensor(5.2471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"113900 tensor(5.0391, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"408000\n",
"114000 tensor(4.6978, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114100 tensor(4.8170, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114200 tensor(4.7943, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"409000\n",
"114300 tensor(4.6678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114400 tensor(4.8687, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114500 tensor(4.9471, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"410000\n",
"114600 tensor(5.1608, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114700 tensor(5.1240, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"114800 tensor(5.0586, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"411000\n",
"114900 tensor(5.0095, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115000 tensor(4.9676, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"412000\n",
"115100 tensor(5.0771, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115200 tensor(5.0396, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115300 tensor(5.0663, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"413000\n",
"115400 tensor(5.2931, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115500 tensor(5.2901, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115600 tensor(5.0953, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"414000\n",
"115700 tensor(4.9706, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115800 tensor(5.0768, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"115900 tensor(5.1668, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"415000\n",
"116000 tensor(5.0913, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116100 tensor(4.8294, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116200 tensor(5.2189, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"416000\n",
"116300 tensor(4.3633, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116400 tensor(5.1168, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"417000\n",
"116500 tensor(5.0554, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116600 tensor(5.1477, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116700 tensor(4.9884, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"418000\n",
"116800 tensor(4.8381, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"116900 tensor(5.0709, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117000 tensor(5.0518, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"419000\n",
"117100 tensor(5.0967, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117200 tensor(4.9430, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117300 tensor(5.2903, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"420000\n",
"117400 tensor(5.2733, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117500 tensor(4.8548, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"421000\n",
"117600 tensor(5.1768, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117700 tensor(5.1013, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"117800 tensor(4.9804, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"422000\n",
"117900 tensor(5.0671, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118000 tensor(5.1332, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118100 tensor(4.9436, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"423000\n",
"118200 tensor(4.9859, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118300 tensor(4.9755, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118400 tensor(5.0902, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"424000\n",
"118500 tensor(4.9845, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118600 tensor(5.1098, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118700 tensor(5.1192, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"425000\n",
"118800 tensor(5.1445, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"118900 tensor(5.1731, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"426000\n",
"119000 tensor(4.6613, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119100 tensor(4.9640, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119200 tensor(5.2998, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"427000\n",
"119300 tensor(4.9427, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119400 tensor(4.9993, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119500 tensor(5.0117, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"428000\n",
"119600 tensor(5.3324, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119700 tensor(4.9375, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"119800 tensor(5.0221, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"429000\n",
"119900 tensor(4.7975, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120000 tensor(4.9253, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120100 tensor(4.9377, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"430000\n",
"120200 tensor(5.0909, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120300 tensor(4.9946, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"431000\n",
"120400 tensor(5.3612, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120500 tensor(4.9678, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"120600 tensor(5.2217, device='cuda:0', grad_fn=<NllLossBackward0>)\n",
"432000\n"
]
}
]
},
{
"cell_type": "markdown",
"source": [
"## Eval"
],
"metadata": {
"id": "fnL8e_NkvwtY"
}
},
{
"cell_type": "code",
"source": [
"\n",
"model = Bigram(vocab_size, embed_size).to(device)\n",
"model.load_state_dict(torch.load('model.bin'))\n",
"model.eval()\n",
"\n",
"res = torch.tensor(vocab.forward(['for'])).to(device)\n",
"\n",
"out = model(res)\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))"
],
"metadata": {
"id": "7LykJxNEWNdJ",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "6b8f9a38-dbcd-4285-ff30-a45d0ee760bf"
},
"execution_count": null,
"outputs": [
{
"output_type": "stream",
"name": "stderr",
"text": [
"/usr/local/lib/python3.10/dist-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"
]
},
{
"output_type": "execute_result",
"data": {
"text/plain": [
"[('<unk>', 0, 0.23156249523162842),\n",
" ('the', 1, 0.2045561522245407),\n",
" ('a', 5, 0.0636623203754425),\n",
" ('his', 20, 0.012841351330280304),\n",
" ('their', 40, 0.012044394388794899),\n",
" ('this', 28, 0.011758995242416859),\n",
" ('tho', 33, 0.010536346584558487),\n",
" ('some', 77, 0.008259670808911324),\n",
" ('any', 49, 0.007337945979088545),\n",
" ('an', 38, 0.007214350625872612)]"
]
},
"metadata": {},
"execution_count": 26
}
]
},
{
"cell_type": "code",
"source": [
"vocab = train_dataset.vocab\n",
"res = torch.tensor(vocab.forward(['wait'])).to(device)\n",
"\n",
"out = model(res)\n",
"top = torch.topk(out[0], 20)\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))"
],
"metadata": {
"id": "wep6U9NYWNfx",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "a089a1e6-a4d3-49cb-d860-2c48a4d8f83b"
},
"execution_count": null,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"[('until', 145, 0.17603398859500885),\n",
" ('for', 9, 0.16400693356990814),\n",
" ('<unk>', 0, 0.15191353857517242),\n",
" ('on', 15, 0.04624223709106445),\n",
" ('till', 677, 0.035729214549064636),\n",
" ('a', 5, 0.03367603197693825),\n",
" ('to', 4, 0.029361305758357048),\n",
" ('upon', 59, 0.01995147578418255),\n",
" ('and', 3, 0.01906605064868927),\n",
" ('in', 6, 0.013167516328394413),\n",
" ('at', 14, 0.011669990606606007),\n",
" ('the', 1, 0.010971800424158573),\n",
" ('of', 2, 0.005925077944993973),\n",
" ('with', 16, 0.0055325529538095),\n",
" ('In', 32, 0.004919056314975023),\n",
" ('until\\\\nthe', 5509, 0.004719363059848547),\n",
" ('tor', 532, 0.004647853318601847),\n",
" ('for\\\\nthe', 389, 0.004400868900120258),\n",
" ('two', 74, 0.0043497709557414055),\n",
" ('patiently', 14401, 0.004239553119987249)]"
]
},
"metadata": {},
"execution_count": 28
}
]
},
{
"cell_type": "code",
"source": [
"cos = nn.CosineSimilarity(dim=1, eps=1e-6)\n",
"\n",
"embeddings = model.model[0].weight\n",
"\n",
"vec = embeddings[vocab['take']]\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))"
],
"metadata": {
"id": "VCkcPiNCWNiZ",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "afcac703-ef72-45d3-b6af-cce72c8d9483"
},
"execution_count": null,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"[('take', 152, 1.0000001192092896),\n",
" ('took', 248, 0.8024641275405884),\n",
" ('taking', 538, 0.775847852230072),\n",
" ('takes', 1108, 0.7307196855545044),\n",
" ('to\\\\ntake', 3165, 0.7301462888717651),\n",
" ('taken', 180, 0.6280043125152588),\n",
" ('will\\\\ntake', 11101, 0.6152595281600952),\n",
" ('tako', 6281, 0.5979241132736206),\n",
" ('have\\\\ntaken', 15483, 0.5244049429893494),\n",
" ('Take', 5203, 0.5183135867118835)]"
]
},
"metadata": {},
"execution_count": 30
}
]
},
{
"cell_type": "code",
"source": [
"vocab = train_dataset.vocab\n",
"res = torch.tensor(vocab.forward(['take'])).to(device)\n",
"\n",
"out = model(res)\n",
"top = torch.topk(out[0], 20)\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))"
],
"metadata": {
"id": "km2U9fR4WNlI",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "ae1fd560-ac8f-480a-a961-f8eb0f265fac"
},
"execution_count": null,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"[('<unk>', 0, 0.18006379902362823),\n",
" ('the', 1, 0.11970410495996475),\n",
" ('a', 5, 0.07213426381349564),\n",
" ('care', 611, 0.027887443080544472),\n",
" ('up', 69, 0.027840441092848778),\n",
" ('it', 17, 0.023985734209418297),\n",
" ('place', 159, 0.020694952458143234),\n",
" ('advantage', 1458, 0.015635941177606583),\n",
" ('his', 20, 0.014868981204926968),\n",
" ('part', 131, 0.013506578281521797),\n",
" ('an', 38, 0.013118326663970947),\n",
" ('their', 40, 0.010849231854081154),\n",
" ('hold', 478, 0.010717789642512798),\n",
" ('them', 72, 0.010186631232500076),\n",
" ('to', 4, 0.009746687486767769),\n",
" ('this', 28, 0.009519988670945168),\n",
" ('any', 49, 0.009436620399355888),\n",
" ('her', 53, 0.008774512447416782),\n",
" ('him', 70, 0.008407332003116608),\n",
" ('all', 34, 0.007673078216612339)]"
]
},
"metadata": {},
"execution_count": 33
}
]
},
{
"cell_type": "markdown",
"source": [
"### Create files for geval"
],
"metadata": {
"id": "GVDTtvY6v2pt"
}
},
{
"cell_type": "code",
"source": [
"def get_values(presc_word, model, vocab):\n",
" ixs = torch.tensor(vocab.forward([presc_word])).to(device)\n",
" out = model(ixs)\n",
" top = torch.topk(out[0], 20)\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 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]\n",
" else:\n",
" last_word += text[i]\n",
" return last_word[::-1]\n",
"\n",
"def 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",
"def sum_prob(dic):\n",
" probsum = sum(float(val) for key, val in dic.items())\n",
" probsum = probsum - float(dic.get('<unk>', 0))\n",
" if \"<unk>\" in dic.keys():\n",
" del dic['<unk>']\n",
" tab = [(key, val) for key, val in dic.items()]\n",
" tab.append(('<unk>', 1-probsum))\n",
" return tab\n",
"\n",
"def format_to(dic):\n",
" tab = sum_prob(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"
],
"metadata": {
"id": "rUqKisSfWNnp"
},
"execution_count": null,
"outputs": []
},
{
"cell_type": "code",
"source": [
"with lzma.open(test_file, 'rt') as file:\n",
" predict_words = []\n",
" results = []\n",
" for line in file:\n",
" split = line.split('\\t')[6:] \n",
" predict_words.append(last_word(split[0]))\n",
" \n",
" vocab = train_dataset.vocab\n",
" for presc_word in predict_words:\n",
" results.append(dict(get_values(presc_word, model ,vocab)))\n",
" \n",
" with open(out_file, 'w') as outfile:\n",
" for elem in results:\n",
" outfile.write(format_to(elem))\n",
"\n"
],
"metadata": {
"id": "cVpm-tRzWNqR"
},
"execution_count": null,
"outputs": []
}
]
}