Skip to content

Instantly share code, notes, and snippets.

@cthoyt
Created August 25, 2020 02:38
Show Gist options
  • Save cthoyt/190233fd98a11306ceb13f2ee0e95a9e to your computer and use it in GitHub Desktop.
Save cthoyt/190233fd98a11306ceb13f2ee0e95a9e to your computer and use it in GitHub Desktop.
Constrained Evaluation on Hetionet.ipynb
Display the source blob
Display the rendered blob
Raw
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "Constrained Evaluation on Hetionet.ipynb",
"provenance": [],
"collapsed_sections": [],
"authorship_tag": "ABX9TyOTzu//BKYj18jmkRcTzkXY",
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/cthoyt/190233fd98a11306ceb13f2ee0e95a9e/constrained-evaluation-on-hetionet.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "code",
"metadata": {
"id": "vFJlGzGGUixD",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 52
},
"outputId": "cda143f5-ba0b-4b2b-dc10-cef7b69de818"
},
"source": [
"!pip install --quiet --upgrade pip\n",
"!pip install --quiet wheel\n",
"!pip install --quiet git+https://github.com/pykeen/pykeen.git@evaluate-on-subset-of-entities "
],
"execution_count": 1,
"outputs": [
{
"output_type": "stream",
"text": [
"\u001b[?25l\r\u001b[K |▏ | 10kB 27.5MB/s eta 0:00:01\r\u001b[K |▍ | 20kB 3.0MB/s eta 0:00:01\r\u001b[K |▋ | 30kB 3.9MB/s eta 0:00:01\r\u001b[K |▉ | 40kB 4.2MB/s eta 0:00:01\r\u001b[K |█ | 51kB 3.1MB/s eta 0:00:01\r\u001b[K |█▎ | 61kB 3.5MB/s eta 0:00:01\r\u001b[K |█▌ | 71kB 4.0MB/s eta 0:00:01\r\u001b[K |█▊ | 81kB 4.2MB/s eta 0:00:01\r\u001b[K |██ | 92kB 4.5MB/s eta 0:00:01\r\u001b[K |██▏ | 102kB 4.6MB/s eta 0:00:01\r\u001b[K |██▍ | 112kB 4.6MB/s eta 0:00:01\r\u001b[K |██▋ | 122kB 4.6MB/s eta 0:00:01\r\u001b[K |██▉ | 133kB 4.6MB/s eta 0:00:01\r\u001b[K |███ | 143kB 4.6MB/s eta 0:00:01\r\u001b[K |███▎ | 153kB 4.6MB/s eta 0:00:01\r\u001b[K |███▌ | 163kB 4.6MB/s eta 0:00:01\r\u001b[K |███▊ | 174kB 4.6MB/s eta 0:00:01\r\u001b[K |████ | 184kB 4.6MB/s eta 0:00:01\r\u001b[K |████▏ | 194kB 4.6MB/s eta 0:00:01\r\u001b[K |████▍ | 204kB 4.6MB/s eta 0:00:01\r\u001b[K |████▋ | 215kB 4.6MB/s eta 0:00:01\r\u001b[K |████▉ | 225kB 4.6MB/s eta 0:00:01\r\u001b[K |█████ | 235kB 4.6MB/s eta 0:00:01\r\u001b[K |█████▎ | 245kB 4.6MB/s eta 0:00:01\r\u001b[K |█████▌ | 256kB 4.6MB/s eta 0:00:01\r\u001b[K |█████▋ | 266kB 4.6MB/s eta 0:00:01\r\u001b[K |█████▉ | 276kB 4.6MB/s eta 0:00:01\r\u001b[K |██████ | 286kB 4.6MB/s eta 0:00:01\r\u001b[K |██████▎ | 296kB 4.6MB/s eta 0:00:01\r\u001b[K |██████▌ | 307kB 4.6MB/s eta 0:00:01\r\u001b[K |██████▊ | 317kB 4.6MB/s eta 0:00:01\r\u001b[K |███████ | 327kB 4.6MB/s eta 0:00:01\r\u001b[K |███████▏ | 337kB 4.6MB/s eta 0:00:01\r\u001b[K |███████▍ | 348kB 4.6MB/s eta 0:00:01\r\u001b[K |███████▋ | 358kB 4.6MB/s eta 0:00:01\r\u001b[K |███████▉ | 368kB 4.6MB/s eta 0:00:01\r\u001b[K |████████ | 378kB 4.6MB/s eta 0:00:01\r\u001b[K |████████▎ | 389kB 4.6MB/s eta 0:00:01\r\u001b[K |████████▌ | 399kB 4.6MB/s eta 0:00:01\r\u001b[K |████████▊ | 409kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████ | 419kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████▏ | 430kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████▍ | 440kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████▋ | 450kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████▉ | 460kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████ | 471kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████▎ | 481kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████▌ | 491kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████▊ | 501kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████ | 512kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████▏ | 522kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████▎ | 532kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████▌ | 542kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████▊ | 552kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████ | 563kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████▏ | 573kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████▍ | 583kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████▋ | 593kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████▉ | 604kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████ | 614kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████▎ | 624kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████▌ | 634kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████▊ | 645kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████ | 655kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████▏ | 665kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████▍ | 675kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████▋ | 686kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████▉ | 696kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████ | 706kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████▎ | 716kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████▌ | 727kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████▊ | 737kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████ | 747kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████▏ | 757kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████▍ | 768kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████▋ | 778kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████▉ | 788kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████ | 798kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████▏ | 808kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████▍ | 819kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████▋ | 829kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████▉ | 839kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████ | 849kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████▎ | 860kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████▌ | 870kB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████▊ | 880kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████ | 890kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████▏ | 901kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████▍ | 911kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████▋ | 921kB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████▉ | 931kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████ | 942kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████▎ | 952kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████▌ | 962kB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████▊ | 972kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████ | 983kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████▏ | 993kB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████▍ | 1.0MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████▋ | 1.0MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████▉ | 1.0MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████ | 1.0MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████▎ | 1.0MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████▌ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████▋ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████▉ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████▎ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████▌ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████▊ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████▏ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████▍ | 1.1MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████▋ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████▉ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████▎ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████▌ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████▊ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████▏ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████▍ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████▋ | 1.2MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████▉ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████▎ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████▌ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████▊ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████▏ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████▎ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████▌ | 1.3MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████▊ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████████ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████████▏ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████████▍ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████████▋ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |█████████████████████████████▉ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████████ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████████▎ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████████▌ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |██████████████████████████████▊ | 1.4MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████████ | 1.5MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████████▏| 1.5MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████████▍| 1.5MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████████▋| 1.5MB 4.6MB/s eta 0:00:01\r\u001b[K |███████████████████████████████▉| 1.5MB 4.6MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 1.5MB 4.6MB/s \n",
"\u001b[?25h Building wheel for pykeen (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "DYMzPewlU9o4",
"colab_type": "code",
"colab": {}
},
"source": [
"import getpass\n",
"import sys\n",
"import time\n",
"\n",
"import pykeen\n",
"from pykeen.pipeline import pipeline"
],
"execution_count": 2,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "UDkPsBmyVCUn",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 52
},
"outputId": "acb873e5-34eb-41d9-ade9-74786e53c266"
},
"source": [
"print(sys.version)"
],
"execution_count": 3,
"outputs": [
{
"output_type": "stream",
"text": [
"3.6.9 (default, Jul 17 2020, 12:50:27) \n",
"[GCC 8.4.0]\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "B9unQGZtVEG-",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "8296654d-c20f-45c3-fab2-8503ca661306"
},
"source": [
"print(time.asctime())"
],
"execution_count": 4,
"outputs": [
{
"output_type": "stream",
"text": [
"Tue Aug 25 01:57:57 2020\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "CvPK1QUmVESl",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
},
"outputId": "905b73b3-e3cf-438b-c739-bc6b5a93bd1e"
},
"source": [
"print(getpass.getuser())"
],
"execution_count": 5,
"outputs": [
{
"output_type": "stream",
"text": [
"root\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "b4zrEyS6VEbn",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 36
},
"outputId": "8c79a2b7-32c0-4414-cb6c-0028bb3d6198"
},
"source": [
"pykeen.get_version()"
],
"execution_count": 6,
"outputs": [
{
"output_type": "execute_result",
"data": {
"application/vnd.google.colaboratory.intrinsic+json": {
"type": "string"
},
"text/plain": [
"'1.0.4-dev'"
]
},
"metadata": {
"tags": []
},
"execution_count": 6
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "f-xHhVhFVKjy",
"colab_type": "code",
"colab": {}
},
"source": [
"restrict_evaluation_to_relations = {'CtD', 'CpD'}"
],
"execution_count": 7,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "SHgB1NU5VUpu",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1000
},
"outputId": "0447c5d4-bc74-4e30-8331-24c77ef7b444"
},
"source": [
"results = pipeline(\n",
" dataset='Hetionet',\n",
" restrict_evaluation_to_relations=restrict_evaluation_to_relations,\n",
" model='RotatE',\n",
")"
],
"execution_count": 8,
"outputs": [
{
"output_type": "stream",
"text": [
"No random seed is specified. Setting to 1534516120.\n",
"Requested ratio[0]=0.800 (equal to size 1800157), but got 0.800 (equal to size 1800806) to ensure that all entities/relations occur in train.\n",
"Requested ratio[1]=0.100 (equal to size 225019), but got 0.100 (equal to size 224647) to ensure that all entities/relations occur in train.\n",
"Training epochs on cuda: 0%| | 0/5 [00:00<?, ?epoch/s]INFO:pykeen.training.training_loop:using stopper: <pykeen.stoppers.stopper.NopStopper object at 0x7fa4841787b8>\n",
"\n",
"Training batches on cuda: 0%| | 0/7035 [00:00<?, ?batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 1/7035 [00:00<14:51, 7.89batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 11/7035 [00:00<10:46, 10.87batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 20/7035 [00:00<07:55, 14.74batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 30/7035 [00:00<05:55, 19.68batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 39/7035 [00:00<04:32, 25.70batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 48/7035 [00:00<03:33, 32.70batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 57/7035 [00:00<02:53, 40.20batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 66/7035 [00:00<02:26, 47.70batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 75/7035 [00:00<02:05, 55.28batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 84/7035 [00:01<01:52, 61.83batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 93/7035 [00:01<01:42, 67.90batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 102/7035 [00:01<01:35, 72.95batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 111/7035 [00:01<01:29, 77.09batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 120/7035 [00:01<01:26, 80.26batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 129/7035 [00:01<01:23, 82.82batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 138/7035 [00:01<01:21, 84.63batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 147/7035 [00:01<01:21, 84.50batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 156/7035 [00:01<01:20, 85.91batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 165/7035 [00:01<01:19, 86.91batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 174/7035 [00:02<01:18, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 183/7035 [00:02<01:17, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 192/7035 [00:02<01:17, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 201/7035 [00:02<01:17, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 210/7035 [00:02<01:17, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 219/7035 [00:02<01:17, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 228/7035 [00:02<01:17, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 237/7035 [00:02<01:18, 86.81batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 246/7035 [00:02<01:18, 86.87batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 255/7035 [00:03<01:18, 86.75batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 264/7035 [00:03<01:17, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 273/7035 [00:03<01:16, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 282/7035 [00:03<01:16, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 291/7035 [00:03<01:16, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 301/7035 [00:03<01:15, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 310/7035 [00:03<01:15, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 319/7035 [00:03<01:17, 86.35batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 328/7035 [00:03<01:18, 85.52batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 337/7035 [00:03<01:17, 86.16batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 346/7035 [00:04<01:16, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 356/7035 [00:04<01:15, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 365/7035 [00:04<01:15, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 375/7035 [00:04<01:14, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 384/7035 [00:04<01:14, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 393/7035 [00:04<01:15, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 402/7035 [00:04<01:14, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 411/7035 [00:04<01:14, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 420/7035 [00:04<01:14, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 429/7035 [00:04<01:14, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 438/7035 [00:05<01:14, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 448/7035 [00:05<01:13, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 457/7035 [00:05<01:14, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 466/7035 [00:05<01:14, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 475/7035 [00:05<01:14, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 484/7035 [00:05<01:14, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 493/7035 [00:05<01:14, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 502/7035 [00:05<01:14, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 511/7035 [00:05<01:15, 86.27batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 521/7035 [00:06<01:14, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 530/7035 [00:06<01:14, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 539/7035 [00:06<01:14, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 549/7035 [00:06<01:13, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 559/7035 [00:06<01:12, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 568/7035 [00:06<01:12, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 578/7035 [00:06<01:12, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 587/7035 [00:06<01:12, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 596/7035 [00:06<01:12, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 606/7035 [00:06<01:11, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 616/7035 [00:07<01:11, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 625/7035 [00:07<01:11, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 634/7035 [00:07<01:11, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 643/7035 [00:07<01:11, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 653/7035 [00:07<01:10, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 662/7035 [00:07<01:11, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 671/7035 [00:07<01:11, 89.57batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 680/7035 [00:07<01:11, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 689/7035 [00:07<01:11, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 698/7035 [00:08<01:12, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 707/7035 [00:08<01:11, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 716/7035 [00:08<01:11, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 725/7035 [00:08<01:11, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 734/7035 [00:08<01:10, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 743/7035 [00:08<01:11, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 753/7035 [00:08<01:10, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 762/7035 [00:08<01:10, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 771/7035 [00:08<01:10, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 780/7035 [00:08<01:10, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 789/7035 [00:09<01:10, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 798/7035 [00:09<01:10, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 807/7035 [00:09<01:10, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 816/7035 [00:09<01:10, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 825/7035 [00:09<01:10, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 834/7035 [00:09<01:10, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 843/7035 [00:09<01:09, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 852/7035 [00:09<01:09, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 861/7035 [00:09<01:09, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 870/7035 [00:09<01:09, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 879/7035 [00:10<01:09, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 888/7035 [00:10<01:09, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 897/7035 [00:10<01:09, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 906/7035 [00:10<01:09, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 915/7035 [00:10<01:09, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 924/7035 [00:10<01:09, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 933/7035 [00:10<01:09, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 942/7035 [00:10<01:09, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 952/7035 [00:10<01:08, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 961/7035 [00:10<01:09, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 970/7035 [00:11<01:09, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 980/7035 [00:11<01:08, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 989/7035 [00:11<01:08, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 998/7035 [00:11<01:08, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1007/7035 [00:11<01:08, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1016/7035 [00:11<01:08, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1025/7035 [00:11<01:07, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1034/7035 [00:11<01:07, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1043/7035 [00:11<01:07, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1052/7035 [00:12<01:08, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1061/7035 [00:12<01:07, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1070/7035 [00:12<01:07, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1079/7035 [00:12<01:07, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1088/7035 [00:12<01:08, 87.29batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1097/7035 [00:12<01:07, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1106/7035 [00:12<01:07, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1116/7035 [00:12<01:06, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1126/7035 [00:12<01:06, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1135/7035 [00:12<01:06, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1144/7035 [00:13<01:07, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1153/7035 [00:13<01:07, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1163/7035 [00:13<01:06, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1173/7035 [00:13<01:05, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1182/7035 [00:13<01:05, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1191/7035 [00:13<01:07, 86.68batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1200/7035 [00:13<01:08, 85.53batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1209/7035 [00:13<01:07, 86.08batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1218/7035 [00:13<01:07, 86.63batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1227/7035 [00:14<01:08, 85.32batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1236/7035 [00:14<01:07, 86.08batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1245/7035 [00:14<01:06, 87.00batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1254/7035 [00:14<01:06, 87.41batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1263/7035 [00:14<01:05, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1272/7035 [00:14<01:05, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1281/7035 [00:14<01:05, 87.47batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1290/7035 [00:14<01:05, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1299/7035 [00:14<01:05, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1309/7035 [00:14<01:04, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1318/7035 [00:15<01:04, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1327/7035 [00:15<01:04, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1337/7035 [00:15<01:03, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1347/7035 [00:15<01:03, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1356/7035 [00:15<01:03, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1365/7035 [00:15<01:03, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1374/7035 [00:15<01:03, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1383/7035 [00:15<01:03, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1392/7035 [00:15<01:03, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1401/7035 [00:15<01:03, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1410/7035 [00:16<01:03, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1420/7035 [00:16<01:03, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1429/7035 [00:16<01:03, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1438/7035 [00:16<01:03, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1447/7035 [00:16<01:03, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1456/7035 [00:16<01:03, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1465/7035 [00:16<01:03, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1475/7035 [00:16<01:02, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1484/7035 [00:16<01:02, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1493/7035 [00:17<01:02, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1502/7035 [00:17<01:02, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1512/7035 [00:17<01:01, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1521/7035 [00:17<01:01, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1531/7035 [00:17<01:01, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1540/7035 [00:17<01:02, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1549/7035 [00:17<01:01, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1559/7035 [00:17<01:01, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1569/7035 [00:17<01:00, 89.83batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1579/7035 [00:17<01:00, 89.98batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1588/7035 [00:18<01:01, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1597/7035 [00:18<01:01, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1606/7035 [00:18<01:01, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1615/7035 [00:18<01:01, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1624/7035 [00:18<01:00, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1633/7035 [00:18<01:00, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1642/7035 [00:18<01:01, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1651/7035 [00:18<01:00, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1660/7035 [00:18<01:00, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1669/7035 [00:19<01:00, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1678/7035 [00:19<01:01, 86.66batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1687/7035 [00:19<01:01, 87.27batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1696/7035 [00:19<01:01, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1705/7035 [00:19<01:00, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1714/7035 [00:19<01:00, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1723/7035 [00:19<01:00, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1732/7035 [00:19<01:00, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1741/7035 [00:19<01:00, 87.54batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1750/7035 [00:19<01:00, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1759/7035 [00:20<01:00, 87.36batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1768/7035 [00:20<00:59, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1777/7035 [00:20<00:59, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1786/7035 [00:20<00:59, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1795/7035 [00:20<00:59, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1804/7035 [00:20<00:59, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1813/7035 [00:20<00:59, 87.31batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1822/7035 [00:20<00:59, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1831/7035 [00:20<00:59, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1840/7035 [00:20<00:58, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1849/7035 [00:21<00:58, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1858/7035 [00:21<00:58, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1867/7035 [00:21<00:58, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1876/7035 [00:21<00:58, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1885/7035 [00:21<00:57, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1894/7035 [00:21<00:57, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1903/7035 [00:21<00:57, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1912/7035 [00:21<00:57, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1921/7035 [00:21<00:57, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1930/7035 [00:21<00:57, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1939/7035 [00:22<00:57, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1948/7035 [00:22<00:57, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1957/7035 [00:22<00:57, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1966/7035 [00:22<00:57, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1975/7035 [00:22<00:57, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1985/7035 [00:22<00:56, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1994/7035 [00:22<00:56, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 2003/7035 [00:22<00:56, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2012/7035 [00:22<00:56, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2021/7035 [00:22<00:56, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2030/7035 [00:23<00:56, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2039/7035 [00:23<00:56, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2048/7035 [00:23<00:56, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2057/7035 [00:23<00:56, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2066/7035 [00:23<00:56, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2075/7035 [00:23<00:56, 87.17batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2084/7035 [00:23<00:56, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2093/7035 [00:23<00:56, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2102/7035 [00:23<00:55, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2112/7035 [00:24<00:55, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2121/7035 [00:24<00:55, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2130/7035 [00:24<00:55, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2139/7035 [00:24<00:56, 87.15batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2148/7035 [00:24<00:55, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2157/7035 [00:24<00:55, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2166/7035 [00:24<00:55, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2175/7035 [00:24<00:55, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2184/7035 [00:24<00:55, 87.03batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2193/7035 [00:24<00:55, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2202/7035 [00:25<00:54, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2211/7035 [00:25<00:54, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2220/7035 [00:25<00:54, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2229/7035 [00:25<00:54, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2238/7035 [00:25<00:54, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2247/7035 [00:25<00:55, 87.05batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2256/7035 [00:25<00:54, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2265/7035 [00:25<00:54, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2274/7035 [00:25<00:54, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2283/7035 [00:25<00:54, 86.68batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2292/7035 [00:26<00:54, 86.37batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2301/7035 [00:26<00:55, 85.58batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2310/7035 [00:26<00:55, 85.59batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2319/7035 [00:26<00:54, 86.84batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2328/7035 [00:26<00:54, 87.04batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2337/7035 [00:26<00:54, 86.65batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2346/7035 [00:26<00:53, 87.04batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2355/7035 [00:26<00:53, 87.06batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2364/7035 [00:26<00:53, 87.35batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2373/7035 [00:27<00:53, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2382/7035 [00:27<00:53, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2391/7035 [00:27<00:53, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2400/7035 [00:27<00:52, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2409/7035 [00:27<00:52, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2418/7035 [00:27<00:52, 87.68batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2427/7035 [00:27<00:52, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2436/7035 [00:27<00:52, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2445/7035 [00:27<00:51, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2454/7035 [00:27<00:51, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2463/7035 [00:28<00:52, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2472/7035 [00:28<00:52, 87.35batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2481/7035 [00:28<00:51, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2490/7035 [00:28<00:51, 87.48batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2499/7035 [00:28<00:51, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2509/7035 [00:28<00:51, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2518/7035 [00:28<00:51, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2527/7035 [00:28<00:51, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2536/7035 [00:28<00:51, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2545/7035 [00:28<00:51, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2554/7035 [00:29<00:50, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2563/7035 [00:29<00:50, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2572/7035 [00:29<00:50, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2581/7035 [00:29<00:50, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2590/7035 [00:29<00:50, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2600/7035 [00:29<00:49, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2609/7035 [00:29<00:49, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2618/7035 [00:29<00:49, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2627/7035 [00:29<00:49, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2636/7035 [00:29<00:49, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2646/7035 [00:30<00:49, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2655/7035 [00:30<00:49, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2664/7035 [00:30<00:49, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2673/7035 [00:30<00:49, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2682/7035 [00:30<00:49, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2691/7035 [00:30<00:49, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2700/7035 [00:30<00:49, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2709/7035 [00:30<00:48, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2718/7035 [00:30<00:48, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2727/7035 [00:31<00:48, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2736/7035 [00:31<00:49, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2745/7035 [00:31<00:48, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2754/7035 [00:31<00:48, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2763/7035 [00:31<00:49, 86.95batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2772/7035 [00:31<00:48, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2781/7035 [00:31<00:48, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2790/7035 [00:31<00:48, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2799/7035 [00:31<00:48, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2809/7035 [00:31<00:47, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2818/7035 [00:32<00:47, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2827/7035 [00:32<00:47, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2836/7035 [00:32<00:47, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2845/7035 [00:32<00:47, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2854/7035 [00:32<00:47, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2863/7035 [00:32<00:47, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2872/7035 [00:32<00:47, 87.42batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2881/7035 [00:32<00:47, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2890/7035 [00:32<00:47, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2899/7035 [00:32<00:47, 86.88batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2908/7035 [00:33<00:47, 87.22batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2917/7035 [00:33<00:47, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2926/7035 [00:33<00:46, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2935/7035 [00:33<00:46, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2944/7035 [00:33<00:46, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2953/7035 [00:33<00:46, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2962/7035 [00:33<00:46, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2971/7035 [00:33<00:46, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2980/7035 [00:33<00:45, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2989/7035 [00:34<00:45, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2998/7035 [00:34<00:45, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3007/7035 [00:34<00:46, 86.36batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3016/7035 [00:34<00:46, 87.08batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3025/7035 [00:34<00:45, 87.72batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3034/7035 [00:34<00:45, 87.03batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3043/7035 [00:34<00:45, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3053/7035 [00:34<00:44, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3062/7035 [00:34<00:44, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3072/7035 [00:34<00:44, 89.48batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3081/7035 [00:35<00:44, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3090/7035 [00:35<00:44, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3099/7035 [00:35<00:44, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3108/7035 [00:35<00:44, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3117/7035 [00:35<00:45, 86.84batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3126/7035 [00:35<00:44, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3135/7035 [00:35<00:44, 87.42batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3145/7035 [00:35<00:44, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3154/7035 [00:35<00:44, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3163/7035 [00:35<00:43, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3172/7035 [00:36<00:43, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3181/7035 [00:36<00:43, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3190/7035 [00:36<00:43, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3199/7035 [00:36<00:43, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3208/7035 [00:36<00:43, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3217/7035 [00:36<00:43, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3226/7035 [00:36<00:42, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3235/7035 [00:36<00:42, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3244/7035 [00:36<00:42, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3253/7035 [00:37<00:42, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3262/7035 [00:37<00:42, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3272/7035 [00:37<00:42, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3281/7035 [00:37<00:42, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3291/7035 [00:37<00:41, 89.63batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3300/7035 [00:37<00:41, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3309/7035 [00:37<00:41, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3318/7035 [00:37<00:41, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3327/7035 [00:37<00:41, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3336/7035 [00:37<00:41, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3345/7035 [00:38<00:41, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3354/7035 [00:38<00:41, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3363/7035 [00:38<00:41, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3372/7035 [00:38<00:41, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3381/7035 [00:38<00:41, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3390/7035 [00:38<00:41, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3399/7035 [00:38<00:41, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3408/7035 [00:38<00:41, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3417/7035 [00:38<00:40, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3426/7035 [00:38<00:40, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3435/7035 [00:39<00:40, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3444/7035 [00:39<00:40, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3453/7035 [00:39<00:40, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3462/7035 [00:39<00:40, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3471/7035 [00:39<00:40, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3480/7035 [00:39<00:40, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3489/7035 [00:39<00:40, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3498/7035 [00:39<00:40, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3507/7035 [00:39<00:40, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3516/7035 [00:39<00:39, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3525/7035 [00:40<00:39, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3534/7035 [00:40<00:39, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3543/7035 [00:40<00:39, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3552/7035 [00:40<00:39, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3562/7035 [00:40<00:38, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3571/7035 [00:40<00:38, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3580/7035 [00:40<00:38, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3589/7035 [00:40<00:38, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3598/7035 [00:40<00:38, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3607/7035 [00:41<00:38, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3616/7035 [00:41<00:38, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3625/7035 [00:41<00:38, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3634/7035 [00:41<00:38, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3643/7035 [00:41<00:38, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3652/7035 [00:41<00:38, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3661/7035 [00:41<00:38, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3670/7035 [00:41<00:38, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3679/7035 [00:41<00:37, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3688/7035 [00:41<00:37, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3697/7035 [00:42<00:37, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3706/7035 [00:42<00:37, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3715/7035 [00:42<00:37, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3724/7035 [00:42<00:37, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3734/7035 [00:42<00:37, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3743/7035 [00:42<00:37, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3752/7035 [00:42<00:36, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3761/7035 [00:42<00:36, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3771/7035 [00:42<00:36, 89.92batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3780/7035 [00:42<00:36, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3789/7035 [00:43<00:36, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3798/7035 [00:43<00:36, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3808/7035 [00:43<00:35, 89.71batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3817/7035 [00:43<00:36, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3826/7035 [00:43<00:36, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3835/7035 [00:43<00:36, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3844/7035 [00:43<00:36, 86.41batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3853/7035 [00:43<00:36, 86.99batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3862/7035 [00:43<00:36, 87.32batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3871/7035 [00:43<00:36, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3880/7035 [00:44<00:35, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3890/7035 [00:44<00:35, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3899/7035 [00:44<00:35, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3908/7035 [00:44<00:35, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3917/7035 [00:44<00:35, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3926/7035 [00:44<00:35, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3935/7035 [00:44<00:35, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3944/7035 [00:44<00:35, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3953/7035 [00:44<00:35, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3962/7035 [00:45<00:34, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3971/7035 [00:45<00:34, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3980/7035 [00:45<00:34, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3989/7035 [00:45<00:34, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3998/7035 [00:45<00:34, 87.51batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4007/7035 [00:45<00:34, 87.20batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4016/7035 [00:45<00:34, 86.91batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4025/7035 [00:45<00:34, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4034/7035 [00:45<00:34, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4043/7035 [00:45<00:34, 87.45batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4052/7035 [00:46<00:34, 87.29batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4061/7035 [00:46<00:33, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4070/7035 [00:46<00:33, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4079/7035 [00:46<00:33, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4088/7035 [00:46<00:33, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4097/7035 [00:46<00:33, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4106/7035 [00:46<00:33, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4115/7035 [00:46<00:32, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4124/7035 [00:46<00:33, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4133/7035 [00:46<00:32, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4142/7035 [00:47<00:32, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4152/7035 [00:47<00:32, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4161/7035 [00:47<00:32, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4170/7035 [00:47<00:32, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4179/7035 [00:47<00:32, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4188/7035 [00:47<00:32, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4197/7035 [00:47<00:32, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4206/7035 [00:47<00:32, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4215/7035 [00:47<00:32, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4224/7035 [00:47<00:31, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4233/7035 [00:48<00:31, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4242/7035 [00:48<00:31, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4251/7035 [00:48<00:31, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4260/7035 [00:48<00:31, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4269/7035 [00:48<00:31, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4278/7035 [00:48<00:31, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4287/7035 [00:48<00:31, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4296/7035 [00:48<00:31, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4305/7035 [00:48<00:30, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4314/7035 [00:49<00:30, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4323/7035 [00:49<00:30, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4332/7035 [00:49<00:30, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4341/7035 [00:49<00:30, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4350/7035 [00:49<00:30, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4359/7035 [00:49<00:30, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4368/7035 [00:49<00:30, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4377/7035 [00:49<00:29, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4386/7035 [00:49<00:29, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4395/7035 [00:49<00:29, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4404/7035 [00:50<00:29, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4413/7035 [00:50<00:29, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4422/7035 [00:50<00:29, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4431/7035 [00:50<00:29, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4440/7035 [00:50<00:29, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4449/7035 [00:50<00:29, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4458/7035 [00:50<00:29, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4467/7035 [00:50<00:29, 87.51batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4477/7035 [00:50<00:28, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4486/7035 [00:50<00:28, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4495/7035 [00:51<00:28, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4504/7035 [00:51<00:28, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4514/7035 [00:51<00:28, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4523/7035 [00:51<00:28, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4532/7035 [00:51<00:28, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4541/7035 [00:51<00:28, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4550/7035 [00:51<00:27, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4559/7035 [00:51<00:28, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4568/7035 [00:51<00:27, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4578/7035 [00:51<00:27, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4587/7035 [00:52<00:27, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4597/7035 [00:52<00:27, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4606/7035 [00:52<00:27, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4615/7035 [00:52<00:27, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4624/7035 [00:52<00:27, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4633/7035 [00:52<00:27, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4642/7035 [00:52<00:27, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4651/7035 [00:52<00:27, 87.60batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4660/7035 [00:52<00:27, 87.54batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4669/7035 [00:53<00:27, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4678/7035 [00:53<00:27, 86.60batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4687/7035 [00:53<00:26, 87.06batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4697/7035 [00:53<00:26, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4706/7035 [00:53<00:26, 86.98batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4715/7035 [00:53<00:26, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4724/7035 [00:53<00:26, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4733/7035 [00:53<00:26, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4742/7035 [00:53<00:26, 86.87batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4752/7035 [00:53<00:25, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4761/7035 [00:54<00:25, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4770/7035 [00:54<00:25, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4779/7035 [00:54<00:25, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4788/7035 [00:54<00:25, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4797/7035 [00:54<00:25, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4806/7035 [00:54<00:25, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4815/7035 [00:54<00:25, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4824/7035 [00:54<00:25, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4833/7035 [00:54<00:24, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4842/7035 [00:55<00:24, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4851/7035 [00:55<00:24, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4861/7035 [00:55<00:24, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4870/7035 [00:55<00:24, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4879/7035 [00:55<00:24, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4888/7035 [00:55<00:24, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4897/7035 [00:55<00:24, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4906/7035 [00:55<00:24, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4915/7035 [00:55<00:24, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4924/7035 [00:55<00:23, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4933/7035 [00:56<00:23, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4942/7035 [00:56<00:23, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4951/7035 [00:56<00:23, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4960/7035 [00:56<00:23, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4969/7035 [00:56<00:23, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4978/7035 [00:56<00:23, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4987/7035 [00:56<00:22, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4996/7035 [00:56<00:22, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5005/7035 [00:56<00:22, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5014/7035 [00:56<00:22, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5023/7035 [00:57<00:22, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5032/7035 [00:57<00:22, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5041/7035 [00:57<00:22, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5050/7035 [00:57<00:22, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5059/7035 [00:57<00:22, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5068/7035 [00:57<00:22, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5077/7035 [00:57<00:22, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5086/7035 [00:57<00:22, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5095/7035 [00:57<00:21, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5104/7035 [00:57<00:21, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5113/7035 [00:58<00:21, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5122/7035 [00:58<00:21, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5131/7035 [00:58<00:21, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5140/7035 [00:58<00:21, 87.12batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5150/7035 [00:58<00:21, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5159/7035 [00:58<00:21, 87.29batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5168/7035 [00:58<00:21, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5177/7035 [00:58<00:21, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5186/7035 [00:58<00:20, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5195/7035 [00:58<00:20, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5205/7035 [00:59<00:20, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5214/7035 [00:59<00:20, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5224/7035 [00:59<00:20, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5233/7035 [00:59<00:20, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5242/7035 [00:59<00:20, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5251/7035 [00:59<00:20, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5260/7035 [00:59<00:20, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5270/7035 [00:59<00:19, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5279/7035 [00:59<00:19, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5288/7035 [01:00<00:19, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5297/7035 [01:00<00:19, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5306/7035 [01:00<00:19, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5315/7035 [01:00<00:19, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5324/7035 [01:00<00:19, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5333/7035 [01:00<00:19, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5342/7035 [01:00<00:19, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5351/7035 [01:00<00:19, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5360/7035 [01:00<00:18, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5369/7035 [01:00<00:19, 86.32batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5378/7035 [01:01<00:19, 87.13batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5387/7035 [01:01<00:18, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5396/7035 [01:01<00:18, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5405/7035 [01:01<00:18, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5414/7035 [01:01<00:18, 87.02batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5423/7035 [01:01<00:18, 86.40batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5433/7035 [01:01<00:18, 87.60batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5442/7035 [01:01<00:18, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5452/7035 [01:01<00:17, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5461/7035 [01:01<00:17, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5470/7035 [01:02<00:17, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5480/7035 [01:02<00:17, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5489/7035 [01:02<00:17, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5499/7035 [01:02<00:17, 89.62batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5508/7035 [01:02<00:17, 89.66batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5517/7035 [01:02<00:16, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5526/7035 [01:02<00:16, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5535/7035 [01:02<00:16, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5544/7035 [01:02<00:16, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5553/7035 [01:03<00:16, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5562/7035 [01:03<00:16, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5571/7035 [01:03<00:16, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5580/7035 [01:03<00:16, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5589/7035 [01:03<00:16, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5598/7035 [01:03<00:16, 86.02batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5607/7035 [01:03<00:16, 86.53batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5616/7035 [01:03<00:16, 87.22batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5625/7035 [01:03<00:16, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5634/7035 [01:03<00:15, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5643/7035 [01:04<00:15, 87.60batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5653/7035 [01:04<00:15, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5662/7035 [01:04<00:15, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5671/7035 [01:04<00:15, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5680/7035 [01:04<00:15, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5689/7035 [01:04<00:15, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5699/7035 [01:04<00:14, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5708/7035 [01:04<00:14, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5717/7035 [01:04<00:14, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5726/7035 [01:04<00:14, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5735/7035 [01:05<00:14, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5745/7035 [01:05<00:14, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5755/7035 [01:05<00:14, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5764/7035 [01:05<00:14, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5774/7035 [01:05<00:14, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5783/7035 [01:05<00:14, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5792/7035 [01:05<00:13, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5801/7035 [01:05<00:13, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5810/7035 [01:05<00:13, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5819/7035 [01:06<00:13, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5828/7035 [01:06<00:13, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5837/7035 [01:06<00:13, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5847/7035 [01:06<00:13, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5856/7035 [01:06<00:13, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5866/7035 [01:06<00:13, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5875/7035 [01:06<00:13, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5884/7035 [01:06<00:13, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5893/7035 [01:06<00:12, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5902/7035 [01:06<00:12, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5911/7035 [01:07<00:12, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5920/7035 [01:07<00:12, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5929/7035 [01:07<00:12, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5938/7035 [01:07<00:12, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5947/7035 [01:07<00:12, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5956/7035 [01:07<00:12, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5965/7035 [01:07<00:12, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5974/7035 [01:07<00:12, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5983/7035 [01:07<00:11, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5993/7035 [01:07<00:11, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6002/7035 [01:08<00:11, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6011/7035 [01:08<00:11, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6020/7035 [01:08<00:11, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6030/7035 [01:08<00:11, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6039/7035 [01:08<00:11, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6048/7035 [01:08<00:11, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6057/7035 [01:08<00:11, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6066/7035 [01:08<00:10, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6075/7035 [01:08<00:10, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6084/7035 [01:09<00:10, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6093/7035 [01:09<00:10, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6103/7035 [01:09<00:10, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6112/7035 [01:09<00:10, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6121/7035 [01:09<00:10, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6130/7035 [01:09<00:10, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6139/7035 [01:09<00:10, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6148/7035 [01:09<00:10, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6157/7035 [01:09<00:10, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6166/7035 [01:09<00:09, 87.40batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6175/7035 [01:10<00:09, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6184/7035 [01:10<00:09, 87.27batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6193/7035 [01:10<00:09, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6203/7035 [01:10<00:09, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6213/7035 [01:10<00:09, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6222/7035 [01:10<00:09, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6231/7035 [01:10<00:09, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6241/7035 [01:10<00:08, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6250/7035 [01:10<00:08, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6259/7035 [01:11<00:08, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6268/7035 [01:11<00:08, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6277/7035 [01:11<00:08, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6286/7035 [01:11<00:08, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6295/7035 [01:11<00:08, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6304/7035 [01:11<00:08, 87.24batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6313/7035 [01:11<00:08, 87.54batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6322/7035 [01:11<00:08, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6332/7035 [01:11<00:07, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6341/7035 [01:11<00:07, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6351/7035 [01:12<00:07, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6360/7035 [01:12<00:07, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6369/7035 [01:12<00:07, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6378/7035 [01:12<00:07, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6387/7035 [01:12<00:07, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6397/7035 [01:12<00:07, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6406/7035 [01:12<00:07, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6415/7035 [01:12<00:07, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6424/7035 [01:12<00:06, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6433/7035 [01:12<00:06, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6442/7035 [01:13<00:06, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6451/7035 [01:13<00:06, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6460/7035 [01:13<00:06, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6469/7035 [01:13<00:06, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6478/7035 [01:13<00:06, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6487/7035 [01:13<00:06, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6496/7035 [01:13<00:06, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6505/7035 [01:13<00:06, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6514/7035 [01:13<00:05, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6523/7035 [01:13<00:05, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6532/7035 [01:14<00:05, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6541/7035 [01:14<00:05, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6550/7035 [01:14<00:05, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6560/7035 [01:14<00:05, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6569/7035 [01:14<00:05, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6578/7035 [01:14<00:05, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6587/7035 [01:14<00:05, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6596/7035 [01:14<00:04, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6605/7035 [01:14<00:04, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6615/7035 [01:15<00:04, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6624/7035 [01:15<00:04, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6633/7035 [01:15<00:04, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6643/7035 [01:15<00:04, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6652/7035 [01:15<00:04, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6662/7035 [01:15<00:04, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6672/7035 [01:15<00:04, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6681/7035 [01:15<00:03, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6690/7035 [01:15<00:03, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6699/7035 [01:15<00:03, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6708/7035 [01:16<00:03, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6718/7035 [01:16<00:03, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6727/7035 [01:16<00:03, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6736/7035 [01:16<00:03, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6746/7035 [01:16<00:03, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6755/7035 [01:16<00:03, 89.37batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6765/7035 [01:16<00:03, 89.48batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6774/7035 [01:16<00:02, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6783/7035 [01:16<00:02, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6792/7035 [01:17<00:02, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6801/7035 [01:17<00:02, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6810/7035 [01:17<00:02, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6819/7035 [01:17<00:02, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6828/7035 [01:17<00:02, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6837/7035 [01:17<00:02, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6846/7035 [01:17<00:02, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6855/7035 [01:17<00:02, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6864/7035 [01:17<00:01, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6873/7035 [01:17<00:01, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6882/7035 [01:18<00:01, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6891/7035 [01:18<00:01, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6900/7035 [01:18<00:01, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6909/7035 [01:18<00:01, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6918/7035 [01:18<00:01, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6927/7035 [01:18<00:01, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6937/7035 [01:18<00:01, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6946/7035 [01:18<00:00, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6955/7035 [01:18<00:00, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6964/7035 [01:18<00:00, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6973/7035 [01:19<00:00, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6982/7035 [01:19<00:00, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6991/7035 [01:19<00:00, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7000/7035 [01:19<00:00, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7009/7035 [01:19<00:00, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7019/7035 [01:19<00:00, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7029/7035 [01:19<00:00, 89.49batch/s]\u001b[A\n",
"Training epochs on cuda: 20%|██ | 1/5 [01:19<05:19, 79.84s/epoch, loss=0.00184, prev_loss=nan]\n",
"Training batches on cuda: 0%| | 0/7035 [00:00<?, ?batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 1/7035 [00:00<15:57, 7.35batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 10/7035 [00:00<11:33, 10.13batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 19/7035 [00:00<08:28, 13.79batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 28/7035 [00:00<06:19, 18.46batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 37/7035 [00:00<04:49, 24.18batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 46/7035 [00:00<03:45, 30.95batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 55/7035 [00:00<03:01, 38.47batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 64/7035 [00:00<02:30, 46.31batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 73/7035 [00:00<02:09, 53.94batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 82/7035 [00:01<01:53, 61.22batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 91/7035 [00:01<01:42, 67.51batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 100/7035 [00:01<01:35, 72.25batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 109/7035 [00:01<01:31, 75.77batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 118/7035 [00:01<01:27, 79.40batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 127/7035 [00:01<01:25, 80.90batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 136/7035 [00:01<01:23, 82.77batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 145/7035 [00:01<01:21, 84.66batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 154/7035 [00:01<01:20, 85.80batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 163/7035 [00:01<01:19, 86.20batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 172/7035 [00:02<01:18, 87.10batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 181/7035 [00:02<01:18, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 190/7035 [00:02<01:17, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 200/7035 [00:02<01:16, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 209/7035 [00:02<01:16, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 218/7035 [00:02<01:16, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 228/7035 [00:02<01:16, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 237/7035 [00:02<01:16, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 247/7035 [00:02<01:15, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 256/7035 [00:03<01:16, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 266/7035 [00:03<01:15, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 276/7035 [00:03<01:15, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 286/7035 [00:03<01:14, 90.02batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 296/7035 [00:03<01:14, 90.15batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 306/7035 [00:03<01:15, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 315/7035 [00:03<01:15, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 324/7035 [00:03<01:15, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 333/7035 [00:03<01:15, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 342/7035 [00:03<01:15, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 351/7035 [00:04<01:15, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 361/7035 [00:04<01:14, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 370/7035 [00:04<01:14, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 379/7035 [00:04<01:14, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 388/7035 [00:04<01:14, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 397/7035 [00:04<01:14, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 406/7035 [00:04<01:14, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 415/7035 [00:04<01:14, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 424/7035 [00:04<01:14, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 433/7035 [00:05<01:14, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 442/7035 [00:05<01:14, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 451/7035 [00:05<01:14, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 460/7035 [00:05<01:14, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 469/7035 [00:05<01:14, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 478/7035 [00:05<01:14, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 487/7035 [00:05<01:14, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 496/7035 [00:05<01:13, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 505/7035 [00:05<01:13, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 514/7035 [00:05<01:13, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 523/7035 [00:06<01:14, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 533/7035 [00:06<01:13, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 542/7035 [00:06<01:12, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 551/7035 [00:06<01:12, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 560/7035 [00:06<01:14, 87.31batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 569/7035 [00:06<01:13, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 578/7035 [00:06<01:13, 87.27batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 587/7035 [00:06<01:13, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 596/7035 [00:06<01:13, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 605/7035 [00:06<01:12, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 614/7035 [00:07<01:12, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 623/7035 [00:07<01:12, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 632/7035 [00:07<01:12, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 641/7035 [00:07<01:11, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 650/7035 [00:07<01:12, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 659/7035 [00:07<01:12, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 668/7035 [00:07<01:12, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 677/7035 [00:07<01:12, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 686/7035 [00:07<01:11, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 695/7035 [00:07<01:12, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 704/7035 [00:08<01:13, 86.68batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 713/7035 [00:08<01:12, 87.42batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 722/7035 [00:08<01:11, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 731/7035 [00:08<01:11, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 740/7035 [00:08<01:11, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 749/7035 [00:08<01:10, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 758/7035 [00:08<01:11, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 767/7035 [00:08<01:10, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 776/7035 [00:08<01:10, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 785/7035 [00:09<01:11, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 794/7035 [00:09<01:11, 86.83batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 803/7035 [00:09<01:11, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 812/7035 [00:09<01:10, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 821/7035 [00:09<01:10, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 830/7035 [00:09<01:10, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 839/7035 [00:09<01:10, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 848/7035 [00:09<01:10, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 857/7035 [00:09<01:10, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 866/7035 [00:09<01:09, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 875/7035 [00:10<01:09, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 884/7035 [00:10<01:09, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 893/7035 [00:10<01:09, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 902/7035 [00:10<01:08, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 911/7035 [00:10<01:09, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 920/7035 [00:10<01:08, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 930/7035 [00:10<01:08, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 939/7035 [00:10<01:08, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 948/7035 [00:10<01:08, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 957/7035 [00:10<01:08, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 966/7035 [00:11<01:08, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 975/7035 [00:11<01:08, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 984/7035 [00:11<01:08, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 993/7035 [00:11<01:08, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1002/7035 [00:11<01:07, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1011/7035 [00:11<01:07, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1020/7035 [00:11<01:07, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1029/7035 [00:11<01:08, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1038/7035 [00:11<01:07, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1047/7035 [00:11<01:08, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1056/7035 [00:12<01:07, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1065/7035 [00:12<01:08, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1074/7035 [00:12<01:07, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1084/7035 [00:12<01:07, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1093/7035 [00:12<01:06, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1102/7035 [00:12<01:06, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1111/7035 [00:12<01:07, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1120/7035 [00:12<01:07, 87.62batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1129/7035 [00:12<01:07, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1138/7035 [00:12<01:07, 87.14batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1148/7035 [00:13<01:06, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1157/7035 [00:13<01:06, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1167/7035 [00:13<01:05, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1176/7035 [00:13<01:05, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1185/7035 [00:13<01:05, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1195/7035 [00:13<01:05, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1204/7035 [00:13<01:05, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1213/7035 [00:13<01:05, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1222/7035 [00:13<01:05, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1231/7035 [00:14<01:05, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1240/7035 [00:14<01:05, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1249/7035 [00:14<01:05, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1258/7035 [00:14<01:05, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1267/7035 [00:14<01:05, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1276/7035 [00:14<01:05, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1286/7035 [00:14<01:04, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1295/7035 [00:14<01:04, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1304/7035 [00:14<01:05, 87.47batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1313/7035 [00:14<01:05, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1322/7035 [00:15<01:04, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1331/7035 [00:15<01:04, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1340/7035 [00:15<01:04, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1349/7035 [00:15<01:04, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1358/7035 [00:15<01:04, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1367/7035 [00:15<01:04, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1376/7035 [00:15<01:04, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1385/7035 [00:15<01:04, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1394/7035 [00:15<01:04, 87.42batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1403/7035 [00:15<01:04, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1412/7035 [00:16<01:03, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1421/7035 [00:16<01:03, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1430/7035 [00:16<01:04, 86.67batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1440/7035 [00:16<01:03, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1449/7035 [00:16<01:03, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1459/7035 [00:16<01:02, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1468/7035 [00:16<01:02, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1477/7035 [00:16<01:02, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1486/7035 [00:16<01:02, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1495/7035 [00:17<01:03, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1504/7035 [00:17<01:02, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1513/7035 [00:17<01:03, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1522/7035 [00:17<01:02, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1531/7035 [00:17<01:02, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1540/7035 [00:17<01:02, 87.33batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1549/7035 [00:17<01:02, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1558/7035 [00:17<01:02, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1567/7035 [00:17<01:02, 87.24batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1576/7035 [00:17<01:02, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1585/7035 [00:18<01:01, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1594/7035 [00:18<01:01, 87.90batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1603/7035 [00:18<01:01, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1612/7035 [00:18<01:01, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1621/7035 [00:18<01:00, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1630/7035 [00:18<01:00, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1639/7035 [00:18<01:00, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1648/7035 [00:18<01:00, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1657/7035 [00:18<01:00, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1666/7035 [00:18<01:00, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1675/7035 [00:19<01:00, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1684/7035 [00:19<01:00, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1693/7035 [00:19<01:00, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1703/7035 [00:19<01:00, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1712/7035 [00:19<01:00, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1722/7035 [00:19<00:59, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1731/7035 [00:19<00:59, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1740/7035 [00:19<00:59, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1749/7035 [00:19<01:00, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1758/7035 [00:20<00:59, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1767/7035 [00:20<00:59, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1776/7035 [00:20<00:59, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1785/7035 [00:20<00:59, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1794/7035 [00:20<00:59, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1804/7035 [00:20<00:58, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1813/7035 [00:20<00:58, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1822/7035 [00:20<00:58, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1831/7035 [00:20<00:58, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1841/7035 [00:20<00:58, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1850/7035 [00:21<00:58, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1860/7035 [00:21<00:57, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1869/7035 [00:21<00:59, 86.99batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1878/7035 [00:21<00:59, 86.92batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1888/7035 [00:21<00:58, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1897/7035 [00:21<00:58, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1906/7035 [00:21<00:58, 87.60batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1915/7035 [00:21<00:58, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1924/7035 [00:21<00:57, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1933/7035 [00:21<00:58, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1942/7035 [00:22<00:57, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1951/7035 [00:22<00:57, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1960/7035 [00:22<00:57, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1969/7035 [00:22<00:57, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1978/7035 [00:22<00:57, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1988/7035 [00:22<00:56, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1997/7035 [00:22<00:56, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2006/7035 [00:22<00:56, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2015/7035 [00:22<00:56, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2024/7035 [00:23<00:56, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2033/7035 [00:23<00:56, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2042/7035 [00:23<00:56, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2051/7035 [00:23<00:57, 87.26batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2060/7035 [00:23<00:56, 87.51batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2069/7035 [00:23<00:56, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2078/7035 [00:23<00:56, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2088/7035 [00:23<00:55, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2097/7035 [00:23<00:55, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2106/7035 [00:23<00:55, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2115/7035 [00:24<00:55, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2124/7035 [00:24<00:55, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2133/7035 [00:24<00:55, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2142/7035 [00:24<00:55, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2151/7035 [00:24<00:55, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2160/7035 [00:24<00:54, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2169/7035 [00:24<00:54, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2178/7035 [00:24<00:54, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2187/7035 [00:24<00:54, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2196/7035 [00:24<00:54, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2205/7035 [00:25<00:55, 86.67batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2214/7035 [00:25<00:55, 87.19batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2223/7035 [00:25<00:55, 86.36batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2232/7035 [00:25<00:54, 87.35batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2241/7035 [00:25<00:54, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2251/7035 [00:25<00:53, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2260/7035 [00:25<00:54, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2269/7035 [00:25<00:53, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2278/7035 [00:25<00:53, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2287/7035 [00:26<00:53, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2296/7035 [00:26<00:53, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2305/7035 [00:26<00:53, 87.59batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2314/7035 [00:26<00:53, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2323/7035 [00:26<00:54, 87.24batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2332/7035 [00:26<00:53, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2341/7035 [00:26<00:53, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2350/7035 [00:26<00:52, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2359/7035 [00:26<00:52, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2368/7035 [00:26<00:52, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2377/7035 [00:27<00:52, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2386/7035 [00:27<00:52, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2395/7035 [00:27<00:52, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2404/7035 [00:27<00:52, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2413/7035 [00:27<00:51, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2422/7035 [00:27<00:52, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2431/7035 [00:27<00:52, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2440/7035 [00:27<00:51, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2449/7035 [00:27<00:51, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2458/7035 [00:27<00:51, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2467/7035 [00:28<00:51, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2476/7035 [00:28<00:51, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2485/7035 [00:28<00:51, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2494/7035 [00:28<00:52, 86.22batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2503/7035 [00:28<00:52, 86.97batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2512/7035 [00:28<00:51, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2521/7035 [00:28<00:51, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2530/7035 [00:28<00:50, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2539/7035 [00:28<00:50, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2548/7035 [00:28<00:50, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2557/7035 [00:29<00:50, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2566/7035 [00:29<00:50, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2575/7035 [00:29<00:50, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2584/7035 [00:29<00:50, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2593/7035 [00:29<00:50, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2602/7035 [00:29<00:50, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2611/7035 [00:29<00:49, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2620/7035 [00:29<00:49, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2629/7035 [00:29<00:49, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2638/7035 [00:29<00:49, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2647/7035 [00:30<00:49, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2656/7035 [00:30<00:49, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2665/7035 [00:30<00:49, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2674/7035 [00:30<00:49, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2683/7035 [00:30<00:49, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2692/7035 [00:30<00:49, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2701/7035 [00:30<00:48, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2710/7035 [00:30<00:48, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2719/7035 [00:30<00:48, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2728/7035 [00:30<00:48, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2737/7035 [00:31<00:48, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2746/7035 [00:31<00:48, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2755/7035 [00:31<00:48, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2764/7035 [00:31<00:48, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2773/7035 [00:31<00:48, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2782/7035 [00:31<00:47, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2791/7035 [00:31<00:47, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2800/7035 [00:31<00:47, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2809/7035 [00:31<00:47, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2818/7035 [00:32<00:47, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2827/7035 [00:32<00:48, 87.39batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2836/7035 [00:32<00:47, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2845/7035 [00:32<00:47, 87.59batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2854/7035 [00:32<00:47, 87.26batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2863/7035 [00:32<00:47, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2872/7035 [00:32<00:47, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2881/7035 [00:32<00:47, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2890/7035 [00:32<00:46, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2899/7035 [00:32<00:46, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2908/7035 [00:33<00:46, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2917/7035 [00:33<00:46, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2926/7035 [00:33<00:46, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2935/7035 [00:33<00:46, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2945/7035 [00:33<00:45, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2954/7035 [00:33<00:46, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2963/7035 [00:33<00:46, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2972/7035 [00:33<00:46, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2981/7035 [00:33<00:46, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2990/7035 [00:33<00:45, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2999/7035 [00:34<00:45, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3008/7035 [00:34<00:45, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3017/7035 [00:34<00:45, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3026/7035 [00:34<00:45, 87.72batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3035/7035 [00:34<00:45, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3044/7035 [00:34<00:45, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3053/7035 [00:34<00:44, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3062/7035 [00:34<00:44, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3071/7035 [00:34<00:44, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3080/7035 [00:34<00:44, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3089/7035 [00:35<00:44, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3098/7035 [00:35<00:44, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3108/7035 [00:35<00:44, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3117/7035 [00:35<00:44, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3127/7035 [00:35<00:43, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3137/7035 [00:35<00:43, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3146/7035 [00:35<00:43, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3155/7035 [00:35<00:43, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3164/7035 [00:35<00:43, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3173/7035 [00:36<00:43, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3183/7035 [00:36<00:42, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3192/7035 [00:36<00:43, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3201/7035 [00:36<00:43, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3210/7035 [00:36<00:43, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3219/7035 [00:36<00:42, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3229/7035 [00:36<00:42, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3238/7035 [00:36<00:42, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3247/7035 [00:36<00:42, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3256/7035 [00:36<00:42, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3265/7035 [00:37<00:42, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3274/7035 [00:37<00:42, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3283/7035 [00:37<00:42, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3293/7035 [00:37<00:42, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3302/7035 [00:37<00:42, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3311/7035 [00:37<00:41, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3320/7035 [00:37<00:41, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3329/7035 [00:37<00:42, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3338/7035 [00:37<00:42, 87.01batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3347/7035 [00:37<00:42, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3356/7035 [00:38<00:41, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3365/7035 [00:38<00:41, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3374/7035 [00:38<00:41, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3383/7035 [00:38<00:41, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3392/7035 [00:38<00:41, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3401/7035 [00:38<00:42, 85.75batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3410/7035 [00:38<00:41, 86.74batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3419/7035 [00:38<00:41, 87.00batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3428/7035 [00:38<00:41, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3437/7035 [00:39<00:40, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3446/7035 [00:39<00:40, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3455/7035 [00:39<00:40, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3464/7035 [00:39<00:40, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3473/7035 [00:39<00:40, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3482/7035 [00:39<00:40, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3491/7035 [00:39<00:40, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3500/7035 [00:39<00:40, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3509/7035 [00:39<00:39, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3518/7035 [00:39<00:39, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3527/7035 [00:40<00:39, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3536/7035 [00:40<00:39, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3545/7035 [00:40<00:39, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3554/7035 [00:40<00:39, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3563/7035 [00:40<00:39, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3572/7035 [00:40<00:39, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3581/7035 [00:40<00:38, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3590/7035 [00:40<00:38, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3599/7035 [00:40<00:38, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3608/7035 [00:40<00:38, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3617/7035 [00:41<00:38, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3626/7035 [00:41<00:38, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3635/7035 [00:41<00:38, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3644/7035 [00:41<00:38, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3653/7035 [00:41<00:38, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3662/7035 [00:41<00:38, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3671/7035 [00:41<00:38, 87.31batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3680/7035 [00:41<00:38, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3689/7035 [00:41<00:37, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3698/7035 [00:41<00:37, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3707/7035 [00:42<00:37, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3716/7035 [00:42<00:37, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3725/7035 [00:42<00:37, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3734/7035 [00:42<00:37, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3743/7035 [00:42<00:37, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3752/7035 [00:42<00:37, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3761/7035 [00:42<00:37, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3770/7035 [00:42<00:36, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3779/7035 [00:42<00:37, 87.12batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3788/7035 [00:43<00:37, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3797/7035 [00:43<00:36, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3806/7035 [00:43<00:36, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3815/7035 [00:43<00:36, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3824/7035 [00:43<00:36, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3833/7035 [00:43<00:36, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3842/7035 [00:43<00:36, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3851/7035 [00:43<00:36, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3860/7035 [00:43<00:35, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3869/7035 [00:43<00:35, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3878/7035 [00:44<00:35, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3887/7035 [00:44<00:35, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3896/7035 [00:44<00:35, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3905/7035 [00:44<00:35, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3914/7035 [00:44<00:35, 87.40batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3923/7035 [00:44<00:35, 86.80batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3932/7035 [00:44<00:35, 87.36batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3941/7035 [00:44<00:35, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3950/7035 [00:44<00:35, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3959/7035 [00:44<00:34, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3968/7035 [00:45<00:34, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3977/7035 [00:45<00:34, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3986/7035 [00:45<00:34, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3995/7035 [00:45<00:34, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4004/7035 [00:45<00:34, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4013/7035 [00:45<00:34, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4022/7035 [00:45<00:34, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4031/7035 [00:45<00:33, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4040/7035 [00:45<00:33, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4049/7035 [00:45<00:33, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4059/7035 [00:46<00:33, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4068/7035 [00:46<00:33, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4077/7035 [00:46<00:33, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4086/7035 [00:46<00:33, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4095/7035 [00:46<00:33, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4104/7035 [00:46<00:33, 87.11batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4113/7035 [00:46<00:33, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4122/7035 [00:46<00:33, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4131/7035 [00:46<00:32, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4140/7035 [00:46<00:32, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4149/7035 [00:47<00:32, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4158/7035 [00:47<00:32, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4167/7035 [00:47<00:32, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4176/7035 [00:47<00:32, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4185/7035 [00:47<00:32, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4194/7035 [00:47<00:31, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4203/7035 [00:47<00:32, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4212/7035 [00:47<00:31, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4221/7035 [00:47<00:31, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4231/7035 [00:48<00:31, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4241/7035 [00:48<00:31, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4251/7035 [00:48<00:30, 89.90batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4260/7035 [00:48<00:31, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4269/7035 [00:48<00:31, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4278/7035 [00:48<00:31, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4287/7035 [00:48<00:31, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4296/7035 [00:48<00:30, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4305/7035 [00:48<00:30, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4314/7035 [00:48<00:30, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4323/7035 [00:49<00:30, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4332/7035 [00:49<00:30, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4341/7035 [00:49<00:30, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4350/7035 [00:49<00:30, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4359/7035 [00:49<00:30, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4368/7035 [00:49<00:30, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4378/7035 [00:49<00:29, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4387/7035 [00:49<00:30, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4396/7035 [00:49<00:29, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4405/7035 [00:49<00:29, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4414/7035 [00:50<00:29, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4423/7035 [00:50<00:29, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4432/7035 [00:50<00:29, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4441/7035 [00:50<00:29, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4450/7035 [00:50<00:29, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4459/7035 [00:50<00:29, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4469/7035 [00:50<00:28, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4478/7035 [00:50<00:28, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4487/7035 [00:50<00:28, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4496/7035 [00:51<00:28, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4505/7035 [00:51<00:28, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4515/7035 [00:51<00:28, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4525/7035 [00:51<00:28, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4534/7035 [00:51<00:27, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4543/7035 [00:51<00:27, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4552/7035 [00:51<00:28, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4561/7035 [00:51<00:28, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4570/7035 [00:51<00:27, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4579/7035 [00:51<00:27, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4588/7035 [00:52<00:27, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4597/7035 [00:52<00:27, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4606/7035 [00:52<00:27, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4615/7035 [00:52<00:27, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4625/7035 [00:52<00:26, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4634/7035 [00:52<00:27, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4643/7035 [00:52<00:26, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4653/7035 [00:52<00:26, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4663/7035 [00:52<00:26, 89.87batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4672/7035 [00:52<00:26, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4682/7035 [00:53<00:26, 90.12batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4692/7035 [00:53<00:25, 90.52batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4702/7035 [00:53<00:25, 90.21batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4712/7035 [00:53<00:25, 90.48batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4722/7035 [00:53<00:25, 90.06batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4732/7035 [00:53<00:25, 90.25batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4742/7035 [00:53<00:25, 90.39batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4752/7035 [00:53<00:25, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4762/7035 [00:53<00:25, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4772/7035 [00:54<00:25, 89.88batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4781/7035 [00:54<00:25, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4790/7035 [00:54<00:25, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4799/7035 [00:54<00:25, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4808/7035 [00:54<00:24, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4817/7035 [00:54<00:25, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4826/7035 [00:54<00:24, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4835/7035 [00:54<00:24, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4844/7035 [00:54<00:24, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4853/7035 [00:55<00:24, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4862/7035 [00:55<00:24, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4871/7035 [00:55<00:24, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4880/7035 [00:55<00:24, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4890/7035 [00:55<00:24, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4899/7035 [00:55<00:24, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4908/7035 [00:55<00:24, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4917/7035 [00:55<00:24, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4926/7035 [00:55<00:23, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4935/7035 [00:55<00:23, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4944/7035 [00:56<00:23, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4953/7035 [00:56<00:23, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4962/7035 [00:56<00:23, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4971/7035 [00:56<00:23, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4980/7035 [00:56<00:23, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4989/7035 [00:56<00:23, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4998/7035 [00:56<00:23, 87.45batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5007/7035 [00:56<00:23, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5016/7035 [00:56<00:22, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5025/7035 [00:56<00:22, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5034/7035 [00:57<00:22, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5043/7035 [00:57<00:22, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5052/7035 [00:57<00:22, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5061/7035 [00:57<00:22, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5070/7035 [00:57<00:22, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5079/7035 [00:57<00:22, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5088/7035 [00:57<00:22, 87.35batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5097/7035 [00:57<00:22, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5106/7035 [00:57<00:21, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5115/7035 [00:57<00:21, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5124/7035 [00:58<00:21, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5133/7035 [00:58<00:21, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5142/7035 [00:58<00:21, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5151/7035 [00:58<00:21, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5160/7035 [00:58<00:21, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5169/7035 [00:58<00:21, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5178/7035 [00:58<00:21, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5187/7035 [00:58<00:20, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5196/7035 [00:58<00:21, 87.48batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5205/7035 [00:58<00:20, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5214/7035 [00:59<00:20, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5223/7035 [00:59<00:20, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5232/7035 [00:59<00:20, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5241/7035 [00:59<00:20, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5250/7035 [00:59<00:20, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5259/7035 [00:59<00:20, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5268/7035 [00:59<00:20, 85.36batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5277/7035 [00:59<00:20, 86.40batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5286/7035 [00:59<00:20, 87.11batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5295/7035 [01:00<00:20, 86.93batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5304/7035 [01:00<00:19, 87.14batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5313/7035 [01:00<00:19, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5322/7035 [01:00<00:19, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5331/7035 [01:00<00:19, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5340/7035 [01:00<00:19, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5349/7035 [01:00<00:19, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5358/7035 [01:00<00:19, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5368/7035 [01:00<00:18, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5378/7035 [01:00<00:18, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5387/7035 [01:01<00:18, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5396/7035 [01:01<00:18, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5405/7035 [01:01<00:18, 89.64batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5414/7035 [01:01<00:18, 89.73batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5423/7035 [01:01<00:17, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5433/7035 [01:01<00:17, 90.06batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5443/7035 [01:01<00:17, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5452/7035 [01:01<00:17, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5462/7035 [01:01<00:17, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5472/7035 [01:02<00:17, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5482/7035 [01:02<00:17, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5492/7035 [01:02<00:17, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5501/7035 [01:02<00:17, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5510/7035 [01:02<00:17, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5520/7035 [01:02<00:16, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5529/7035 [01:02<00:16, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5538/7035 [01:02<00:16, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5547/7035 [01:02<00:16, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5557/7035 [01:02<00:16, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5567/7035 [01:03<00:16, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5577/7035 [01:03<00:16, 89.77batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5586/7035 [01:03<00:16, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5595/7035 [01:03<00:16, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5605/7035 [01:03<00:15, 90.05batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5615/7035 [01:03<00:15, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5624/7035 [01:03<00:15, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5633/7035 [01:03<00:15, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5643/7035 [01:03<00:15, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5652/7035 [01:04<00:15, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5661/7035 [01:04<00:15, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5670/7035 [01:04<00:15, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5679/7035 [01:04<00:15, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5688/7035 [01:04<00:15, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5697/7035 [01:04<00:15, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5706/7035 [01:04<00:14, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5715/7035 [01:04<00:14, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5724/7035 [01:04<00:14, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5734/7035 [01:04<00:14, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5744/7035 [01:05<00:14, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5753/7035 [01:05<00:14, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5762/7035 [01:05<00:14, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5771/7035 [01:05<00:14, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5780/7035 [01:05<00:14, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5789/7035 [01:05<00:13, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5798/7035 [01:05<00:13, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5807/7035 [01:05<00:13, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5816/7035 [01:05<00:13, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5825/7035 [01:05<00:13, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5834/7035 [01:06<00:13, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5843/7035 [01:06<00:13, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5852/7035 [01:06<00:13, 87.47batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5861/7035 [01:06<00:13, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5870/7035 [01:06<00:13, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5879/7035 [01:06<00:13, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5888/7035 [01:06<00:12, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5897/7035 [01:06<00:12, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5906/7035 [01:06<00:12, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5915/7035 [01:06<00:12, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5924/7035 [01:07<00:12, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5933/7035 [01:07<00:12, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5942/7035 [01:07<00:12, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5951/7035 [01:07<00:12, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5960/7035 [01:07<00:12, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5969/7035 [01:07<00:12, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5978/7035 [01:07<00:11, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5987/7035 [01:07<00:11, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5996/7035 [01:07<00:11, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6006/7035 [01:08<00:11, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6015/7035 [01:08<00:11, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6024/7035 [01:08<00:11, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6033/7035 [01:08<00:11, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6042/7035 [01:08<00:11, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6051/7035 [01:08<00:11, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6060/7035 [01:08<00:10, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6069/7035 [01:08<00:10, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6078/7035 [01:08<00:10, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6087/7035 [01:08<00:10, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6096/7035 [01:09<00:10, 86.92batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6105/7035 [01:09<00:10, 86.67batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6114/7035 [01:09<00:10, 87.59batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6123/7035 [01:09<00:10, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6132/7035 [01:09<00:10, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6141/7035 [01:09<00:10, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6150/7035 [01:09<00:09, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6159/7035 [01:09<00:09, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6168/7035 [01:09<00:09, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6177/7035 [01:09<00:09, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6186/7035 [01:10<00:09, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6195/7035 [01:10<00:09, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6204/7035 [01:10<00:09, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6213/7035 [01:10<00:09, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6222/7035 [01:10<00:09, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6231/7035 [01:10<00:09, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6240/7035 [01:10<00:08, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6250/7035 [01:10<00:08, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6259/7035 [01:10<00:08, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6268/7035 [01:10<00:08, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6278/7035 [01:11<00:08, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6288/7035 [01:11<00:08, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6297/7035 [01:11<00:08, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6307/7035 [01:11<00:08, 89.80batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6316/7035 [01:11<00:08, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6325/7035 [01:11<00:07, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6334/7035 [01:11<00:07, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6343/7035 [01:11<00:07, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6352/7035 [01:11<00:07, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6361/7035 [01:12<00:07, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6370/7035 [01:12<00:07, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6379/7035 [01:12<00:07, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6389/7035 [01:12<00:07, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6398/7035 [01:12<00:07, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6407/7035 [01:12<00:07, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6416/7035 [01:12<00:06, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6425/7035 [01:12<00:06, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6434/7035 [01:12<00:06, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6443/7035 [01:12<00:06, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6452/7035 [01:13<00:06, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6462/7035 [01:13<00:06, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6472/7035 [01:13<00:06, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6481/7035 [01:13<00:06, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6491/7035 [01:13<00:06, 89.76batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6500/7035 [01:13<00:05, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6510/7035 [01:13<00:05, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6519/7035 [01:13<00:05, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6528/7035 [01:13<00:05, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6537/7035 [01:13<00:05, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6547/7035 [01:14<00:05, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6556/7035 [01:14<00:05, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6565/7035 [01:14<00:05, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6574/7035 [01:14<00:05, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6583/7035 [01:14<00:05, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6592/7035 [01:14<00:04, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6602/7035 [01:14<00:04, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6611/7035 [01:14<00:04, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6620/7035 [01:14<00:04, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6629/7035 [01:15<00:04, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6638/7035 [01:15<00:04, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6647/7035 [01:15<00:04, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6656/7035 [01:15<00:04, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6665/7035 [01:15<00:04, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6674/7035 [01:15<00:04, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6683/7035 [01:15<00:03, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6692/7035 [01:15<00:03, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6702/7035 [01:15<00:03, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6711/7035 [01:15<00:03, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6720/7035 [01:16<00:03, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6729/7035 [01:16<00:03, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6738/7035 [01:16<00:03, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6747/7035 [01:16<00:03, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6756/7035 [01:16<00:03, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6765/7035 [01:16<00:03, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6774/7035 [01:16<00:02, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6783/7035 [01:16<00:02, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6792/7035 [01:16<00:02, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6801/7035 [01:16<00:02, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6810/7035 [01:17<00:02, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6819/7035 [01:17<00:02, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6828/7035 [01:17<00:02, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6837/7035 [01:17<00:02, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6846/7035 [01:17<00:02, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6855/7035 [01:17<00:02, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6864/7035 [01:17<00:01, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6873/7035 [01:17<00:01, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6882/7035 [01:17<00:01, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6891/7035 [01:17<00:01, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6900/7035 [01:18<00:01, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6909/7035 [01:18<00:01, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6918/7035 [01:18<00:01, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6928/7035 [01:18<00:01, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6937/7035 [01:18<00:01, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6946/7035 [01:18<00:00, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6955/7035 [01:18<00:00, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6964/7035 [01:18<00:00, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6973/7035 [01:18<00:00, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6982/7035 [01:19<00:00, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6991/7035 [01:19<00:00, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7001/7035 [01:19<00:00, 89.37batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7010/7035 [01:19<00:00, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7020/7035 [01:19<00:00, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7030/7035 [01:19<00:00, 89.99batch/s]\u001b[A\n",
"Training epochs on cuda: 40%|████ | 2/5 [02:39<03:59, 79.80s/epoch, loss=0.00143, prev_loss=nan]\n",
"Training batches on cuda: 0%| | 0/7035 [00:00<?, ?batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 1/7035 [00:00<16:21, 7.16batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 10/7035 [00:00<11:50, 9.89batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 19/7035 [00:00<08:41, 13.46batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 28/7035 [00:00<06:29, 18.01batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 37/7035 [00:00<04:55, 23.66batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 46/7035 [00:00<03:50, 30.33batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 55/7035 [00:00<03:04, 37.83batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 64/7035 [00:00<02:32, 45.72batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 73/7035 [00:00<02:09, 53.58batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 82/7035 [00:01<01:54, 60.90batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 91/7035 [00:01<01:43, 66.86batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 100/7035 [00:01<01:36, 71.96batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 109/7035 [00:01<01:30, 76.47batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 118/7035 [00:01<01:26, 79.91batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 127/7035 [00:01<01:24, 82.19batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 136/7035 [00:01<01:22, 83.96batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 145/7035 [00:01<01:20, 85.50batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 154/7035 [00:01<01:19, 86.67batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 163/7035 [00:01<01:19, 86.77batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 172/7035 [00:02<01:18, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 181/7035 [00:02<01:17, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 190/7035 [00:02<01:17, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 199/7035 [00:02<01:17, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 208/7035 [00:02<01:17, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 217/7035 [00:02<01:17, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 226/7035 [00:02<01:16, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 235/7035 [00:02<01:16, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 244/7035 [00:02<01:16, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 253/7035 [00:02<01:16, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 262/7035 [00:03<01:16, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 271/7035 [00:03<01:16, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 280/7035 [00:03<01:16, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 289/7035 [00:03<01:16, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 298/7035 [00:03<01:16, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 308/7035 [00:03<01:15, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 317/7035 [00:03<01:15, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 326/7035 [00:03<01:15, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 336/7035 [00:03<01:15, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 345/7035 [00:04<01:15, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 354/7035 [00:04<01:14, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 363/7035 [00:04<01:14, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 372/7035 [00:04<01:15, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 381/7035 [00:04<01:15, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 390/7035 [00:04<01:15, 87.62batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 399/7035 [00:04<01:15, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 408/7035 [00:04<01:14, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 417/7035 [00:04<01:14, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 426/7035 [00:04<01:14, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 435/7035 [00:05<01:14, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 444/7035 [00:05<01:14, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 453/7035 [00:05<01:14, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 462/7035 [00:05<01:15, 87.31batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 471/7035 [00:05<01:15, 87.32batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 480/7035 [00:05<01:14, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 489/7035 [00:05<01:14, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 498/7035 [00:05<01:14, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 507/7035 [00:05<01:13, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 516/7035 [00:05<01:13, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 525/7035 [00:06<01:14, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 534/7035 [00:06<01:13, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 543/7035 [00:06<01:14, 87.35batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 552/7035 [00:06<01:14, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 562/7035 [00:06<01:13, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 571/7035 [00:06<01:14, 86.85batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 580/7035 [00:06<01:14, 87.16batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 590/7035 [00:06<01:13, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 599/7035 [00:06<01:12, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 609/7035 [00:07<01:12, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 618/7035 [00:07<01:11, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 628/7035 [00:07<01:11, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 637/7035 [00:07<01:11, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 646/7035 [00:07<01:11, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 655/7035 [00:07<01:12, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 664/7035 [00:07<01:11, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 674/7035 [00:07<01:11, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 684/7035 [00:07<01:10, 89.80batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 693/7035 [00:07<01:10, 89.77batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 702/7035 [00:08<01:10, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 711/7035 [00:08<01:12, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 720/7035 [00:08<01:11, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 729/7035 [00:08<01:11, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 738/7035 [00:08<01:11, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 747/7035 [00:08<01:12, 87.16batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 756/7035 [00:08<01:11, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 766/7035 [00:08<01:10, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 776/7035 [00:08<01:10, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 786/7035 [00:09<01:09, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 796/7035 [00:09<01:09, 89.98batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 806/7035 [00:09<01:09, 90.25batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 816/7035 [00:09<01:08, 90.15batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 826/7035 [00:09<01:08, 90.18batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 836/7035 [00:09<01:09, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 845/7035 [00:09<01:09, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 854/7035 [00:09<01:09, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 863/7035 [00:09<01:09, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 872/7035 [00:09<01:09, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 881/7035 [00:10<01:09, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 890/7035 [00:10<01:09, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 899/7035 [00:10<01:09, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 908/7035 [00:10<01:09, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 917/7035 [00:10<01:10, 87.20batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 926/7035 [00:10<01:09, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 935/7035 [00:10<01:09, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 944/7035 [00:10<01:08, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 953/7035 [00:10<01:08, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 963/7035 [00:11<01:08, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 972/7035 [00:11<01:08, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 981/7035 [00:11<01:08, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 991/7035 [00:11<01:07, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1000/7035 [00:11<01:07, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1009/7035 [00:11<01:07, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1018/7035 [00:11<01:07, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1027/7035 [00:11<01:08, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1036/7035 [00:11<01:07, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1045/7035 [00:11<01:07, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1054/7035 [00:12<01:07, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1063/7035 [00:12<01:07, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1072/7035 [00:12<01:06, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1081/7035 [00:12<01:06, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1090/7035 [00:12<01:06, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1099/7035 [00:12<01:07, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1108/7035 [00:12<01:07, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1117/7035 [00:12<01:06, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1126/7035 [00:12<01:06, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1135/7035 [00:12<01:06, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1144/7035 [00:13<01:06, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1153/7035 [00:13<01:06, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1162/7035 [00:13<01:06, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1171/7035 [00:13<01:06, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1180/7035 [00:13<01:06, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1189/7035 [00:13<01:05, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1198/7035 [00:13<01:05, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1207/7035 [00:13<01:05, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1216/7035 [00:13<01:05, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1225/7035 [00:13<01:05, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1234/7035 [00:14<01:05, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1243/7035 [00:14<01:05, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1252/7035 [00:14<01:04, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1262/7035 [00:14<01:04, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1271/7035 [00:14<01:05, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1280/7035 [00:14<01:04, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1289/7035 [00:14<01:04, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1298/7035 [00:14<01:04, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1307/7035 [00:14<01:05, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1316/7035 [00:14<01:04, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1325/7035 [00:15<01:04, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1334/7035 [00:15<01:03, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1343/7035 [00:15<01:04, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1352/7035 [00:15<01:04, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1361/7035 [00:15<01:04, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1370/7035 [00:15<01:04, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1379/7035 [00:15<01:03, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1389/7035 [00:15<01:03, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1398/7035 [00:15<01:03, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1407/7035 [00:16<01:03, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1416/7035 [00:16<01:03, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1425/7035 [00:16<01:02, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1434/7035 [00:16<01:02, 89.37batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1443/7035 [00:16<01:03, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1452/7035 [00:16<01:03, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1461/7035 [00:16<01:02, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1470/7035 [00:16<01:02, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1479/7035 [00:16<01:02, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1488/7035 [00:16<01:02, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1498/7035 [00:17<01:02, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1507/7035 [00:17<01:02, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1516/7035 [00:17<01:01, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1525/7035 [00:17<01:01, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1534/7035 [00:17<01:02, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1543/7035 [00:17<01:02, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1552/7035 [00:17<01:02, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1561/7035 [00:17<01:02, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1571/7035 [00:17<01:01, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1580/7035 [00:17<01:01, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1589/7035 [00:18<01:01, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1598/7035 [00:18<01:01, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1607/7035 [00:18<01:01, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1616/7035 [00:18<01:01, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1625/7035 [00:18<01:01, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1634/7035 [00:18<01:01, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1643/7035 [00:18<01:00, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1652/7035 [00:18<01:00, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1661/7035 [00:18<01:00, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1670/7035 [00:18<01:00, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1679/7035 [00:19<01:00, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1688/7035 [00:19<01:00, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1697/7035 [00:19<01:00, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1706/7035 [00:19<01:00, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1715/7035 [00:19<01:00, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1724/7035 [00:19<01:00, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1733/7035 [00:19<01:00, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1742/7035 [00:19<01:00, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1751/7035 [00:19<00:59, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1760/7035 [00:19<00:59, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1769/7035 [00:20<00:59, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1778/7035 [00:20<00:59, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1788/7035 [00:20<00:58, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1797/7035 [00:20<00:59, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1806/7035 [00:20<00:59, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1816/7035 [00:20<00:58, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1825/7035 [00:20<00:58, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1835/7035 [00:20<00:58, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1844/7035 [00:20<00:58, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1853/7035 [00:21<00:58, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1862/7035 [00:21<00:58, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1871/7035 [00:21<00:58, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1880/7035 [00:21<00:58, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1889/7035 [00:21<00:58, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1898/7035 [00:21<00:57, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1907/7035 [00:21<00:57, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1916/7035 [00:21<00:57, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1925/7035 [00:21<00:57, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1934/7035 [00:21<00:57, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1943/7035 [00:22<00:57, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1952/7035 [00:22<00:57, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1961/7035 [00:22<00:57, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1970/7035 [00:22<00:57, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1979/7035 [00:22<00:56, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1988/7035 [00:22<00:57, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1997/7035 [00:22<00:56, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2006/7035 [00:22<00:56, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2015/7035 [00:22<00:56, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2024/7035 [00:22<00:56, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2033/7035 [00:23<00:56, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2043/7035 [00:23<00:56, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2053/7035 [00:23<00:55, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2062/7035 [00:23<00:55, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2071/7035 [00:23<00:55, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2081/7035 [00:23<00:55, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2090/7035 [00:23<00:55, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2100/7035 [00:23<00:54, 90.01batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2110/7035 [00:23<00:54, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2119/7035 [00:24<00:54, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2129/7035 [00:24<00:54, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2138/7035 [00:24<00:54, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2147/7035 [00:24<00:54, 89.73batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2156/7035 [00:24<00:54, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2165/7035 [00:24<00:54, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2174/7035 [00:24<00:55, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2183/7035 [00:24<00:54, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2192/7035 [00:24<00:54, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2201/7035 [00:24<00:55, 86.57batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2210/7035 [00:25<00:55, 86.87batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2219/7035 [00:25<00:54, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2228/7035 [00:25<00:54, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2238/7035 [00:25<00:53, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2247/7035 [00:25<00:53, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2256/7035 [00:25<00:53, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2265/7035 [00:25<00:53, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2274/7035 [00:25<00:53, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2283/7035 [00:25<00:53, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2292/7035 [00:25<00:53, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2301/7035 [00:26<00:53, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2310/7035 [00:26<00:53, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2319/7035 [00:26<00:53, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2328/7035 [00:26<00:52, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2337/7035 [00:26<00:52, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2346/7035 [00:26<00:52, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2355/7035 [00:26<00:53, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2364/7035 [00:26<00:52, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2373/7035 [00:26<00:53, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2382/7035 [00:27<00:53, 87.15batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2391/7035 [00:27<00:53, 86.96batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2400/7035 [00:27<00:53, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2409/7035 [00:27<00:52, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2418/7035 [00:27<00:52, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2428/7035 [00:27<00:51, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2437/7035 [00:27<00:52, 87.36batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2446/7035 [00:27<00:52, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2455/7035 [00:27<00:52, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2464/7035 [00:27<00:52, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2473/7035 [00:28<00:51, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2482/7035 [00:28<00:51, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2491/7035 [00:28<00:51, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2500/7035 [00:28<00:51, 87.86batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2509/7035 [00:28<00:51, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2518/7035 [00:28<00:51, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2527/7035 [00:28<00:52, 86.55batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2536/7035 [00:28<00:51, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2545/7035 [00:28<00:51, 87.30batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2554/7035 [00:28<00:51, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2563/7035 [00:29<00:50, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2572/7035 [00:29<00:50, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2581/7035 [00:29<00:50, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2590/7035 [00:29<00:50, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2599/7035 [00:29<00:49, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2608/7035 [00:29<00:50, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2617/7035 [00:29<00:50, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2626/7035 [00:29<00:50, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2635/7035 [00:29<00:50, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2644/7035 [00:29<00:50, 87.11batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2653/7035 [00:30<00:50, 87.26batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2662/7035 [00:30<00:50, 86.73batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2671/7035 [00:30<00:50, 87.23batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2680/7035 [00:30<00:49, 87.51batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2689/7035 [00:30<00:50, 86.35batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2698/7035 [00:30<00:50, 86.71batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2707/7035 [00:30<00:49, 87.01batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2716/7035 [00:30<00:49, 87.32batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2725/7035 [00:30<00:49, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2734/7035 [00:31<00:48, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2743/7035 [00:31<00:48, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2752/7035 [00:31<00:48, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2761/7035 [00:31<00:48, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2770/7035 [00:31<00:48, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2779/7035 [00:31<00:48, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2788/7035 [00:31<00:48, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2797/7035 [00:31<00:48, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2806/7035 [00:31<00:47, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2815/7035 [00:31<00:47, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2824/7035 [00:32<00:47, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2833/7035 [00:32<00:47, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2842/7035 [00:32<00:47, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2851/7035 [00:32<00:47, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2860/7035 [00:32<00:47, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2869/7035 [00:32<00:46, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2878/7035 [00:32<00:46, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2887/7035 [00:32<00:46, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2896/7035 [00:32<00:46, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2905/7035 [00:32<00:46, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2914/7035 [00:33<00:46, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2924/7035 [00:33<00:46, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2933/7035 [00:33<00:45, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2942/7035 [00:33<00:46, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2951/7035 [00:33<00:45, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2960/7035 [00:33<00:45, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2969/7035 [00:33<00:46, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2978/7035 [00:33<00:46, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2987/7035 [00:33<00:45, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2996/7035 [00:33<00:45, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3005/7035 [00:34<00:45, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3014/7035 [00:34<00:45, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3023/7035 [00:34<00:45, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3032/7035 [00:34<00:45, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3041/7035 [00:34<00:45, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3050/7035 [00:34<00:45, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3059/7035 [00:34<00:45, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3068/7035 [00:34<00:45, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3077/7035 [00:34<00:45, 86.37batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3086/7035 [00:35<00:45, 86.99batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3095/7035 [00:35<00:45, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3104/7035 [00:35<00:44, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3113/7035 [00:35<00:44, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3122/7035 [00:35<00:44, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3131/7035 [00:35<00:44, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3140/7035 [00:35<00:44, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3149/7035 [00:35<00:44, 87.41batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3158/7035 [00:35<00:44, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3167/7035 [00:35<00:43, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3176/7035 [00:36<00:44, 87.13batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3185/7035 [00:36<00:43, 87.54batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3195/7035 [00:36<00:43, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3204/7035 [00:36<00:43, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3214/7035 [00:36<00:42, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3223/7035 [00:36<00:42, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3232/7035 [00:36<00:42, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3241/7035 [00:36<00:42, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3250/7035 [00:36<00:42, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3259/7035 [00:36<00:42, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3268/7035 [00:37<00:42, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3277/7035 [00:37<00:42, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3286/7035 [00:37<00:41, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3295/7035 [00:37<00:41, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3305/7035 [00:37<00:41, 89.87batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3314/7035 [00:37<00:41, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3323/7035 [00:37<00:42, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3332/7035 [00:37<00:41, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3341/7035 [00:37<00:41, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3350/7035 [00:37<00:41, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3359/7035 [00:38<00:41, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3368/7035 [00:38<00:41, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3377/7035 [00:38<00:41, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3386/7035 [00:38<00:41, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3395/7035 [00:38<00:41, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3404/7035 [00:38<00:40, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3413/7035 [00:38<00:41, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3423/7035 [00:38<00:40, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3432/7035 [00:38<00:40, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3441/7035 [00:39<00:40, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3450/7035 [00:39<00:40, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3459/7035 [00:39<00:40, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3468/7035 [00:39<00:40, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3477/7035 [00:39<00:40, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3486/7035 [00:39<00:39, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3495/7035 [00:39<00:39, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3504/7035 [00:39<00:39, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3513/7035 [00:39<00:39, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3522/7035 [00:39<00:39, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3531/7035 [00:40<00:39, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3540/7035 [00:40<00:39, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3549/7035 [00:40<00:39, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3558/7035 [00:40<00:39, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3567/7035 [00:40<00:39, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3577/7035 [00:40<00:39, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3586/7035 [00:40<00:39, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3595/7035 [00:40<00:38, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3604/7035 [00:40<00:38, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3613/7035 [00:40<00:38, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3622/7035 [00:41<00:38, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3631/7035 [00:41<00:38, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3640/7035 [00:41<00:38, 87.25batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3649/7035 [00:41<00:38, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3659/7035 [00:41<00:38, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3668/7035 [00:41<00:38, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3677/7035 [00:41<00:38, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3686/7035 [00:41<00:37, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3695/7035 [00:41<00:38, 87.24batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3704/7035 [00:41<00:38, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3713/7035 [00:42<00:37, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3722/7035 [00:42<00:37, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3731/7035 [00:42<00:37, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3740/7035 [00:42<00:37, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3749/7035 [00:42<00:37, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3758/7035 [00:42<00:37, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3767/7035 [00:42<00:36, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3776/7035 [00:42<00:37, 87.30batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3785/7035 [00:42<00:36, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3794/7035 [00:43<00:36, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3803/7035 [00:43<00:36, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3812/7035 [00:43<00:36, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3821/7035 [00:43<00:36, 87.03batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3830/7035 [00:43<00:36, 87.37batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3839/7035 [00:43<00:36, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3848/7035 [00:43<00:36, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3857/7035 [00:43<00:36, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3866/7035 [00:43<00:35, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3875/7035 [00:43<00:35, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3884/7035 [00:44<00:35, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3893/7035 [00:44<00:35, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3902/7035 [00:44<00:35, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3911/7035 [00:44<00:35, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3920/7035 [00:44<00:34, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3929/7035 [00:44<00:34, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3938/7035 [00:44<00:34, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3947/7035 [00:44<00:35, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3957/7035 [00:44<00:34, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3966/7035 [00:44<00:34, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3975/7035 [00:45<00:34, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3984/7035 [00:45<00:34, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3993/7035 [00:45<00:34, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4002/7035 [00:45<00:34, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4011/7035 [00:45<00:33, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4021/7035 [00:45<00:33, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4030/7035 [00:45<00:33, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4039/7035 [00:45<00:33, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4048/7035 [00:45<00:33, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4057/7035 [00:45<00:33, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4066/7035 [00:46<00:33, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4075/7035 [00:46<00:33, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4084/7035 [00:46<00:33, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4093/7035 [00:46<00:33, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4102/7035 [00:46<00:32, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4111/7035 [00:46<00:32, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4120/7035 [00:46<00:32, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4129/7035 [00:46<00:32, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4138/7035 [00:46<00:32, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4147/7035 [00:46<00:32, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4156/7035 [00:47<00:32, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4165/7035 [00:47<00:32, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4175/7035 [00:47<00:32, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4184/7035 [00:47<00:32, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4194/7035 [00:47<00:31, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4203/7035 [00:47<00:31, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4212/7035 [00:47<00:32, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4221/7035 [00:47<00:32, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4230/7035 [00:47<00:31, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4239/7035 [00:48<00:31, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4248/7035 [00:48<00:31, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4258/7035 [00:48<00:31, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4267/7035 [00:48<00:31, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4276/7035 [00:48<00:31, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4285/7035 [00:48<00:30, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4294/7035 [00:48<00:30, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4303/7035 [00:48<00:30, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4312/7035 [00:48<00:30, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4322/7035 [00:48<00:30, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4331/7035 [00:49<00:30, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4340/7035 [00:49<00:30, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4350/7035 [00:49<00:29, 89.76batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4359/7035 [00:49<00:29, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4368/7035 [00:49<00:29, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4377/7035 [00:49<00:29, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4386/7035 [00:49<00:29, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4395/7035 [00:49<00:29, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4404/7035 [00:49<00:29, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4413/7035 [00:49<00:29, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4422/7035 [00:50<00:29, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4431/7035 [00:50<00:29, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4441/7035 [00:50<00:29, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4450/7035 [00:50<00:29, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4459/7035 [00:50<00:29, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4468/7035 [00:50<00:28, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4477/7035 [00:50<00:28, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4486/7035 [00:50<00:28, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4495/7035 [00:50<00:28, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4504/7035 [00:51<00:28, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4513/7035 [00:51<00:28, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4522/7035 [00:51<00:28, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4531/7035 [00:51<00:28, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4540/7035 [00:51<00:28, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4549/7035 [00:51<00:28, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4559/7035 [00:51<00:27, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4569/7035 [00:51<00:27, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4578/7035 [00:51<00:27, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4587/7035 [00:51<00:27, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4596/7035 [00:52<00:27, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4605/7035 [00:52<00:27, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4614/7035 [00:52<00:27, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4623/7035 [00:52<00:27, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4632/7035 [00:52<00:27, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4641/7035 [00:52<00:27, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4651/7035 [00:52<00:26, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4660/7035 [00:52<00:26, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4669/7035 [00:52<00:26, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4678/7035 [00:52<00:26, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4687/7035 [00:53<00:26, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4696/7035 [00:53<00:26, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4705/7035 [00:53<00:26, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4714/7035 [00:53<00:26, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4723/7035 [00:53<00:26, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4732/7035 [00:53<00:25, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4741/7035 [00:53<00:25, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4750/7035 [00:53<00:25, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4759/7035 [00:53<00:25, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4768/7035 [00:53<00:25, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4777/7035 [00:54<00:25, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4786/7035 [00:54<00:25, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4796/7035 [00:54<00:25, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4805/7035 [00:54<00:24, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4814/7035 [00:54<00:24, 89.37batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4823/7035 [00:54<00:24, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4832/7035 [00:54<00:24, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4841/7035 [00:54<00:24, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4850/7035 [00:54<00:24, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4859/7035 [00:55<00:24, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4868/7035 [00:55<00:24, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4877/7035 [00:55<00:24, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4886/7035 [00:55<00:24, 87.90batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4895/7035 [00:55<00:24, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4904/7035 [00:55<00:24, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4913/7035 [00:55<00:23, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4922/7035 [00:55<00:23, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4932/7035 [00:55<00:23, 89.48batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4941/7035 [00:55<00:23, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4951/7035 [00:56<00:23, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4960/7035 [00:56<00:23, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4969/7035 [00:56<00:23, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4978/7035 [00:56<00:23, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4987/7035 [00:56<00:23, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4996/7035 [00:56<00:23, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5006/7035 [00:56<00:22, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5015/7035 [00:56<00:22, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5024/7035 [00:56<00:22, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5033/7035 [00:56<00:22, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5042/7035 [00:57<00:22, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5051/7035 [00:57<00:22, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5060/7035 [00:57<00:22, 87.51batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5069/7035 [00:57<00:22, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5078/7035 [00:57<00:22, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5087/7035 [00:57<00:22, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5096/7035 [00:57<00:21, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5105/7035 [00:57<00:21, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5114/7035 [00:57<00:21, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5123/7035 [00:57<00:21, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5132/7035 [00:58<00:21, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5141/7035 [00:58<00:21, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5150/7035 [00:58<00:21, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5159/7035 [00:58<00:21, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5169/7035 [00:58<00:20, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5178/7035 [00:58<00:21, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5187/7035 [00:58<00:21, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5196/7035 [00:58<00:20, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5205/7035 [00:58<00:21, 86.92batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5214/7035 [00:59<00:20, 87.00batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5223/7035 [00:59<00:20, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5232/7035 [00:59<00:20, 87.31batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5241/7035 [00:59<00:20, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5250/7035 [00:59<00:20, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5259/7035 [00:59<00:20, 87.84batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5268/7035 [00:59<00:20, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5277/7035 [00:59<00:19, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5286/7035 [00:59<00:19, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5295/7035 [00:59<00:19, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5304/7035 [01:00<00:19, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5313/7035 [01:00<00:19, 87.27batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5322/7035 [01:00<00:19, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5331/7035 [01:00<00:19, 86.92batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5340/7035 [01:00<00:19, 87.21batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5349/7035 [01:00<00:19, 86.02batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5358/7035 [01:00<00:19, 86.45batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5367/7035 [01:00<00:19, 86.84batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5376/7035 [01:00<00:18, 87.45batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5385/7035 [01:00<00:18, 87.37batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5394/7035 [01:01<00:18, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5403/7035 [01:01<00:18, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5412/7035 [01:01<00:18, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5421/7035 [01:01<00:18, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5430/7035 [01:01<00:18, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5439/7035 [01:01<00:18, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5448/7035 [01:01<00:18, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5457/7035 [01:01<00:17, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5466/7035 [01:01<00:17, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5475/7035 [01:02<00:17, 86.67batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5484/7035 [01:02<00:17, 86.96batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5493/7035 [01:02<00:17, 87.33batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5503/7035 [01:02<00:17, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5512/7035 [01:02<00:17, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5521/7035 [01:02<00:17, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5530/7035 [01:02<00:17, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5539/7035 [01:02<00:17, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5548/7035 [01:02<00:16, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5557/7035 [01:02<00:16, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5566/7035 [01:03<00:16, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5575/7035 [01:03<00:16, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5584/7035 [01:03<00:16, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5593/7035 [01:03<00:16, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5602/7035 [01:03<00:16, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5611/7035 [01:03<00:16, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5620/7035 [01:03<00:16, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5629/7035 [01:03<00:16, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5638/7035 [01:03<00:15, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5647/7035 [01:03<00:15, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5656/7035 [01:04<00:15, 86.89batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5665/7035 [01:04<00:15, 87.26batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5675/7035 [01:04<00:15, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5685/7035 [01:04<00:15, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5694/7035 [01:04<00:15, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5703/7035 [01:04<00:15, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5712/7035 [01:04<00:14, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5721/7035 [01:04<00:14, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5730/7035 [01:04<00:14, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5739/7035 [01:05<00:14, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5748/7035 [01:05<00:14, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5757/7035 [01:05<00:14, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5766/7035 [01:05<00:14, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5775/7035 [01:05<00:14, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5784/7035 [01:05<00:14, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5793/7035 [01:05<00:13, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5802/7035 [01:05<00:13, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5811/7035 [01:05<00:13, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5820/7035 [01:05<00:13, 88.06batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5829/7035 [01:06<00:13, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5838/7035 [01:06<00:13, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5847/7035 [01:06<00:13, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5856/7035 [01:06<00:13, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5865/7035 [01:06<00:13, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5875/7035 [01:06<00:13, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5884/7035 [01:06<00:12, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5893/7035 [01:06<00:12, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5902/7035 [01:06<00:12, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5911/7035 [01:06<00:12, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5920/7035 [01:07<00:12, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5929/7035 [01:07<00:12, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5939/7035 [01:07<00:12, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5948/7035 [01:07<00:12, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5957/7035 [01:07<00:12, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5966/7035 [01:07<00:12, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5975/7035 [01:07<00:11, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5984/7035 [01:07<00:11, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5993/7035 [01:07<00:11, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6002/7035 [01:07<00:11, 86.79batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6011/7035 [01:08<00:11, 86.91batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6020/7035 [01:08<00:11, 87.36batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6029/7035 [01:08<00:11, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6038/7035 [01:08<00:11, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6047/7035 [01:08<00:11, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6056/7035 [01:08<00:11, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6065/7035 [01:08<00:10, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6074/7035 [01:08<00:10, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6083/7035 [01:08<00:10, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6092/7035 [01:09<00:10, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6101/7035 [01:09<00:10, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6110/7035 [01:09<00:10, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6120/7035 [01:09<00:10, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6129/7035 [01:09<00:10, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6139/7035 [01:09<00:10, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6148/7035 [01:09<00:09, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6157/7035 [01:09<00:09, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6166/7035 [01:09<00:09, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6175/7035 [01:09<00:09, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6184/7035 [01:10<00:09, 87.15batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6193/7035 [01:10<00:09, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6203/7035 [01:10<00:09, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6212/7035 [01:10<00:09, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6221/7035 [01:10<00:09, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6230/7035 [01:10<00:09, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6239/7035 [01:10<00:08, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6248/7035 [01:10<00:08, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6257/7035 [01:10<00:08, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6267/7035 [01:10<00:08, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6276/7035 [01:11<00:08, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6285/7035 [01:11<00:08, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6294/7035 [01:11<00:08, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6303/7035 [01:11<00:08, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6312/7035 [01:11<00:08, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6321/7035 [01:11<00:07, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6330/7035 [01:11<00:07, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6339/7035 [01:11<00:07, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6349/7035 [01:11<00:07, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6358/7035 [01:12<00:07, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6367/7035 [01:12<00:07, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6376/7035 [01:12<00:07, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6386/7035 [01:12<00:07, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6396/7035 [01:12<00:07, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6405/7035 [01:12<00:07, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6415/7035 [01:12<00:06, 89.72batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6424/7035 [01:12<00:06, 89.78batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6433/7035 [01:12<00:06, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6442/7035 [01:12<00:06, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6451/7035 [01:13<00:06, 89.53batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6460/7035 [01:13<00:06, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6469/7035 [01:13<00:06, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6478/7035 [01:13<00:06, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6487/7035 [01:13<00:06, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6496/7035 [01:13<00:06, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6505/7035 [01:13<00:05, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6514/7035 [01:13<00:05, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6523/7035 [01:13<00:05, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6532/7035 [01:13<00:05, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6541/7035 [01:14<00:05, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6550/7035 [01:14<00:05, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6559/7035 [01:14<00:05, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6568/7035 [01:14<00:05, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6577/7035 [01:14<00:05, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6586/7035 [01:14<00:05, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6595/7035 [01:14<00:04, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6604/7035 [01:14<00:04, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6613/7035 [01:14<00:04, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6622/7035 [01:14<00:04, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6631/7035 [01:15<00:04, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6640/7035 [01:15<00:04, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6649/7035 [01:15<00:04, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6658/7035 [01:15<00:04, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6668/7035 [01:15<00:04, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6677/7035 [01:15<00:04, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6686/7035 [01:15<00:03, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6695/7035 [01:15<00:03, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6704/7035 [01:15<00:03, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6713/7035 [01:15<00:03, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6722/7035 [01:16<00:03, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6731/7035 [01:16<00:03, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6741/7035 [01:16<00:03, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6751/7035 [01:16<00:03, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6760/7035 [01:16<00:03, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6769/7035 [01:16<00:02, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6778/7035 [01:16<00:02, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6787/7035 [01:16<00:02, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6796/7035 [01:16<00:02, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6805/7035 [01:17<00:02, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6814/7035 [01:17<00:02, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6823/7035 [01:17<00:02, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6833/7035 [01:17<00:02, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6842/7035 [01:17<00:02, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6851/7035 [01:17<00:02, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6860/7035 [01:17<00:01, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6869/7035 [01:17<00:01, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6879/7035 [01:17<00:01, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6888/7035 [01:17<00:01, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6897/7035 [01:18<00:01, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6906/7035 [01:18<00:01, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6915/7035 [01:18<00:01, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6924/7035 [01:18<00:01, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6933/7035 [01:18<00:01, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6942/7035 [01:18<00:01, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6951/7035 [01:18<00:00, 87.68batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6960/7035 [01:18<00:00, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6969/7035 [01:18<00:00, 87.07batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6978/7035 [01:18<00:00, 84.74batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6987/7035 [01:19<00:00, 85.83batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6996/7035 [01:19<00:00, 86.53batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7005/7035 [01:19<00:00, 87.47batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7014/7035 [01:19<00:00, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7024/7035 [01:19<00:00, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7033/7035 [01:19<00:00, 88.74batch/s]\u001b[A\n",
"Training epochs on cuda: 60%|██████ | 3/5 [03:59<02:39, 79.77s/epoch, loss=0.00137, prev_loss=0.00143]\n",
"Training batches on cuda: 0%| | 0/7035 [00:00<?, ?batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 1/7035 [00:00<14:17, 8.20batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 10/7035 [00:00<10:22, 11.28batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 19/7035 [00:00<07:39, 15.26batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 28/7035 [00:00<05:44, 20.31batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 37/7035 [00:00<04:24, 26.42batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 46/7035 [00:00<03:28, 33.48batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 55/7035 [00:00<02:49, 41.22batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 64/7035 [00:00<02:21, 49.13batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 74/7035 [00:00<02:02, 56.99batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 83/7035 [00:01<01:48, 63.93batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 92/7035 [00:01<01:39, 69.48batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 101/7035 [00:01<01:33, 74.15batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 110/7035 [00:01<01:29, 77.36batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 119/7035 [00:01<01:26, 80.17batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 128/7035 [00:01<01:23, 82.67batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 137/7035 [00:01<01:21, 84.19batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 146/7035 [00:01<01:20, 85.83batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 156/7035 [00:01<01:18, 87.32batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 165/7035 [00:01<01:18, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 175/7035 [00:02<01:17, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 185/7035 [00:02<01:16, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 194/7035 [00:02<01:16, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 203/7035 [00:02<01:16, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 213/7035 [00:02<01:16, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 223/7035 [00:02<01:15, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 233/7035 [00:02<01:15, 89.90batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 243/7035 [00:02<01:15, 90.07batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 253/7035 [00:02<01:14, 90.52batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 263/7035 [00:03<01:15, 90.14batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 273/7035 [00:03<01:15, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 283/7035 [00:03<01:15, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 292/7035 [00:03<01:15, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 301/7035 [00:03<01:15, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 310/7035 [00:03<01:16, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 320/7035 [00:03<01:15, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 329/7035 [00:03<01:15, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 338/7035 [00:03<01:15, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 347/7035 [00:04<01:16, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 356/7035 [00:04<01:16, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 365/7035 [00:04<01:15, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 374/7035 [00:04<01:15, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 383/7035 [00:04<01:15, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 392/7035 [00:04<01:15, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 401/7035 [00:04<01:15, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 410/7035 [00:04<01:15, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 419/7035 [00:04<01:14, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 429/7035 [00:04<01:14, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 438/7035 [00:05<01:14, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 447/7035 [00:05<01:14, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 456/7035 [00:05<01:13, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 465/7035 [00:05<01:14, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 474/7035 [00:05<01:14, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 483/7035 [00:05<01:15, 87.20batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 492/7035 [00:05<01:14, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 502/7035 [00:05<01:13, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 511/7035 [00:05<01:13, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 520/7035 [00:05<01:13, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 529/7035 [00:06<01:13, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 539/7035 [00:06<01:12, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 548/7035 [00:06<01:12, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 557/7035 [00:06<01:12, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 567/7035 [00:06<01:12, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 576/7035 [00:06<01:12, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 585/7035 [00:06<01:12, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 594/7035 [00:06<01:12, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 603/7035 [00:06<01:12, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 612/7035 [00:06<01:12, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 621/7035 [00:07<01:12, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 630/7035 [00:07<01:14, 85.83batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 639/7035 [00:07<01:13, 86.49batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 648/7035 [00:07<01:13, 87.04batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 657/7035 [00:07<01:13, 87.00batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 666/7035 [00:07<01:12, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 676/7035 [00:07<01:11, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 686/7035 [00:07<01:11, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 695/7035 [00:07<01:12, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 704/7035 [00:08<01:11, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 713/7035 [00:08<01:11, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 722/7035 [00:08<01:11, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 731/7035 [00:08<01:11, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 740/7035 [00:08<01:11, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 749/7035 [00:08<01:11, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 758/7035 [00:08<01:11, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 767/7035 [00:08<01:11, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 776/7035 [00:08<01:11, 87.40batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 785/7035 [00:08<01:11, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 794/7035 [00:09<01:11, 87.28batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 803/7035 [00:09<01:11, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 812/7035 [00:09<01:11, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 821/7035 [00:09<01:10, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 830/7035 [00:09<01:11, 87.14batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 839/7035 [00:09<01:11, 86.66batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 848/7035 [00:09<01:11, 86.78batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 857/7035 [00:09<01:11, 86.63batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 866/7035 [00:09<01:10, 87.45batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 875/7035 [00:10<01:09, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 884/7035 [00:10<01:09, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 893/7035 [00:10<01:09, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 902/7035 [00:10<01:09, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 911/7035 [00:10<01:09, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 920/7035 [00:10<01:09, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 929/7035 [00:10<01:09, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 938/7035 [00:10<01:09, 87.72batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 947/7035 [00:10<01:09, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 957/7035 [00:10<01:08, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 966/7035 [00:11<01:08, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 976/7035 [00:11<01:08, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 985/7035 [00:11<01:07, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 994/7035 [00:11<01:07, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1003/7035 [00:11<01:09, 86.97batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1012/7035 [00:11<01:09, 87.00batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1021/7035 [00:11<01:08, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1030/7035 [00:11<01:08, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1039/7035 [00:11<01:08, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1048/7035 [00:11<01:07, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1057/7035 [00:12<01:07, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1066/7035 [00:12<01:07, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1075/7035 [00:12<01:07, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1084/7035 [00:12<01:07, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1093/7035 [00:12<01:06, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1102/7035 [00:12<01:07, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1111/7035 [00:12<01:07, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1121/7035 [00:12<01:06, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1130/7035 [00:12<01:06, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1139/7035 [00:12<01:06, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1148/7035 [00:13<01:06, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1157/7035 [00:13<01:06, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1166/7035 [00:13<01:06, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1175/7035 [00:13<01:05, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1184/7035 [00:13<01:05, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1193/7035 [00:13<01:06, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1202/7035 [00:13<01:06, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1211/7035 [00:13<01:05, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1220/7035 [00:13<01:05, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1229/7035 [00:14<01:05, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1239/7035 [00:14<01:04, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1248/7035 [00:14<01:04, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1257/7035 [00:14<01:04, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1266/7035 [00:14<01:04, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1275/7035 [00:14<01:05, 87.60batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1284/7035 [00:14<01:05, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1293/7035 [00:14<01:05, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1302/7035 [00:14<01:06, 86.78batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1311/7035 [00:14<01:05, 86.99batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1320/7035 [00:15<01:05, 86.71batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1329/7035 [00:15<01:05, 87.30batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1338/7035 [00:15<01:04, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1347/7035 [00:15<01:04, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1357/7035 [00:15<01:04, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1366/7035 [00:15<01:04, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1375/7035 [00:15<01:04, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1385/7035 [00:15<01:03, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1395/7035 [00:15<01:03, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1404/7035 [00:15<01:03, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1414/7035 [00:16<01:02, 89.66batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1423/7035 [00:16<01:02, 89.68batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1433/7035 [00:16<01:02, 90.04batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1443/7035 [00:16<01:01, 90.23batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1453/7035 [00:16<01:01, 90.28batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1463/7035 [00:16<01:02, 89.77batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1473/7035 [00:16<01:01, 90.02batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1483/7035 [00:16<01:01, 90.07batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1493/7035 [00:16<01:02, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1502/7035 [00:17<01:02, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1511/7035 [00:17<01:01, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1520/7035 [00:17<01:01, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1529/7035 [00:17<01:01, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1538/7035 [00:17<01:01, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1547/7035 [00:17<01:02, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1556/7035 [00:17<01:02, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1566/7035 [00:17<01:01, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1575/7035 [00:17<01:01, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1584/7035 [00:18<01:01, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1593/7035 [00:18<01:01, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1602/7035 [00:18<01:00, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1611/7035 [00:18<01:00, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1620/7035 [00:18<01:00, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1629/7035 [00:18<01:00, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1638/7035 [00:18<01:00, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1647/7035 [00:18<01:01, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1656/7035 [00:18<01:00, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1665/7035 [00:18<01:00, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1674/7035 [00:19<01:00, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1684/7035 [00:19<01:00, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1693/7035 [00:19<00:59, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1702/7035 [00:19<00:59, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1711/7035 [00:19<00:59, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1720/7035 [00:19<00:59, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1729/7035 [00:19<00:59, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1738/7035 [00:19<00:59, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1747/7035 [00:19<00:59, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1756/7035 [00:19<00:59, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1765/7035 [00:20<00:59, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1774/7035 [00:20<00:59, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1783/7035 [00:20<00:59, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1792/7035 [00:20<00:59, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1801/7035 [00:20<00:59, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1811/7035 [00:20<00:58, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1820/7035 [00:20<00:59, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1829/7035 [00:20<00:59, 87.34batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1838/7035 [00:20<00:59, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1847/7035 [00:20<00:58, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1857/7035 [00:21<00:58, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1867/7035 [00:21<00:57, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1876/7035 [00:21<00:57, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1885/7035 [00:21<00:57, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1894/7035 [00:21<00:57, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1903/7035 [00:21<00:57, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1912/7035 [00:21<00:57, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1922/7035 [00:21<00:57, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1931/7035 [00:21<00:57, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1940/7035 [00:22<00:57, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1949/7035 [00:22<00:57, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1958/7035 [00:22<00:57, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1967/7035 [00:22<00:57, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1976/7035 [00:22<00:57, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1985/7035 [00:22<00:57, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1994/7035 [00:22<00:56, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 2003/7035 [00:22<00:57, 87.83batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2012/7035 [00:22<00:56, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2022/7035 [00:22<00:56, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2031/7035 [00:23<00:56, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2040/7035 [00:23<00:56, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2050/7035 [00:23<00:55, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2059/7035 [00:23<00:56, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2068/7035 [00:23<00:56, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2078/7035 [00:23<00:55, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2088/7035 [00:23<00:55, 89.71batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2097/7035 [00:23<00:55, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2106/7035 [00:23<00:55, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2116/7035 [00:23<00:55, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2125/7035 [00:24<00:55, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2134/7035 [00:24<00:55, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2143/7035 [00:24<00:54, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2152/7035 [00:24<00:55, 87.52batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2161/7035 [00:24<00:55, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2170/7035 [00:24<00:55, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2179/7035 [00:24<00:54, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2188/7035 [00:24<00:55, 87.15batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2198/7035 [00:24<00:54, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2207/7035 [00:25<00:55, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2216/7035 [00:25<00:54, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2225/7035 [00:25<00:54, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2234/7035 [00:25<00:54, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2243/7035 [00:25<00:53, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2253/7035 [00:25<00:53, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2262/7035 [00:25<00:53, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2271/7035 [00:25<00:53, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2281/7035 [00:25<00:53, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2291/7035 [00:25<00:52, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2300/7035 [00:26<00:52, 89.64batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2310/7035 [00:26<00:52, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2319/7035 [00:26<00:52, 89.76batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2328/7035 [00:26<00:53, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2337/7035 [00:26<00:53, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2346/7035 [00:26<00:52, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2355/7035 [00:26<00:52, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2364/7035 [00:26<00:53, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2374/7035 [00:26<00:52, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2383/7035 [00:27<00:52, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2393/7035 [00:27<00:52, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2402/7035 [00:27<00:51, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2411/7035 [00:27<00:51, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2420/7035 [00:27<00:51, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2429/7035 [00:27<00:51, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2438/7035 [00:27<00:51, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2447/7035 [00:27<00:51, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2456/7035 [00:27<00:52, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2466/7035 [00:27<00:51, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2475/7035 [00:28<00:51, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2484/7035 [00:28<00:51, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2493/7035 [00:28<00:51, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2502/7035 [00:28<00:51, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2511/7035 [00:28<00:51, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2520/7035 [00:28<00:50, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2529/7035 [00:28<00:51, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2538/7035 [00:28<00:50, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2547/7035 [00:28<00:51, 87.72batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2556/7035 [00:28<00:51, 87.10batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2565/7035 [00:29<00:50, 87.90batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2574/7035 [00:29<00:50, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2583/7035 [00:29<00:50, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2592/7035 [00:29<00:50, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2601/7035 [00:29<00:49, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2610/7035 [00:29<00:49, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2619/7035 [00:29<00:49, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2629/7035 [00:29<00:49, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2638/7035 [00:29<00:49, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2648/7035 [00:29<00:49, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2657/7035 [00:30<00:49, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2666/7035 [00:30<00:48, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2675/7035 [00:30<00:48, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2684/7035 [00:30<00:48, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2693/7035 [00:30<00:48, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2702/7035 [00:30<00:48, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2711/7035 [00:30<00:48, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2720/7035 [00:30<00:49, 87.61batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2729/7035 [00:30<00:49, 86.77batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2738/7035 [00:31<00:49, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2747/7035 [00:31<00:48, 88.00batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2756/7035 [00:31<00:48, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2765/7035 [00:31<00:48, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2774/7035 [00:31<00:48, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2783/7035 [00:31<00:47, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2792/7035 [00:31<00:47, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2801/7035 [00:31<00:47, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2810/7035 [00:31<00:47, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2819/7035 [00:31<00:47, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2828/7035 [00:32<00:47, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2837/7035 [00:32<00:47, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2846/7035 [00:32<00:47, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2856/7035 [00:32<00:46, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2865/7035 [00:32<00:46, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2874/7035 [00:32<00:46, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2883/7035 [00:32<00:46, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2892/7035 [00:32<00:46, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2901/7035 [00:32<00:46, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2910/7035 [00:32<00:47, 87.30batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2919/7035 [00:33<00:46, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2928/7035 [00:33<00:46, 87.68batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2937/7035 [00:33<00:47, 87.14batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2946/7035 [00:33<00:46, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2955/7035 [00:33<00:46, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2964/7035 [00:33<00:46, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2973/7035 [00:33<00:46, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2982/7035 [00:33<00:46, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2991/7035 [00:33<00:45, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3000/7035 [00:33<00:45, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3009/7035 [00:34<00:45, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3019/7035 [00:34<00:45, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3028/7035 [00:34<00:45, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3037/7035 [00:34<00:45, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3046/7035 [00:34<00:45, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3055/7035 [00:34<00:45, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3064/7035 [00:34<00:44, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3073/7035 [00:34<00:44, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3082/7035 [00:34<00:44, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3091/7035 [00:35<00:44, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3100/7035 [00:35<00:44, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3110/7035 [00:35<00:43, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3119/7035 [00:35<00:43, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3128/7035 [00:35<00:43, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3137/7035 [00:35<00:44, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3146/7035 [00:35<00:43, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3156/7035 [00:35<00:43, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3166/7035 [00:35<00:43, 89.63batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3175/7035 [00:35<00:43, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3184/7035 [00:36<00:43, 87.79batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3193/7035 [00:36<00:43, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3202/7035 [00:36<00:43, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3211/7035 [00:36<00:42, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3220/7035 [00:36<00:42, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3229/7035 [00:36<00:42, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3239/7035 [00:36<00:42, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3249/7035 [00:36<00:42, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3258/7035 [00:36<00:42, 89.77batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3267/7035 [00:36<00:43, 87.38batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3276/7035 [00:37<00:42, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3285/7035 [00:37<00:42, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3294/7035 [00:37<00:42, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3303/7035 [00:37<00:42, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3312/7035 [00:37<00:42, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3321/7035 [00:37<00:41, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3330/7035 [00:37<00:41, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3339/7035 [00:37<00:41, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3348/7035 [00:37<00:42, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3357/7035 [00:38<00:42, 87.19batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3366/7035 [00:38<00:41, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3375/7035 [00:38<00:41, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3384/7035 [00:38<00:41, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3393/7035 [00:38<00:41, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3402/7035 [00:38<00:41, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3411/7035 [00:38<00:41, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3420/7035 [00:38<00:40, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3429/7035 [00:38<00:40, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3438/7035 [00:38<00:40, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3447/7035 [00:39<00:40, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3456/7035 [00:39<00:40, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3465/7035 [00:39<00:40, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3474/7035 [00:39<00:39, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3483/7035 [00:39<00:39, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3492/7035 [00:39<00:39, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3501/7035 [00:39<00:39, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3510/7035 [00:39<00:39, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3519/7035 [00:39<00:39, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3528/7035 [00:39<00:39, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3537/7035 [00:40<00:39, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3546/7035 [00:40<00:39, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3556/7035 [00:40<00:39, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3565/7035 [00:40<00:38, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3574/7035 [00:40<00:38, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3583/7035 [00:40<00:39, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3592/7035 [00:40<00:39, 87.80batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3601/7035 [00:40<00:38, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3611/7035 [00:40<00:38, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3620/7035 [00:40<00:38, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3629/7035 [00:41<00:38, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3638/7035 [00:41<00:38, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3648/7035 [00:41<00:37, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3657/7035 [00:41<00:37, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3666/7035 [00:41<00:37, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3675/7035 [00:41<00:37, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3684/7035 [00:41<00:37, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3693/7035 [00:41<00:37, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3702/7035 [00:41<00:37, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3711/7035 [00:41<00:37, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3720/7035 [00:42<00:37, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3729/7035 [00:42<00:37, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3738/7035 [00:42<00:37, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3747/7035 [00:42<00:36, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3757/7035 [00:42<00:36, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3767/7035 [00:42<00:36, 89.92batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3776/7035 [00:42<00:36, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3785/7035 [00:42<00:36, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3794/7035 [00:42<00:36, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3803/7035 [00:43<00:36, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3812/7035 [00:43<00:36, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3822/7035 [00:43<00:36, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3831/7035 [00:43<00:36, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3840/7035 [00:43<00:35, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3850/7035 [00:43<00:35, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3859/7035 [00:43<00:35, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3868/7035 [00:43<00:35, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3877/7035 [00:43<00:35, 89.68batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3886/7035 [00:43<00:35, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3896/7035 [00:44<00:34, 89.85batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3906/7035 [00:44<00:34, 90.15batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3916/7035 [00:44<00:34, 90.49batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3926/7035 [00:44<00:34, 89.96batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3935/7035 [00:44<00:34, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3944/7035 [00:44<00:34, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3953/7035 [00:44<00:34, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3963/7035 [00:44<00:34, 89.73batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3972/7035 [00:44<00:34, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3981/7035 [00:45<00:34, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3990/7035 [00:45<00:34, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3999/7035 [00:45<00:34, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4008/7035 [00:45<00:34, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4017/7035 [00:45<00:34, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4026/7035 [00:45<00:33, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4036/7035 [00:45<00:33, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4046/7035 [00:45<00:33, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4056/7035 [00:45<00:33, 89.92batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4066/7035 [00:45<00:32, 89.98batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4075/7035 [00:46<00:33, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4084/7035 [00:46<00:33, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4093/7035 [00:46<00:33, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4102/7035 [00:46<00:33, 87.32batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4112/7035 [00:46<00:33, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4121/7035 [00:46<00:32, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4130/7035 [00:46<00:32, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4139/7035 [00:46<00:32, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4148/7035 [00:46<00:32, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4157/7035 [00:46<00:32, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4166/7035 [00:47<00:32, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4175/7035 [00:47<00:32, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4184/7035 [00:47<00:32, 88.49batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4193/7035 [00:47<00:32, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4202/7035 [00:47<00:31, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4211/7035 [00:47<00:31, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4220/7035 [00:47<00:31, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4229/7035 [00:47<00:31, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4239/7035 [00:47<00:31, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4248/7035 [00:48<00:31, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4257/7035 [00:48<00:31, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4266/7035 [00:48<00:31, 87.28batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4275/7035 [00:48<00:31, 87.57batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4285/7035 [00:48<00:31, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4294/7035 [00:48<00:31, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4303/7035 [00:48<00:30, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4312/7035 [00:48<00:30, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4321/7035 [00:48<00:30, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4330/7035 [00:48<00:30, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4339/7035 [00:49<00:30, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4348/7035 [00:49<00:30, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4357/7035 [00:49<00:30, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4366/7035 [00:49<00:30, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4375/7035 [00:49<00:30, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4385/7035 [00:49<00:29, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4395/7035 [00:49<00:29, 89.66batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4404/7035 [00:49<00:29, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4413/7035 [00:49<00:29, 89.63batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4422/7035 [00:49<00:29, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4431/7035 [00:50<00:29, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4440/7035 [00:50<00:29, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4449/7035 [00:50<00:29, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4458/7035 [00:50<00:29, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4467/7035 [00:50<00:28, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4476/7035 [00:50<00:28, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4485/7035 [00:50<00:28, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4494/7035 [00:50<00:28, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4504/7035 [00:50<00:28, 89.53batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4513/7035 [00:51<00:28, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4522/7035 [00:51<00:28, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4531/7035 [00:51<00:27, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4541/7035 [00:51<00:27, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4550/7035 [00:51<00:27, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4559/7035 [00:51<00:27, 89.88batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4569/7035 [00:51<00:27, 90.05batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4579/7035 [00:51<00:27, 90.14batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4589/7035 [00:51<00:27, 90.12batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4599/7035 [00:51<00:27, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4608/7035 [00:52<00:27, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4617/7035 [00:52<00:27, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4626/7035 [00:52<00:27, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4635/7035 [00:52<00:27, 87.56batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4644/7035 [00:52<00:27, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4654/7035 [00:52<00:26, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4663/7035 [00:52<00:26, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4672/7035 [00:52<00:26, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4681/7035 [00:52<00:26, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4691/7035 [00:52<00:26, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4700/7035 [00:53<00:26, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4709/7035 [00:53<00:26, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4718/7035 [00:53<00:25, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4727/7035 [00:53<00:25, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4736/7035 [00:53<00:25, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4745/7035 [00:53<00:25, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4754/7035 [00:53<00:25, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4763/7035 [00:53<00:25, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4772/7035 [00:53<00:25, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4781/7035 [00:54<00:25, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4790/7035 [00:54<00:25, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4799/7035 [00:54<00:25, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4808/7035 [00:54<00:25, 87.43batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4817/7035 [00:54<00:25, 86.66batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4826/7035 [00:54<00:25, 87.30batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4835/7035 [00:54<00:25, 87.98batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4844/7035 [00:54<00:24, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4853/7035 [00:54<00:24, 87.45batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4862/7035 [00:54<00:24, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4871/7035 [00:55<00:24, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4880/7035 [00:55<00:24, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4890/7035 [00:55<00:24, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4899/7035 [00:55<00:24, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4909/7035 [00:55<00:23, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4918/7035 [00:55<00:23, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4928/7035 [00:55<00:23, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4937/7035 [00:55<00:23, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4947/7035 [00:55<00:23, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4957/7035 [00:55<00:23, 90.05batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4967/7035 [00:56<00:22, 89.97batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4976/7035 [00:56<00:22, 89.79batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4985/7035 [00:56<00:22, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4994/7035 [00:56<00:22, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5003/7035 [00:56<00:22, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5012/7035 [00:56<00:22, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5021/7035 [00:56<00:22, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5030/7035 [00:56<00:22, 87.71batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5039/7035 [00:56<00:22, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5048/7035 [00:57<00:22, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5057/7035 [00:57<00:22, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5066/7035 [00:57<00:22, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5076/7035 [00:57<00:21, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5085/7035 [00:57<00:21, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5094/7035 [00:57<00:21, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5103/7035 [00:57<00:21, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5112/7035 [00:57<00:21, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5121/7035 [00:57<00:21, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5130/7035 [00:57<00:21, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5139/7035 [00:58<00:21, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5148/7035 [00:58<00:21, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5157/7035 [00:58<00:21, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5166/7035 [00:58<00:21, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5175/7035 [00:58<00:21, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5184/7035 [00:58<00:21, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5193/7035 [00:58<00:20, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5203/7035 [00:58<00:20, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5212/7035 [00:58<00:20, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5221/7035 [00:58<00:20, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5231/7035 [00:59<00:20, 89.71batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5240/7035 [00:59<00:20, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5249/7035 [00:59<00:20, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5258/7035 [00:59<00:19, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5267/7035 [00:59<00:19, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5276/7035 [00:59<00:19, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5285/7035 [00:59<00:19, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5295/7035 [00:59<00:19, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5305/7035 [00:59<00:19, 90.16batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5315/7035 [01:00<00:19, 90.28batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5325/7035 [01:00<00:18, 90.17batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5335/7035 [01:00<00:18, 90.09batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5345/7035 [01:00<00:18, 89.83batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5355/7035 [01:00<00:18, 90.09batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5365/7035 [01:00<00:18, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5375/7035 [01:00<00:18, 89.99batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5385/7035 [01:00<00:18, 90.17batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5395/7035 [01:00<00:18, 90.41batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5405/7035 [01:01<00:18, 90.26batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5415/7035 [01:01<00:17, 90.40batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5425/7035 [01:01<00:17, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5434/7035 [01:01<00:18, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5444/7035 [01:01<00:17, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5453/7035 [01:01<00:17, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5463/7035 [01:01<00:17, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5473/7035 [01:01<00:17, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5483/7035 [01:01<00:17, 89.79batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5492/7035 [01:01<00:17, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5501/7035 [01:02<00:17, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5510/7035 [01:02<00:17, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5519/7035 [01:02<00:17, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5528/7035 [01:02<00:16, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5537/7035 [01:02<00:16, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5547/7035 [01:02<00:16, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5556/7035 [01:02<00:16, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5565/7035 [01:02<00:16, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5574/7035 [01:02<00:16, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5583/7035 [01:03<00:16, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5592/7035 [01:03<00:16, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5601/7035 [01:03<00:16, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5611/7035 [01:03<00:15, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5620/7035 [01:03<00:16, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5629/7035 [01:03<00:15, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5638/7035 [01:03<00:15, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5647/7035 [01:03<00:15, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5656/7035 [01:03<00:15, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5665/7035 [01:03<00:15, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5674/7035 [01:04<00:15, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5683/7035 [01:04<00:15, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5692/7035 [01:04<00:15, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5701/7035 [01:04<00:15, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5710/7035 [01:04<00:14, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5719/7035 [01:04<00:14, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5728/7035 [01:04<00:14, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5737/7035 [01:04<00:14, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5746/7035 [01:04<00:14, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5755/7035 [01:04<00:14, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5764/7035 [01:05<00:14, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5773/7035 [01:05<00:14, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5782/7035 [01:05<00:14, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5791/7035 [01:05<00:14, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5800/7035 [01:05<00:14, 87.95batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5809/7035 [01:05<00:14, 87.37batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5818/7035 [01:05<00:13, 88.04batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5828/7035 [01:05<00:13, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5837/7035 [01:05<00:13, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5846/7035 [01:05<00:13, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5855/7035 [01:06<00:13, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5864/7035 [01:06<00:13, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5873/7035 [01:06<00:13, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5882/7035 [01:06<00:13, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5891/7035 [01:06<00:12, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5900/7035 [01:06<00:12, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5910/7035 [01:06<00:12, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5919/7035 [01:06<00:12, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5928/7035 [01:06<00:12, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5938/7035 [01:07<00:12, 89.87batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5947/7035 [01:07<00:12, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5956/7035 [01:07<00:12, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5965/7035 [01:07<00:12, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5975/7035 [01:07<00:11, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5984/7035 [01:07<00:11, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5993/7035 [01:07<00:11, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6002/7035 [01:07<00:11, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6011/7035 [01:07<00:11, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6020/7035 [01:07<00:11, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6029/7035 [01:08<00:11, 88.17batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6038/7035 [01:08<00:11, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6047/7035 [01:08<00:11, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6056/7035 [01:08<00:11, 87.37batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6065/7035 [01:08<00:11, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6074/7035 [01:08<00:10, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6083/7035 [01:08<00:10, 88.12batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6092/7035 [01:08<00:10, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6101/7035 [01:08<00:10, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6111/7035 [01:08<00:10, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6121/7035 [01:09<00:10, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6130/7035 [01:09<00:10, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6139/7035 [01:09<00:10, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6148/7035 [01:09<00:09, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6157/7035 [01:09<00:09, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6166/7035 [01:09<00:09, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6175/7035 [01:09<00:09, 87.46batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6184/7035 [01:09<00:09, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6193/7035 [01:09<00:09, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6203/7035 [01:10<00:09, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6212/7035 [01:10<00:09, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6222/7035 [01:10<00:09, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6231/7035 [01:10<00:09, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6240/7035 [01:10<00:08, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6249/7035 [01:10<00:08, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6258/7035 [01:10<00:08, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6267/7035 [01:10<00:08, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6276/7035 [01:10<00:08, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6285/7035 [01:10<00:08, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6294/7035 [01:11<00:08, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6303/7035 [01:11<00:08, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6313/7035 [01:11<00:08, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6322/7035 [01:11<00:08, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6331/7035 [01:11<00:07, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6341/7035 [01:11<00:07, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6350/7035 [01:11<00:07, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6359/7035 [01:11<00:07, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6368/7035 [01:11<00:07, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6377/7035 [01:11<00:07, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6386/7035 [01:12<00:07, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6395/7035 [01:12<00:07, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6405/7035 [01:12<00:07, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6414/7035 [01:12<00:06, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6423/7035 [01:12<00:06, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6432/7035 [01:12<00:06, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6441/7035 [01:12<00:06, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6450/7035 [01:12<00:06, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6459/7035 [01:12<00:06, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6469/7035 [01:13<00:06, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6478/7035 [01:13<00:06, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6487/7035 [01:13<00:06, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6496/7035 [01:13<00:06, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6505/7035 [01:13<00:05, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6514/7035 [01:13<00:05, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6523/7035 [01:13<00:05, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6532/7035 [01:13<00:05, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6542/7035 [01:13<00:05, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6551/7035 [01:13<00:05, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6560/7035 [01:14<00:05, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6569/7035 [01:14<00:05, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6578/7035 [01:14<00:05, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6587/7035 [01:14<00:05, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6596/7035 [01:14<00:05, 87.66batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6605/7035 [01:14<00:04, 88.05batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6614/7035 [01:14<00:04, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6623/7035 [01:14<00:04, 87.67batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6632/7035 [01:14<00:04, 88.33batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6641/7035 [01:14<00:04, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6650/7035 [01:15<00:04, 86.65batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6659/7035 [01:15<00:04, 87.17batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6668/7035 [01:15<00:04, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6677/7035 [01:15<00:04, 87.39batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6686/7035 [01:15<00:03, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6696/7035 [01:15<00:03, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6705/7035 [01:15<00:03, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6714/7035 [01:15<00:03, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6724/7035 [01:15<00:03, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6733/7035 [01:15<00:03, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6743/7035 [01:16<00:03, 89.78batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6753/7035 [01:16<00:03, 89.97batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6762/7035 [01:16<00:03, 89.68batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6771/7035 [01:16<00:02, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6780/7035 [01:16<00:02, 89.53batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6790/7035 [01:16<00:02, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6799/7035 [01:16<00:02, 89.68batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6808/7035 [01:16<00:02, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6818/7035 [01:16<00:02, 89.64batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6827/7035 [01:17<00:02, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6836/7035 [01:17<00:02, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6846/7035 [01:17<00:02, 89.22batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6855/7035 [01:17<00:02, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6864/7035 [01:17<00:01, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6874/7035 [01:17<00:01, 89.59batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6884/7035 [01:17<00:01, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6894/7035 [01:17<00:01, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6903/7035 [01:17<00:01, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6913/7035 [01:18<00:01, 89.57batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6923/7035 [01:18<00:01, 89.99batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6933/7035 [01:18<00:01, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6942/7035 [01:18<00:01, 89.62batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6951/7035 [01:18<00:00, 89.16batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6961/7035 [01:18<00:00, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6970/7035 [01:18<00:00, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6979/7035 [01:18<00:00, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6988/7035 [01:18<00:00, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6997/7035 [01:18<00:00, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7006/7035 [01:19<00:00, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7015/7035 [01:19<00:00, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7025/7035 [01:19<00:00, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7034/7035 [01:19<00:00, 89.02batch/s]\u001b[A\n",
"Training epochs on cuda: 80%|████████ | 4/5 [05:18<01:19, 79.68s/epoch, loss=0.00134, prev_loss=0.00137]\n",
"Training batches on cuda: 0%| | 0/7035 [00:00<?, ?batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 1/7035 [00:00<14:48, 7.92batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 10/7035 [00:00<10:44, 10.90batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 20/7035 [00:00<07:53, 14.81batch/s]\u001b[A\n",
"Training batches on cuda: 0%| | 29/7035 [00:00<05:54, 19.76batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 38/7035 [00:00<04:31, 25.80batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 48/7035 [00:00<03:32, 32.87batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 58/7035 [00:00<02:51, 40.63batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 67/7035 [00:00<02:23, 48.63batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 76/7035 [00:00<02:03, 56.20batch/s]\u001b[A\n",
"Training batches on cuda: 1%| | 85/7035 [00:01<01:49, 63.19batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 94/7035 [00:01<01:40, 69.30batch/s]\u001b[A\n",
"Training batches on cuda: 1%|▏ | 103/7035 [00:01<01:33, 73.84batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 112/7035 [00:01<01:28, 78.02batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 121/7035 [00:01<01:26, 80.24batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 130/7035 [00:01<01:23, 82.63batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 139/7035 [00:01<01:21, 84.45batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 148/7035 [00:01<01:20, 85.71batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 157/7035 [00:01<01:19, 86.61batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 166/7035 [00:01<01:18, 86.98batch/s]\u001b[A\n",
"Training batches on cuda: 2%|▏ | 175/7035 [00:02<01:18, 87.73batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 184/7035 [00:02<01:17, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 193/7035 [00:02<01:17, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 202/7035 [00:02<01:16, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 211/7035 [00:02<01:18, 86.56batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 220/7035 [00:02<01:17, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 229/7035 [00:02<01:17, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 3%|▎ | 238/7035 [00:02<01:17, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 248/7035 [00:02<01:16, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▎ | 257/7035 [00:03<01:16, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 267/7035 [00:03<01:15, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 276/7035 [00:03<01:15, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 285/7035 [00:03<01:15, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 294/7035 [00:03<01:15, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 303/7035 [00:03<01:15, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 4%|▍ | 312/7035 [00:03<01:15, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 321/7035 [00:03<01:15, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 330/7035 [00:03<01:15, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 339/7035 [00:03<01:14, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▍ | 348/7035 [00:04<01:14, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 358/7035 [00:04<01:14, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 367/7035 [00:04<01:14, 89.83batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 376/7035 [00:04<01:14, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 5%|▌ | 385/7035 [00:04<01:14, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 394/7035 [00:04<01:14, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 403/7035 [00:04<01:14, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 412/7035 [00:04<01:14, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 421/7035 [00:04<01:14, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 430/7035 [00:04<01:14, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▌ | 439/7035 [00:05<01:14, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 448/7035 [00:05<01:14, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 6%|▋ | 457/7035 [00:05<01:14, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 466/7035 [00:05<01:13, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 475/7035 [00:05<01:13, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 484/7035 [00:05<01:14, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 493/7035 [00:05<01:14, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 502/7035 [00:05<01:14, 87.91batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 511/7035 [00:05<01:14, 87.64batch/s]\u001b[A\n",
"Training batches on cuda: 7%|▋ | 520/7035 [00:05<01:13, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 529/7035 [00:06<01:13, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 539/7035 [00:06<01:12, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 549/7035 [00:06<01:12, 89.90batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 558/7035 [00:06<01:12, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 567/7035 [00:06<01:12, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 577/7035 [00:06<01:12, 89.41batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 586/7035 [00:06<01:12, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 8%|▊ | 595/7035 [00:06<01:12, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 604/7035 [00:06<01:12, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▊ | 613/7035 [00:07<01:11, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 622/7035 [00:07<01:12, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 632/7035 [00:07<01:11, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 642/7035 [00:07<01:11, 89.85batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 651/7035 [00:07<01:11, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 9%|▉ | 660/7035 [00:07<01:11, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 669/7035 [00:07<01:11, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 678/7035 [00:07<01:11, 89.17batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 687/7035 [00:07<01:11, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 10%|▉ | 696/7035 [00:07<01:11, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 705/7035 [00:08<01:11, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 714/7035 [00:08<01:11, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 723/7035 [00:08<01:11, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 10%|█ | 732/7035 [00:08<01:10, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 741/7035 [00:08<01:10, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 750/7035 [00:08<01:11, 87.44batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 759/7035 [00:08<01:11, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 768/7035 [00:08<01:11, 87.97batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 777/7035 [00:08<01:10, 88.15batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█ | 786/7035 [00:08<01:10, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 795/7035 [00:09<01:11, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 11%|█▏ | 804/7035 [00:09<01:10, 88.42batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 814/7035 [00:09<01:09, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 823/7035 [00:09<01:09, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 833/7035 [00:09<01:09, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 843/7035 [00:09<01:08, 89.88batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 852/7035 [00:09<01:08, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 861/7035 [00:09<01:09, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 870/7035 [00:09<01:08, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 12%|█▏ | 879/7035 [00:09<01:09, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 888/7035 [00:10<01:09, 88.09batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 897/7035 [00:10<01:09, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 906/7035 [00:10<01:09, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 916/7035 [00:10<01:08, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 926/7035 [00:10<01:08, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 936/7035 [00:10<01:07, 90.18batch/s]\u001b[A\n",
"Training batches on cuda: 13%|█▎ | 946/7035 [00:10<01:07, 90.51batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 956/7035 [00:10<01:07, 90.48batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▎ | 966/7035 [00:10<01:07, 89.95batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 975/7035 [00:11<01:07, 89.95batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 984/7035 [00:11<01:07, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 993/7035 [00:11<01:07, 89.81batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1002/7035 [00:11<01:07, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1011/7035 [00:11<01:07, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 14%|█▍ | 1020/7035 [00:11<01:07, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1029/7035 [00:11<01:07, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1039/7035 [00:11<01:07, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▍ | 1048/7035 [00:11<01:07, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1057/7035 [00:11<01:06, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1066/7035 [00:12<01:07, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1075/7035 [00:12<01:07, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 15%|█▌ | 1084/7035 [00:12<01:07, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1093/7035 [00:12<01:07, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1102/7035 [00:12<01:06, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1111/7035 [00:12<01:06, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1120/7035 [00:12<01:07, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1129/7035 [00:12<01:06, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▌ | 1138/7035 [00:12<01:06, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1148/7035 [00:13<01:05, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 16%|█▋ | 1157/7035 [00:13<01:05, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1166/7035 [00:13<01:05, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1175/7035 [00:13<01:05, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1185/7035 [00:13<01:05, 89.71batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1194/7035 [00:13<01:05, 89.66batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1203/7035 [00:13<01:05, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1212/7035 [00:13<01:05, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1221/7035 [00:13<01:04, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 17%|█▋ | 1230/7035 [00:13<01:05, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1239/7035 [00:14<01:05, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1248/7035 [00:14<01:05, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1257/7035 [00:14<01:05, 87.62batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1266/7035 [00:14<01:05, 87.76batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1275/7035 [00:14<01:05, 87.29batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1284/7035 [00:14<01:05, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 18%|█▊ | 1294/7035 [00:14<01:04, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1303/7035 [00:14<01:04, 88.62batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▊ | 1312/7035 [00:14<01:04, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1322/7035 [00:14<01:03, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1331/7035 [00:15<01:03, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1340/7035 [00:15<01:04, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1349/7035 [00:15<01:03, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1358/7035 [00:15<01:03, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 19%|█▉ | 1367/7035 [00:15<01:03, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1376/7035 [00:15<01:03, 88.65batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1385/7035 [00:15<01:03, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1394/7035 [00:15<01:03, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 20%|█▉ | 1403/7035 [00:15<01:03, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1413/7035 [00:15<01:02, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1422/7035 [00:16<01:02, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1431/7035 [00:16<01:03, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 20%|██ | 1440/7035 [00:16<01:02, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1450/7035 [00:16<01:02, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1460/7035 [00:16<01:01, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1470/7035 [00:16<01:01, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1480/7035 [00:16<01:01, 90.10batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██ | 1490/7035 [00:16<01:01, 89.99batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1500/7035 [00:16<01:01, 89.83batch/s]\u001b[A\n",
"Training batches on cuda: 21%|██▏ | 1509/7035 [00:17<01:01, 89.78batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1518/7035 [00:17<01:01, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1527/7035 [00:17<01:01, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1537/7035 [00:17<01:01, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1546/7035 [00:17<01:02, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1555/7035 [00:17<01:01, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1564/7035 [00:17<01:01, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 22%|██▏ | 1574/7035 [00:17<01:01, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1583/7035 [00:17<01:01, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1592/7035 [00:17<01:01, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1601/7035 [00:18<01:01, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1610/7035 [00:18<01:01, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1619/7035 [00:18<01:01, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1628/7035 [00:18<01:00, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1638/7035 [00:18<01:00, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 23%|██▎ | 1647/7035 [00:18<01:00, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1656/7035 [00:18<01:00, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▎ | 1665/7035 [00:18<01:00, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1674/7035 [00:18<01:00, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1683/7035 [00:19<01:00, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1692/7035 [00:19<01:00, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1701/7035 [00:19<01:00, 88.28batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1710/7035 [00:19<01:00, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 24%|██▍ | 1719/7035 [00:19<00:59, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1729/7035 [00:19<00:59, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1738/7035 [00:19<00:59, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1747/7035 [00:19<00:59, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▍ | 1757/7035 [00:19<00:59, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1767/7035 [00:19<00:58, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1777/7035 [00:20<00:58, 89.93batch/s]\u001b[A\n",
"Training batches on cuda: 25%|██▌ | 1786/7035 [00:20<00:58, 89.86batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1795/7035 [00:20<00:58, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1805/7035 [00:20<00:58, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1815/7035 [00:20<00:58, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1825/7035 [00:20<00:57, 90.01batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1835/7035 [00:20<00:58, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▌ | 1844/7035 [00:20<00:58, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1853/7035 [00:20<00:58, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 26%|██▋ | 1862/7035 [00:21<00:58, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1871/7035 [00:21<00:58, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1880/7035 [00:21<00:58, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1889/7035 [00:21<00:57, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1898/7035 [00:21<00:57, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1907/7035 [00:21<00:57, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1916/7035 [00:21<00:57, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1925/7035 [00:21<00:57, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 27%|██▋ | 1934/7035 [00:21<00:57, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1943/7035 [00:21<00:57, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1952/7035 [00:22<00:56, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1961/7035 [00:22<00:56, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1970/7035 [00:22<00:57, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1979/7035 [00:22<00:57, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1988/7035 [00:22<00:56, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 28%|██▊ | 1997/7035 [00:22<00:56, 88.53batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2006/7035 [00:22<00:56, 88.34batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▊ | 2015/7035 [00:22<00:56, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2024/7035 [00:22<00:56, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2033/7035 [00:22<00:56, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2043/7035 [00:23<00:56, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2052/7035 [00:23<00:56, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2061/7035 [00:23<00:56, 88.40batch/s]\u001b[A\n",
"Training batches on cuda: 29%|██▉ | 2070/7035 [00:23<00:55, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2080/7035 [00:23<00:55, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2089/7035 [00:23<00:55, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2098/7035 [00:23<00:55, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 30%|██▉ | 2107/7035 [00:23<00:55, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2116/7035 [00:23<00:55, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2125/7035 [00:23<00:55, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2134/7035 [00:24<00:55, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 30%|███ | 2143/7035 [00:24<00:54, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2152/7035 [00:24<00:55, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2161/7035 [00:24<00:55, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2170/7035 [00:24<00:54, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2179/7035 [00:24<00:54, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2188/7035 [00:24<00:54, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███ | 2197/7035 [00:24<00:54, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2206/7035 [00:24<00:54, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 31%|███▏ | 2215/7035 [00:25<00:54, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2225/7035 [00:25<00:53, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2235/7035 [00:25<00:53, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2244/7035 [00:25<00:53, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2253/7035 [00:25<00:53, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2262/7035 [00:25<00:53, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2271/7035 [00:25<00:53, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 32%|███▏ | 2280/7035 [00:25<00:54, 88.02batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2289/7035 [00:25<00:53, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2299/7035 [00:25<00:53, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2308/7035 [00:26<00:53, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2318/7035 [00:26<00:52, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2327/7035 [00:26<00:52, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2336/7035 [00:26<00:52, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2345/7035 [00:26<00:52, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 33%|███▎ | 2354/7035 [00:26<00:52, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2363/7035 [00:26<00:52, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▎ | 2372/7035 [00:26<00:52, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2381/7035 [00:26<00:52, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2390/7035 [00:26<00:52, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2400/7035 [00:27<00:51, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2410/7035 [00:27<00:51, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 34%|███▍ | 2420/7035 [00:27<00:51, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2430/7035 [00:27<00:51, 90.01batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2440/7035 [00:27<00:51, 89.80batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2450/7035 [00:27<00:50, 90.04batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▍ | 2460/7035 [00:27<00:50, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2469/7035 [00:27<00:51, 88.71batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2478/7035 [00:27<00:51, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2487/7035 [00:28<00:51, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 35%|███▌ | 2496/7035 [00:28<00:51, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2506/7035 [00:28<00:50, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2515/7035 [00:28<00:51, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2525/7035 [00:28<00:50, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2535/7035 [00:28<00:50, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▌ | 2545/7035 [00:28<00:49, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2555/7035 [00:28<00:49, 89.98batch/s]\u001b[A\n",
"Training batches on cuda: 36%|███▋ | 2564/7035 [00:28<00:49, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2573/7035 [00:29<00:49, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2582/7035 [00:29<00:49, 89.62batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2591/7035 [00:29<00:49, 89.64batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2600/7035 [00:29<00:49, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2609/7035 [00:29<00:49, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2618/7035 [00:29<00:49, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2627/7035 [00:29<00:49, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 37%|███▋ | 2636/7035 [00:29<00:49, 88.60batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2646/7035 [00:29<00:49, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2655/7035 [00:29<00:49, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2664/7035 [00:30<00:49, 87.74batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2673/7035 [00:30<00:49, 88.14batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2683/7035 [00:30<00:48, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2692/7035 [00:30<00:48, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 38%|███▊ | 2701/7035 [00:30<00:48, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2710/7035 [00:30<00:49, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▊ | 2719/7035 [00:30<00:48, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2728/7035 [00:30<00:48, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2737/7035 [00:30<00:48, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2746/7035 [00:30<00:48, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2755/7035 [00:31<00:48, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2764/7035 [00:31<00:48, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 39%|███▉ | 2773/7035 [00:31<00:48, 88.45batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2782/7035 [00:31<00:48, 87.27batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2791/7035 [00:31<00:48, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2800/7035 [00:31<00:48, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 40%|███▉ | 2809/7035 [00:31<00:47, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2818/7035 [00:31<00:47, 88.13batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2827/7035 [00:31<00:47, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2836/7035 [00:31<00:47, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 40%|████ | 2845/7035 [00:32<00:47, 87.78batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2854/7035 [00:32<00:47, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2863/7035 [00:32<00:47, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2872/7035 [00:32<00:47, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2881/7035 [00:32<00:47, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2890/7035 [00:32<00:46, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████ | 2900/7035 [00:32<00:46, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2909/7035 [00:32<00:46, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 41%|████▏ | 2918/7035 [00:32<00:46, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2927/7035 [00:33<00:46, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2936/7035 [00:33<00:46, 87.58batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2945/7035 [00:33<00:46, 87.53batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2954/7035 [00:33<00:46, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2963/7035 [00:33<00:46, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2972/7035 [00:33<00:45, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 42%|████▏ | 2981/7035 [00:33<00:45, 88.50batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2990/7035 [00:33<00:45, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 2999/7035 [00:33<00:45, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3008/7035 [00:33<00:45, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3017/7035 [00:34<00:45, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3026/7035 [00:34<00:45, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3035/7035 [00:34<00:45, 88.68batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3045/7035 [00:34<00:44, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 43%|████▎ | 3054/7035 [00:34<00:44, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3063/7035 [00:34<00:44, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▎ | 3072/7035 [00:34<00:44, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3081/7035 [00:34<00:44, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3090/7035 [00:34<00:44, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3099/7035 [00:34<00:44, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3108/7035 [00:35<00:44, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3117/7035 [00:35<00:44, 87.09batch/s]\u001b[A\n",
"Training batches on cuda: 44%|████▍ | 3127/7035 [00:35<00:44, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3136/7035 [00:35<00:44, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3145/7035 [00:35<00:44, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3154/7035 [00:35<00:44, 87.85batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▍ | 3164/7035 [00:35<00:43, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3173/7035 [00:35<00:43, 88.85batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3182/7035 [00:35<00:43, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3191/7035 [00:35<00:43, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 45%|████▌ | 3200/7035 [00:36<00:43, 87.75batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3209/7035 [00:36<00:43, 87.89batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3218/7035 [00:36<00:43, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3227/7035 [00:36<00:42, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3236/7035 [00:36<00:42, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▌ | 3245/7035 [00:36<00:42, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3254/7035 [00:36<00:42, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 46%|████▋ | 3263/7035 [00:36<00:42, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3272/7035 [00:36<00:42, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3281/7035 [00:37<00:42, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3290/7035 [00:37<00:42, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3299/7035 [00:37<00:42, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3308/7035 [00:37<00:42, 88.21batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3317/7035 [00:37<00:42, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3326/7035 [00:37<00:41, 88.52batch/s]\u001b[A\n",
"Training batches on cuda: 47%|████▋ | 3335/7035 [00:37<00:42, 87.65batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3344/7035 [00:37<00:42, 87.49batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3354/7035 [00:37<00:41, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3363/7035 [00:37<00:41, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3373/7035 [00:38<00:41, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3382/7035 [00:38<00:41, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3391/7035 [00:38<00:40, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3401/7035 [00:38<00:40, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 48%|████▊ | 3410/7035 [00:38<00:40, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3419/7035 [00:38<00:40, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▊ | 3428/7035 [00:38<00:40, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3437/7035 [00:38<00:40, 88.73batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3446/7035 [00:38<00:40, 88.67batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3455/7035 [00:38<00:40, 88.78batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3464/7035 [00:39<00:40, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3473/7035 [00:39<00:40, 87.47batch/s]\u001b[A\n",
"Training batches on cuda: 49%|████▉ | 3482/7035 [00:39<00:40, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3492/7035 [00:39<00:39, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3501/7035 [00:39<00:39, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 50%|████▉ | 3510/7035 [00:39<00:40, 88.08batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3519/7035 [00:39<00:39, 88.01batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3528/7035 [00:39<00:39, 87.81batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3537/7035 [00:39<00:39, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 50%|█████ | 3547/7035 [00:40<00:39, 88.91batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3557/7035 [00:40<00:38, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3567/7035 [00:40<00:38, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3577/7035 [00:40<00:38, 89.87batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3586/7035 [00:40<00:38, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████ | 3596/7035 [00:40<00:38, 90.07batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3606/7035 [00:40<00:37, 90.32batch/s]\u001b[A\n",
"Training batches on cuda: 51%|█████▏ | 3616/7035 [00:40<00:37, 90.20batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3626/7035 [00:40<00:38, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3635/7035 [00:40<00:38, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3644/7035 [00:41<00:38, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3653/7035 [00:41<00:38, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3662/7035 [00:41<00:38, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3671/7035 [00:41<00:37, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3680/7035 [00:41<00:37, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 52%|█████▏ | 3689/7035 [00:41<00:37, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3698/7035 [00:41<00:37, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3708/7035 [00:41<00:37, 89.57batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3717/7035 [00:41<00:37, 89.34batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3726/7035 [00:42<00:37, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3735/7035 [00:42<00:36, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3744/7035 [00:42<00:37, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3753/7035 [00:42<00:37, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 53%|█████▎ | 3763/7035 [00:42<00:36, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3772/7035 [00:42<00:36, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▎ | 3781/7035 [00:42<00:36, 88.55batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3790/7035 [00:42<00:36, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3799/7035 [00:42<00:36, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3808/7035 [00:42<00:36, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3817/7035 [00:43<00:36, 89.10batch/s]\u001b[A\n",
"Training batches on cuda: 54%|█████▍ | 3826/7035 [00:43<00:35, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3835/7035 [00:43<00:35, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3844/7035 [00:43<00:35, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3853/7035 [00:43<00:35, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▍ | 3862/7035 [00:43<00:35, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3872/7035 [00:43<00:35, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3882/7035 [00:43<00:35, 89.66batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3892/7035 [00:43<00:34, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 55%|█████▌ | 3901/7035 [00:43<00:34, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3910/7035 [00:44<00:34, 89.73batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3919/7035 [00:44<00:34, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3928/7035 [00:44<00:34, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3937/7035 [00:44<00:34, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3946/7035 [00:44<00:34, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▌ | 3955/7035 [00:44<00:34, 88.96batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3964/7035 [00:44<00:34, 89.02batch/s]\u001b[A\n",
"Training batches on cuda: 56%|█████▋ | 3973/7035 [00:44<00:34, 88.89batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3982/7035 [00:44<00:34, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 3992/7035 [00:45<00:34, 89.48batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4001/7035 [00:45<00:34, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4010/7035 [00:45<00:33, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4019/7035 [00:45<00:34, 87.55batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4028/7035 [00:45<00:34, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 57%|█████▋ | 4037/7035 [00:45<00:33, 88.38batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4046/7035 [00:45<00:33, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4055/7035 [00:45<00:33, 88.32batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4065/7035 [00:45<00:33, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4074/7035 [00:45<00:33, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4084/7035 [00:46<00:32, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4093/7035 [00:46<00:32, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4102/7035 [00:46<00:32, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 58%|█████▊ | 4111/7035 [00:46<00:32, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4121/7035 [00:46<00:32, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▊ | 4130/7035 [00:46<00:32, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4139/7035 [00:46<00:32, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4148/7035 [00:46<00:32, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4157/7035 [00:46<00:32, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4166/7035 [00:46<00:32, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4175/7035 [00:47<00:32, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 59%|█████▉ | 4184/7035 [00:47<00:32, 89.01batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4193/7035 [00:47<00:31, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4202/7035 [00:47<00:31, 88.84batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4211/7035 [00:47<00:31, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 60%|█████▉ | 4220/7035 [00:47<00:31, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4229/7035 [00:47<00:31, 88.77batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4238/7035 [00:47<00:31, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 60%|██████ | 4247/7035 [00:47<00:31, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4257/7035 [00:47<00:31, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4267/7035 [00:48<00:30, 89.71batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4276/7035 [00:48<00:30, 89.40batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4286/7035 [00:48<00:30, 89.89batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4295/7035 [00:48<00:30, 88.74batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████ | 4304/7035 [00:48<00:30, 88.88batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4313/7035 [00:48<00:30, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 61%|██████▏ | 4322/7035 [00:48<00:30, 88.29batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4331/7035 [00:48<00:30, 87.99batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4340/7035 [00:48<00:30, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4349/7035 [00:49<00:30, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4359/7035 [00:49<00:29, 89.38batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4368/7035 [00:49<00:29, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4377/7035 [00:49<00:30, 88.58batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4386/7035 [00:49<00:30, 87.70batch/s]\u001b[A\n",
"Training batches on cuda: 62%|██████▏ | 4395/7035 [00:49<00:29, 88.03batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4404/7035 [00:49<00:29, 88.41batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4413/7035 [00:49<00:29, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4422/7035 [00:49<00:29, 88.54batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4431/7035 [00:49<00:29, 88.75batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4440/7035 [00:50<00:29, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4450/7035 [00:50<00:28, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 63%|██████▎ | 4459/7035 [00:50<00:28, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4468/7035 [00:50<00:28, 88.63batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▎ | 4477/7035 [00:50<00:28, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4486/7035 [00:50<00:28, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4495/7035 [00:50<00:28, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4504/7035 [00:50<00:28, 88.31batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4513/7035 [00:50<00:28, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4522/7035 [00:50<00:28, 87.94batch/s]\u001b[A\n",
"Training batches on cuda: 64%|██████▍ | 4531/7035 [00:51<00:28, 88.47batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4540/7035 [00:51<00:28, 88.18batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4549/7035 [00:51<00:28, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4558/7035 [00:51<00:27, 88.51batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▍ | 4567/7035 [00:51<00:28, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4576/7035 [00:51<00:27, 87.92batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4585/7035 [00:51<00:27, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4595/7035 [00:51<00:27, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 65%|██████▌ | 4604/7035 [00:51<00:27, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4614/7035 [00:52<00:26, 89.77batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4624/7035 [00:52<00:26, 90.15batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4634/7035 [00:52<00:26, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4643/7035 [00:52<00:26, 89.11batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▌ | 4653/7035 [00:52<00:26, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4662/7035 [00:52<00:26, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 66%|██████▋ | 4671/7035 [00:52<00:26, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4681/7035 [00:52<00:26, 89.64batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4690/7035 [00:52<00:26, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4699/7035 [00:52<00:26, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4709/7035 [00:53<00:26, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4718/7035 [00:53<00:25, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4727/7035 [00:53<00:25, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4736/7035 [00:53<00:25, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 67%|██████▋ | 4745/7035 [00:53<00:26, 87.96batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4754/7035 [00:53<00:26, 87.72batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4763/7035 [00:53<00:25, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4772/7035 [00:53<00:25, 88.27batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4781/7035 [00:53<00:25, 88.10batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4790/7035 [00:53<00:25, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4799/7035 [00:54<00:25, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4808/7035 [00:54<00:25, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 68%|██████▊ | 4818/7035 [00:54<00:24, 89.50batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4827/7035 [00:54<00:24, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▊ | 4836/7035 [00:54<00:24, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4845/7035 [00:54<00:24, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4855/7035 [00:54<00:24, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4864/7035 [00:54<00:24, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4874/7035 [00:54<00:24, 89.57batch/s]\u001b[A\n",
"Training batches on cuda: 69%|██████▉ | 4883/7035 [00:55<00:24, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4893/7035 [00:55<00:23, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4902/7035 [00:55<00:23, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4911/7035 [00:55<00:23, 89.09batch/s]\u001b[A\n",
"Training batches on cuda: 70%|██████▉ | 4921/7035 [00:55<00:23, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4930/7035 [00:55<00:23, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4939/7035 [00:55<00:23, 88.07batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4948/7035 [00:55<00:23, 88.39batch/s]\u001b[A\n",
"Training batches on cuda: 70%|███████ | 4957/7035 [00:55<00:23, 88.69batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4966/7035 [00:55<00:23, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4975/7035 [00:56<00:23, 88.80batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4985/7035 [00:56<00:22, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 4995/7035 [00:56<00:22, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████ | 5005/7035 [00:56<00:22, 90.07batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5015/7035 [00:56<00:22, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 71%|███████▏ | 5024/7035 [00:56<00:22, 88.64batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5034/7035 [00:56<00:22, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5043/7035 [00:56<00:22, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5052/7035 [00:56<00:22, 89.47batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5062/7035 [00:57<00:21, 89.84batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5072/7035 [00:57<00:21, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5082/7035 [00:57<00:21, 90.14batch/s]\u001b[A\n",
"Training batches on cuda: 72%|███████▏ | 5092/7035 [00:57<00:21, 90.16batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5102/7035 [00:57<00:21, 89.55batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5111/7035 [00:57<00:21, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5120/7035 [00:57<00:21, 89.07batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5129/7035 [00:57<00:21, 89.18batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5138/7035 [00:57<00:21, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5147/7035 [00:57<00:21, 88.70batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5157/7035 [00:58<00:21, 89.25batch/s]\u001b[A\n",
"Training batches on cuda: 73%|███████▎ | 5167/7035 [00:58<00:20, 89.72batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5177/7035 [00:58<00:20, 90.07batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▎ | 5187/7035 [00:58<00:20, 90.25batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5197/7035 [00:58<00:20, 90.27batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5207/7035 [00:58<00:20, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5216/7035 [00:58<00:20, 89.51batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5225/7035 [00:58<00:20, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 74%|███████▍ | 5234/7035 [00:58<00:20, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5243/7035 [00:59<00:20, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5252/7035 [00:59<00:20, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5262/7035 [00:59<00:19, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▍ | 5272/7035 [00:59<00:19, 90.06batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5282/7035 [00:59<00:19, 90.08batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5292/7035 [00:59<00:19, 90.16batch/s]\u001b[A\n",
"Training batches on cuda: 75%|███████▌ | 5302/7035 [00:59<00:19, 90.12batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5312/7035 [00:59<00:19, 90.24batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5322/7035 [00:59<00:19, 90.13batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5332/7035 [01:00<00:18, 90.40batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5342/7035 [01:00<00:18, 90.36batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5352/7035 [01:00<00:18, 90.08batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▌ | 5362/7035 [01:00<00:18, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5371/7035 [01:00<00:18, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 76%|███████▋ | 5380/7035 [01:00<00:18, 89.13batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5389/7035 [01:00<00:18, 88.79batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5398/7035 [01:00<00:18, 89.00batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5407/7035 [01:00<00:18, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5416/7035 [01:00<00:18, 87.69batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5425/7035 [01:01<00:18, 88.30batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5434/7035 [01:01<00:18, 88.25batch/s]\u001b[A\n",
"Training batches on cuda: 77%|███████▋ | 5444/7035 [01:01<00:17, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5454/7035 [01:01<00:17, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5464/7035 [01:01<00:17, 90.02batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5474/7035 [01:01<00:17, 89.54batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5483/7035 [01:01<00:17, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5492/7035 [01:01<00:17, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5501/7035 [01:01<00:17, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5510/7035 [01:02<00:17, 89.15batch/s]\u001b[A\n",
"Training batches on cuda: 78%|███████▊ | 5519/7035 [01:02<00:16, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5528/7035 [01:02<00:16, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▊ | 5537/7035 [01:02<00:16, 89.35batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5546/7035 [01:02<00:16, 89.33batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5555/7035 [01:02<00:16, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5564/7035 [01:02<00:16, 87.63batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5573/7035 [01:02<00:16, 87.18batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5582/7035 [01:02<00:16, 86.60batch/s]\u001b[A\n",
"Training batches on cuda: 79%|███████▉ | 5592/7035 [01:02<00:16, 87.88batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5601/7035 [01:03<00:16, 88.20batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5611/7035 [01:03<00:16, 88.98batch/s]\u001b[A\n",
"Training batches on cuda: 80%|███████▉ | 5620/7035 [01:03<00:15, 88.66batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5630/7035 [01:03<00:15, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5639/7035 [01:03<00:15, 89.30batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5648/7035 [01:03<00:15, 88.82batch/s]\u001b[A\n",
"Training batches on cuda: 80%|████████ | 5657/7035 [01:03<00:15, 89.03batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5666/7035 [01:03<00:16, 84.92batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5675/7035 [01:03<00:15, 86.21batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5685/7035 [01:04<00:15, 87.42batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5695/7035 [01:04<00:15, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5705/7035 [01:04<00:14, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████ | 5714/7035 [01:04<00:14, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 81%|████████▏ | 5724/7035 [01:04<00:14, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5734/7035 [01:04<00:14, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5743/7035 [01:04<00:14, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5752/7035 [01:04<00:14, 88.92batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5761/7035 [01:04<00:14, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5771/7035 [01:04<00:14, 89.45batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5780/7035 [01:05<00:14, 89.60batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5789/7035 [01:05<00:13, 89.32batch/s]\u001b[A\n",
"Training batches on cuda: 82%|████████▏ | 5799/7035 [01:05<00:13, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5809/7035 [01:05<00:13, 89.91batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5819/7035 [01:05<00:13, 90.17batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5829/7035 [01:05<00:13, 89.91batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5838/7035 [01:05<00:13, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5848/7035 [01:05<00:13, 89.68batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5858/7035 [01:05<00:13, 90.11batch/s]\u001b[A\n",
"Training batches on cuda: 83%|████████▎ | 5868/7035 [01:06<00:13, 89.69batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5878/7035 [01:06<00:12, 90.19batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▎ | 5888/7035 [01:06<00:12, 90.38batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5898/7035 [01:06<00:12, 90.55batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5908/7035 [01:06<00:12, 90.68batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5918/7035 [01:06<00:12, 90.71batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5928/7035 [01:06<00:12, 90.38batch/s]\u001b[A\n",
"Training batches on cuda: 84%|████████▍ | 5938/7035 [01:06<00:12, 90.21batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5948/7035 [01:06<00:12, 90.27batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5958/7035 [01:07<00:11, 90.45batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5968/7035 [01:07<00:11, 89.91batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▍ | 5978/7035 [01:07<00:11, 90.12batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5988/7035 [01:07<00:11, 89.91batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 5997/7035 [01:07<00:11, 89.76batch/s]\u001b[A\n",
"Training batches on cuda: 85%|████████▌ | 6006/7035 [01:07<00:11, 89.75batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6015/7035 [01:07<00:11, 89.52batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6024/7035 [01:07<00:11, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6033/7035 [01:07<00:11, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6043/7035 [01:08<00:11, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6053/7035 [01:08<00:10, 89.44batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▌ | 6062/7035 [01:08<00:11, 88.37batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6071/7035 [01:08<00:10, 88.16batch/s]\u001b[A\n",
"Training batches on cuda: 86%|████████▋ | 6080/7035 [01:08<00:10, 88.56batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6089/7035 [01:08<00:10, 88.72batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6098/7035 [01:08<00:10, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6107/7035 [01:08<00:10, 87.87batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6116/7035 [01:08<00:10, 87.77batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6125/7035 [01:08<00:10, 88.11batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6134/7035 [01:09<00:10, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6143/7035 [01:09<00:10, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 87%|████████▋ | 6153/7035 [01:09<00:09, 89.39batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6162/7035 [01:09<00:09, 89.04batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6171/7035 [01:09<00:09, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6181/7035 [01:09<00:09, 89.58batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6190/7035 [01:09<00:09, 89.62batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6199/7035 [01:09<00:09, 89.06batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6208/7035 [01:09<00:09, 87.50batch/s]\u001b[A\n",
"Training batches on cuda: 88%|████████▊ | 6217/7035 [01:09<00:09, 87.93batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6226/7035 [01:10<00:09, 88.43batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▊ | 6235/7035 [01:10<00:09, 88.36batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6244/7035 [01:10<00:08, 88.48batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6253/7035 [01:10<00:08, 88.23batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6262/7035 [01:10<00:08, 87.82batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6271/7035 [01:10<00:08, 88.19batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6281/7035 [01:10<00:08, 88.94batch/s]\u001b[A\n",
"Training batches on cuda: 89%|████████▉ | 6290/7035 [01:10<00:08, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6300/7035 [01:10<00:08, 89.19batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6309/7035 [01:11<00:08, 88.90batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6319/7035 [01:11<00:08, 89.43batch/s]\u001b[A\n",
"Training batches on cuda: 90%|████████▉ | 6328/7035 [01:11<00:07, 89.56batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6337/7035 [01:11<00:07, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6346/7035 [01:11<00:07, 89.49batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6355/7035 [01:11<00:07, 89.42batch/s]\u001b[A\n",
"Training batches on cuda: 90%|█████████ | 6365/7035 [01:11<00:07, 89.83batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6375/7035 [01:11<00:07, 90.12batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6385/7035 [01:11<00:07, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6394/7035 [01:11<00:07, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6404/7035 [01:12<00:07, 89.70batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████ | 6414/7035 [01:12<00:06, 90.09batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6424/7035 [01:12<00:06, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 91%|█████████▏| 6434/7035 [01:12<00:06, 90.03batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6444/7035 [01:12<00:06, 90.15batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6454/7035 [01:12<00:06, 89.82batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6463/7035 [01:12<00:06, 88.95batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6472/7035 [01:12<00:06, 88.22batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6481/7035 [01:12<00:06, 88.46batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6490/7035 [01:13<00:06, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 92%|█████████▏| 6499/7035 [01:13<00:06, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6508/7035 [01:13<00:05, 89.21batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6517/7035 [01:13<00:05, 89.29batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6526/7035 [01:13<00:05, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6535/7035 [01:13<00:05, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6544/7035 [01:13<00:05, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6553/7035 [01:13<00:05, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6562/7035 [01:13<00:05, 88.35batch/s]\u001b[A\n",
"Training batches on cuda: 93%|█████████▎| 6571/7035 [01:13<00:05, 88.59batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6580/7035 [01:14<00:05, 88.83batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▎| 6589/7035 [01:14<00:05, 89.14batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6598/7035 [01:14<00:04, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6607/7035 [01:14<00:04, 89.36batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6616/7035 [01:14<00:04, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6625/7035 [01:14<00:04, 88.97batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6634/7035 [01:14<00:04, 89.24batch/s]\u001b[A\n",
"Training batches on cuda: 94%|█████████▍| 6643/7035 [01:14<00:04, 89.28batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6652/7035 [01:14<00:04, 89.27batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6661/7035 [01:14<00:04, 87.37batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6671/7035 [01:15<00:04, 88.26batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▍| 6681/7035 [01:15<00:03, 88.99batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6690/7035 [01:15<00:03, 89.08batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6700/7035 [01:15<00:03, 89.48batch/s]\u001b[A\n",
"Training batches on cuda: 95%|█████████▌| 6710/7035 [01:15<00:03, 89.85batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6719/7035 [01:15<00:03, 89.85batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6728/7035 [01:15<00:03, 89.12batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6737/7035 [01:15<00:03, 88.87batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6746/7035 [01:15<00:03, 88.24batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6756/7035 [01:16<00:03, 88.93batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▌| 6765/7035 [01:16<00:03, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6775/7035 [01:16<00:02, 89.53batch/s]\u001b[A\n",
"Training batches on cuda: 96%|█████████▋| 6784/7035 [01:16<00:02, 89.61batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6794/7035 [01:16<00:02, 90.01batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6804/7035 [01:16<00:02, 90.13batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6814/7035 [01:16<00:02, 90.00batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6824/7035 [01:16<00:02, 90.01batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6834/7035 [01:16<00:02, 89.31batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6843/7035 [01:16<00:02, 88.57batch/s]\u001b[A\n",
"Training batches on cuda: 97%|█████████▋| 6852/7035 [01:17<00:02, 88.86batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6861/7035 [01:17<00:01, 89.05batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6870/7035 [01:17<00:01, 88.44batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6879/7035 [01:17<00:01, 88.76batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6889/7035 [01:17<00:01, 89.23batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6898/7035 [01:17<00:01, 89.20batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6908/7035 [01:17<00:01, 89.65batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6918/7035 [01:17<00:01, 89.90batch/s]\u001b[A\n",
"Training batches on cuda: 98%|█████████▊| 6927/7035 [01:17<00:01, 88.81batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6936/7035 [01:18<00:01, 88.61batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▊| 6946/7035 [01:18<00:00, 89.26batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6955/7035 [01:18<00:00, 89.46batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6965/7035 [01:18<00:00, 89.90batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6974/7035 [01:18<00:00, 89.67batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6984/7035 [01:18<00:00, 90.04batch/s]\u001b[A\n",
"Training batches on cuda: 99%|█████████▉| 6994/7035 [01:18<00:00, 90.29batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7004/7035 [01:18<00:00, 89.98batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7013/7035 [01:18<00:00, 89.74batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7022/7035 [01:18<00:00, 89.62batch/s]\u001b[A\n",
"Training batches on cuda: 100%|█████████▉| 7031/7035 [01:19<00:00, 89.48batch/s]\u001b[A\n",
"Training epochs on cuda: 100%|██████████| 5/5 [06:37<00:00, 79.59s/epoch, loss=0.00132, prev_loss=0.00134]\n",
"INFO:pykeen.evaluation.evaluator:Starting batch_size search for evaluation now...\n",
"/usr/local/lib/python3.6/dist-packages/pykeen/evaluation/evaluator.py:379: UserWarning: This overload of nonzero is deprecated:\n",
"\tnonzero()\n",
"Consider using one of the following signatures instead:\n",
"\tnonzero(*, bool as_tuple) (Triggered internally at /pytorch/torch/csrc/utils/python_arg_parser.cpp:766.)\n",
" filter_batch = (entity_filter_test & relation_filter).nonzero()\n",
"INFO:pykeen.evaluation.evaluator:Concluded batch_size search with batch_size=61.\n",
"Evaluating on cuda: 100%|██████████| 122/122 [00:00<00:00, 252triple/s]\n",
"INFO:pykeen.evaluation.evaluator:Evaluation took 0.52s seconds\n"
],
"name": "stderr"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "XnM8jztkbV_o",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1000
},
"outputId": "3268483e-1c27-483c-c48e-eaa3f76383b7"
},
"source": [
"results.metric_results.to_df()"
],
"execution_count": 11,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Side</th>\n",
" <th>Type</th>\n",
" <th>Metric</th>\n",
" <th>Value</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>adjusted_mean_rank</td>\n",
" <td>0.376230</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>adjusted_mean_rank</td>\n",
" <td>0.231138</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>adjusted_mean_rank</td>\n",
" <td>0.303684</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>mean_rank</td>\n",
" <td>8491.758197</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.000654</td>\n",
" </tr>\n",
" <tr>\n",
" <th>5</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>6</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>7</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>8</th>\n",
" <td>head</td>\n",
" <td>avg</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>9</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>mean_rank</td>\n",
" <td>8491.762295</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.000654</td>\n",
" </tr>\n",
" <tr>\n",
" <th>11</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>12</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>13</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>14</th>\n",
" <td>head</td>\n",
" <td>worst</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>15</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>mean_rank</td>\n",
" <td>8491.754098</td>\n",
" </tr>\n",
" <tr>\n",
" <th>16</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.000654</td>\n",
" </tr>\n",
" <tr>\n",
" <th>17</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>18</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>19</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>20</th>\n",
" <td>head</td>\n",
" <td>best</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>21</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>mean_rank</td>\n",
" <td>5218.696721</td>\n",
" </tr>\n",
" <tr>\n",
" <th>22</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001418</td>\n",
" </tr>\n",
" <tr>\n",
" <th>23</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>24</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>25</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>26</th>\n",
" <td>tail</td>\n",
" <td>avg</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>27</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>mean_rank</td>\n",
" <td>5218.696721</td>\n",
" </tr>\n",
" <tr>\n",
" <th>28</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001418</td>\n",
" </tr>\n",
" <tr>\n",
" <th>29</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>30</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>31</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>32</th>\n",
" <td>tail</td>\n",
" <td>worst</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>33</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>mean_rank</td>\n",
" <td>5218.696721</td>\n",
" </tr>\n",
" <tr>\n",
" <th>34</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001418</td>\n",
" </tr>\n",
" <tr>\n",
" <th>35</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>36</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>37</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>38</th>\n",
" <td>tail</td>\n",
" <td>best</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>39</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>mean_rank</td>\n",
" <td>6855.227459</td>\n",
" </tr>\n",
" <tr>\n",
" <th>40</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001036</td>\n",
" </tr>\n",
" <tr>\n",
" <th>41</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>42</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>43</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>44</th>\n",
" <td>both</td>\n",
" <td>avg</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>45</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>mean_rank</td>\n",
" <td>6855.229508</td>\n",
" </tr>\n",
" <tr>\n",
" <th>46</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001036</td>\n",
" </tr>\n",
" <tr>\n",
" <th>47</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>48</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>49</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>50</th>\n",
" <td>both</td>\n",
" <td>worst</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>51</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>mean_rank</td>\n",
" <td>6855.225410</td>\n",
" </tr>\n",
" <tr>\n",
" <th>52</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>mean_reciprocal_rank</td>\n",
" <td>0.001036</td>\n",
" </tr>\n",
" <tr>\n",
" <th>53</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>hits_at_1</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>54</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>hits_at_3</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>55</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>hits_at_5</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" <tr>\n",
" <th>56</th>\n",
" <td>both</td>\n",
" <td>best</td>\n",
" <td>hits_at_10</td>\n",
" <td>0.000000</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Side Type Metric Value\n",
"0 head avg adjusted_mean_rank 0.376230\n",
"1 tail avg adjusted_mean_rank 0.231138\n",
"2 both avg adjusted_mean_rank 0.303684\n",
"3 head avg mean_rank 8491.758197\n",
"4 head avg mean_reciprocal_rank 0.000654\n",
"5 head avg hits_at_1 0.000000\n",
"6 head avg hits_at_3 0.000000\n",
"7 head avg hits_at_5 0.000000\n",
"8 head avg hits_at_10 0.000000\n",
"9 head worst mean_rank 8491.762295\n",
"10 head worst mean_reciprocal_rank 0.000654\n",
"11 head worst hits_at_1 0.000000\n",
"12 head worst hits_at_3 0.000000\n",
"13 head worst hits_at_5 0.000000\n",
"14 head worst hits_at_10 0.000000\n",
"15 head best mean_rank 8491.754098\n",
"16 head best mean_reciprocal_rank 0.000654\n",
"17 head best hits_at_1 0.000000\n",
"18 head best hits_at_3 0.000000\n",
"19 head best hits_at_5 0.000000\n",
"20 head best hits_at_10 0.000000\n",
"21 tail avg mean_rank 5218.696721\n",
"22 tail avg mean_reciprocal_rank 0.001418\n",
"23 tail avg hits_at_1 0.000000\n",
"24 tail avg hits_at_3 0.000000\n",
"25 tail avg hits_at_5 0.000000\n",
"26 tail avg hits_at_10 0.000000\n",
"27 tail worst mean_rank 5218.696721\n",
"28 tail worst mean_reciprocal_rank 0.001418\n",
"29 tail worst hits_at_1 0.000000\n",
"30 tail worst hits_at_3 0.000000\n",
"31 tail worst hits_at_5 0.000000\n",
"32 tail worst hits_at_10 0.000000\n",
"33 tail best mean_rank 5218.696721\n",
"34 tail best mean_reciprocal_rank 0.001418\n",
"35 tail best hits_at_1 0.000000\n",
"36 tail best hits_at_3 0.000000\n",
"37 tail best hits_at_5 0.000000\n",
"38 tail best hits_at_10 0.000000\n",
"39 both avg mean_rank 6855.227459\n",
"40 both avg mean_reciprocal_rank 0.001036\n",
"41 both avg hits_at_1 0.000000\n",
"42 both avg hits_at_3 0.000000\n",
"43 both avg hits_at_5 0.000000\n",
"44 both avg hits_at_10 0.000000\n",
"45 both worst mean_rank 6855.229508\n",
"46 both worst mean_reciprocal_rank 0.001036\n",
"47 both worst hits_at_1 0.000000\n",
"48 both worst hits_at_3 0.000000\n",
"49 both worst hits_at_5 0.000000\n",
"50 both worst hits_at_10 0.000000\n",
"51 both best mean_rank 6855.225410\n",
"52 both best mean_reciprocal_rank 0.001036\n",
"53 both best hits_at_1 0.000000\n",
"54 both best hits_at_3 0.000000\n",
"55 both best hits_at_5 0.000000\n",
"56 both best hits_at_10 0.000000"
]
},
"metadata": {
"tags": []
},
"execution_count": 11
}
]
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment