MegaByte_pytroch.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "authorship_tag": "ABX9TyPhPSQfnkQMshQV1wDb4iPH",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/weedge/doraemon-nb/blob/main/MegaByte_pytroch.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "base: 基础普适研究\n",
        "\n",
        "- hourglass transformers: [2021. Hierarchical Transformers Are More Efficient Language Models](https://arxiv.org/abs/2110.13711) | [lucidrains/simple-hierarchical-transformer](https://github.com/lucidrains/simple-hierarchical-transformer) vanilla layers and shortened layers use GPT  AR  GLM 🤞\n",
        "- [2023.5 **MEGABYTE: Predicting Million-byte Sequences with Multiscale Transformers**](https://arxiv.org/abs/2305.07185) | [lucidrains/MEGABYTE-pytorch](https://github.com/lucidrains/MEGABYTE-pytorch)\n",
        "\n",
        "PS: 想法和自己整理的[simple LM](https://github.com/ai-bot-pro/baby-llm/tree/main/simpleLM)相似~"
      ],
      "metadata": {
        "id": "evNk1AfC-jmf"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# hourglass transformer\n",
        "\n",
        "hourglass transformers: [2021. Hierarchical Transformers Are More Efficient Language Models](https://arxiv.org/abs/2110.13711) | [lucidrains/simple-hierarchical-transformer](https://github.com/lucidrains/simple-hierarchical-transformer) vanilla layers and shortened layers use GPT  AR  GLM 🤞\n"
      ],
      "metadata": {
        "id": "JxqKWTOI-qHX"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!git clone https://github.com/lucidrains/simple-hierarchical-transformer.git"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NlvniYnc-peD",
        "outputId": "c8c63a0d-6a50-43a6-b9ed-ec018087e8f8"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Cloning into 'simple-hierarchical-transformer'...\n",
            "remote: Enumerating objects: 312, done.\u001b[K\n",
            "remote: Counting objects: 100% (71/71), done.\u001b[K\n",
            "remote: Compressing objects: 100% (3/3), done.\u001b[K\n",
            "remote: Total 312 (delta 68), reused 69 (delta 68), pack-reused 241 (from 1)\u001b[K\n",
            "Receiving objects: 100% (312/312), 34.92 MiB | 24.52 MiB/s, done.\n",
            "Resolving deltas: 100% (224/224), done.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "%cd /content/simple-hierarchical-transformer"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ksTSlONwAIYU",
        "outputId": "a4c52955-339b-4f62-cf97-221e3c6cf4e7"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/simple-hierarchical-transformer\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install -e ."
      ],
      "metadata": {
        "id": "CWp389ps-jIe"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "Yfaoorlt-PHh"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "from simple_hierarchical_transformer import HierarchicalTransformer\n",
        "\n",
        "model = HierarchicalTransformer(\n",
        "    num_tokens = 20000,                # number of tokens\n",
        "    dim = 512,                         # model dimensions\n",
        "    depth = 6,                         # depth\n",
        "    dim_head = 64,                     # dimension per attention head\n",
        "    heads = 8,                         # attention heads\n",
        "    seq_len = 2048,                    # sequence lengths\n",
        "    hierarchies = (1, 2, 8),           # hierarchies - here we have 1x (like in a regular transformer), then 2x and 8x compressed hierarchical tokens that undergo their own transformer blocks. information is pooled into one hierarchy at each layer\n",
        "    window_sizes = (32, 64, None)      # local attention window sizes - the idea is that the higher hierarchies can pass distant information to the local one. None stands for full receptive field. Setting 0 would turn off attention at this hierarchy altogether (while token shift will still be in effect in each layer)\n",
        ")\n",
        "print(model)\n",
        "\n",
        "ids = torch.randint(0, 20000, (1, 2048))\n",
        "print(ids.shpae)\n",
        "loss, _ = model(ids, return_loss = True)\n",
        "print(loss)\n",
        "loss.backward()\n",
        "\n",
        "# after much training\n",
        "\n",
        "logits = model(ids)\n",
        "print(logits)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# non-hierarchical transformer\n",
        "\n",
        "model = HierarchicalTransformer(\n",
        "    num_tokens = 20000,\n",
        "    dim = 512,\n",
        "    depth = 8,\n",
        "    dim_head = 64,\n",
        "    heads = 8,\n",
        "    seq_len = 2048,\n",
        "    hierarchies = 1,        # implied 1 if not set\n",
        "    window_sizes = None     # implied None (full sequence length) if not set\n",
        ")\n",
        "print(model)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7vVLV2qPAXRt",
        "outputId": "e5f8ca85-935c-4daf-88a4-31a06260294d"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "HierarchicalTransformer(\n",
            "  (token_emb): Embedding(20000, 512)\n",
            "  (to_hierarchical_preds): ModuleList(\n",
            "    (0): None\n",
            "  )\n",
            "  (compressors): ModuleList(\n",
            "    (0): Compress(\n",
            "      (compress_fn): Identity()\n",
            "    )\n",
            "  )\n",
            "  (post_token_emb_norms): ModuleList(\n",
            "    (0): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "  )\n",
            "  (layers): ModuleList(\n",
            "    (0-7): 8 x ModuleList(\n",
            "      (0): HierarchicalBlock(\n",
            "        (attn): Attention(\n",
            "          (norm): RMSNorm()\n",
            "          (rotary_emb): RotaryEmbedding()\n",
            "          (attend): Attend()\n",
            "          (to_qkv): Linear(in_features=512, out_features=1536, bias=False)\n",
            "          (to_out): Linear(in_features=512, out_features=512, bias=False)\n",
            "        )\n",
            "        (ff): FeedForward(\n",
            "          (net): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=512, out_features=2048, bias=False)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Linear(in_features=2048, out_features=512, bias=False)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "    )\n",
            "  )\n",
            "  (hierarchical_merges): ModuleList()\n",
            "  (norms): ModuleList(\n",
            "    (0): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "  )\n",
            "  (to_logits): Linear(in_features=512, out_features=20000, bias=False)\n",
            ")\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model = HierarchicalTransformer(\n",
        "    num_tokens = 256,\n",
        "    dim = (128, 256, 512, 1024),\n",
        "    depth = 8,\n",
        "    seq_len = 1024,\n",
        "    use_flash_attn = True,\n",
        "    ff_mult = (2, 2, 4, 4),\n",
        "    dim_head = (16, 32, 64, 64),\n",
        "    heads = (2, 4, 8, 8),\n",
        "    hierarchies = (1, 2, 4, 16),\n",
        "    hierarchical_stride = (1, 1, 1, 8),  # this would determine the stride when compressing, and when concatting the hierarchical tokens to the fine tokens, the past tokens will be repeated this amount of time. causality is not violated as using the trick from hourglass transformers where sequence is shifted by compression factor - 1. recommend sticking with 1 except for highly compressed hierarchies, as it becomes very uncompetitive with baseline and generations look off\n",
        "    window_sizes = (16, 32, 64, None)\n",
        ").cuda()\n",
        "\n",
        "print(model)\n",
        "\n",
        "# hierarchies\n",
        "# 1x - dim 128 - attention (2 heads, 16 dim, receptive field 16)\n",
        "# 2x - dim 256 - attention (4 heads, 32 dim, receptive field 32)\n",
        "# 4x - dim 512 - attention (8 heads, 64 dim, receptive field 64)\n",
        "# 8x - dim 1024 - attention (8 heads, 64 dim, receptive field of all)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EIMQFngnAh1H",
        "outputId": "5884a7e7-1a0d-4cd8-f129-1db64059dd34"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "HierarchicalTransformer(\n",
            "  (token_emb): Embedding(256, 1024)\n",
            "  (to_hierarchical_preds): ModuleList(\n",
            "    (0): None\n",
            "    (1): Linear(in_features=256, out_features=256, bias=True)\n",
            "    (2): Linear(in_features=512, out_features=512, bias=True)\n",
            "    (3): Linear(in_features=1024, out_features=1024, bias=True)\n",
            "  )\n",
            "  (compressors): ModuleList(\n",
            "    (0): Compress(\n",
            "      (compress_fn): Linear(in_features=1024, out_features=128, bias=False)\n",
            "    )\n",
            "    (1): Compress(\n",
            "      (compress_fn): Sequential(\n",
            "        (0): Rearrange('b n d -> b d n')\n",
            "        (1): CausalConv(\n",
            "          (conv): Conv1d(1024, 4096, kernel_size=(2,), stride=(1,))\n",
            "        )\n",
            "        (2): SiLU()\n",
            "        (3): Conv1d(4096, 256, kernel_size=(1,), stride=(1,))\n",
            "        (4): Rearrange('b d n -> b n d')\n",
            "      )\n",
            "      (to_recon): Linear(in_features=256, out_features=512, bias=False)\n",
            "    )\n",
            "    (2): Compress(\n",
            "      (compress_fn): Sequential(\n",
            "        (0): Rearrange('b n d -> b d n')\n",
            "        (1): CausalConv(\n",
            "          (conv): Conv1d(1024, 4096, kernel_size=(4,), stride=(1,))\n",
            "        )\n",
            "        (2): SiLU()\n",
            "        (3): Conv1d(4096, 512, kernel_size=(1,), stride=(1,))\n",
            "        (4): Rearrange('b d n -> b n d')\n",
            "      )\n",
            "      (to_recon): Linear(in_features=512, out_features=1024, bias=False)\n",
            "    )\n",
            "    (3): Compress(\n",
            "      (compress_fn): Sequential(\n",
            "        (0): Rearrange('b n d -> b d n')\n",
            "        (1): CausalConv(\n",
            "          (conv): Conv1d(1024, 4096, kernel_size=(16,), stride=(8,))\n",
            "        )\n",
            "        (2): SiLU()\n",
            "        (3): Conv1d(4096, 1024, kernel_size=(1,), stride=(1,))\n",
            "        (4): Rearrange('b d n -> b n d')\n",
            "      )\n",
            "      (to_recon): Linear(in_features=1024, out_features=4096, bias=False)\n",
            "    )\n",
            "  )\n",
            "  (post_token_emb_norms): ModuleList(\n",
            "    (0): LayerNorm((128,), eps=1e-05, elementwise_affine=True)\n",
            "    (1): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n",
            "    (2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "    (3): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
            "  )\n",
            "  (layers): ModuleList(\n",
            "    (0-7): 8 x ModuleList(\n",
            "      (0): HierarchicalBlock(\n",
            "        (attn): LocalMHA(\n",
            "          (norm): LayerNorm((128,), eps=1e-05, elementwise_affine=True)\n",
            "          (to_qkv): Linear(in_features=128, out_features=96, bias=False)\n",
            "          (attn_fn): LocalAttention(\n",
            "            (dropout): Dropout(p=0.0, inplace=False)\n",
            "            (rel_pos): SinusoidalEmbeddings()\n",
            "          )\n",
            "          (to_out): Linear(in_features=32, out_features=128, bias=False)\n",
            "        )\n",
            "        (ff): FeedForward(\n",
            "          (net): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=128, out_features=256, bias=False)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Linear(in_features=256, out_features=128, bias=False)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "      (1): HierarchicalBlock(\n",
            "        (attn): LocalMHA(\n",
            "          (norm): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n",
            "          (to_qkv): Linear(in_features=256, out_features=384, bias=False)\n",
            "          (attn_fn): LocalAttention(\n",
            "            (dropout): Dropout(p=0.0, inplace=False)\n",
            "            (rel_pos): SinusoidalEmbeddings()\n",
            "          )\n",
            "          (to_out): Linear(in_features=128, out_features=256, bias=False)\n",
            "        )\n",
            "        (ff): FeedForward(\n",
            "          (net): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=256, out_features=512, bias=False)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Linear(in_features=512, out_features=256, bias=False)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "      (2): HierarchicalBlock(\n",
            "        (attn): LocalMHA(\n",
            "          (norm): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "          (to_qkv): Linear(in_features=512, out_features=1536, bias=False)\n",
            "          (attn_fn): LocalAttention(\n",
            "            (dropout): Dropout(p=0.0, inplace=False)\n",
            "            (rel_pos): SinusoidalEmbeddings()\n",
            "          )\n",
            "          (to_out): Linear(in_features=512, out_features=512, bias=False)\n",
            "        )\n",
            "        (ff): FeedForward(\n",
            "          (net): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=512, out_features=2048, bias=False)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Linear(in_features=2048, out_features=512, bias=False)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "      (3): HierarchicalBlock(\n",
            "        (attn): Attention(\n",
            "          (norm): RMSNorm()\n",
            "          (rotary_emb): RotaryEmbedding()\n",
            "          (attend): Attend()\n",
            "          (to_qkv): Linear(in_features=1024, out_features=1536, bias=False)\n",
            "          (to_out): Linear(in_features=512, out_features=1024, bias=False)\n",
            "        )\n",
            "        (ff): FeedForward(\n",
            "          (net): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=1024, out_features=4096, bias=False)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Linear(in_features=4096, out_features=1024, bias=False)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "    )\n",
            "  )\n",
            "  (hierarchical_merges): ModuleList(\n",
            "    (0-7): 8 x HierarchicalMerge(\n",
            "      (net): Sequential(\n",
            "        (0): RMSNorm()\n",
            "        (1): Linear(in_features=1920, out_features=256, bias=True)\n",
            "        (2): SiLU()\n",
            "        (3): Linear(in_features=256, out_features=128, bias=True)\n",
            "      )\n",
            "    )\n",
            "  )\n",
            "  (norms): ModuleList(\n",
            "    (0): LayerNorm((128,), eps=1e-05, elementwise_affine=True)\n",
            "    (1): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n",
            "    (2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "    (3): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
            "  )\n",
            "  (to_logits): Linear(in_features=128, out_features=256, bias=False)\n",
            ")\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!python train.py\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4sT_H1WQBLYW",
        "outputId": "ed37aad2-586e-498c-e4ad-6f3729eed286"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/simple-hierarchical-transformer/simple_hierarchical_transformer/simple_hierarchical_transformer.py:109: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.\n",
            "  @autocast(enabled = False)\n",
            "/content/simple-hierarchical-transformer/simple_hierarchical_transformer/simple_hierarchical_transformer.py:138: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.\n",
            "  @autocast(enabled = False)\n",
            "training:   0% 0/100000 [00:00<?, ?it/s]training loss: 5.7180633544921875\n",
            "validation loss: 4.1848015785217285\n",
            "%s \n",
            "\n",
            " %s ('rthodox iconography in the interpretation of [[Simon Ushakov]] (1658).]]  In addition to the legend that Pilate had made an imag', '****************************************************************************************************')\n",
            "   t  a  e te   t o   o      or    o th  ho   |hM   s|   a  ietit t  [ oPi   [  or th1 e tz e       e      l    h  oá s      t   a  e oo n  ao     [   u X a    h   i  ?s  a  o n    t eon  aæ t    a     o t    l   [il a  ai  | o    r o  a         zura   o a   e zi   n en æ  on          n'       t        thao    a   io    tos   o      on  o  z t r tf    ar o  on  an  á     t  r      h  t o    to h[[  o o    o th      n  tha  ou Äf tit oe    t t     t    a      o n t   n t io t[  o ei  io  o  i  hJ      o h    a   t= r t t   |e   arz  ue     t    a | o            li tor io    n n  or  n  f t r n  lh  e s   e   a[  o     ilo     o   on   a  X  [o  o   f  t t o t  e of    a   a t  e     h oe   e   n  a      e re   a    t r  a  r     r   or     o   n  of ii  n  u  l t t  t e    t     t ti [ a    o Xe  ole     z     o  the    on  oh e    att        he    t  [hh´ t   a   oPo    l f ot th  ar               ha  h u f aæoá á   a  te  u tha        ar    l  u  o ohe  n   a ar  f  hu e o     h   e  o      or   a  ohi&    \n",
            "\n",
            "training:   0% 1/100000 [02:01<3383:32:53, 121.81s/it]training loss: 4.331456184387207\n",
            "training:   0% 2/100000 [02:15<1620:47:33, 58.35s/it] training loss: 3.984083652496338\n",
            "training:   0% 3/100000 [02:29<1057:06:53, 38.06s/it]training loss: 3.7945539951324463\n",
            "training:   0% 4/100000 [02:43<789:57:30, 28.44s/it] training loss: 3.399085521697998\n",
            "training:   0% 5/100000 [02:56<641:16:03, 23.09s/it]training loss: 3.168203353881836\n",
            "training:   0% 6/100000 [03:10<552:00:23, 19.87s/it]training loss: 3.524209976196289\n",
            "training:   0% 7/100000 [03:24<496:22:59, 17.87s/it]training loss: 2.8885233402252197\n",
            "training:   0% 8/100000 [03:38<460:10:15, 16.57s/it]training loss: 2.9572205543518066\n",
            "training:   0% 9/100000 [03:51<435:43:16, 15.69s/it]training loss: 3.3888423442840576\n",
            "training:   0% 10/100000 [04:16<713:34:35, 25.69s/it]\n",
            "Traceback (most recent call last):\n",
            "  File \"/content/simple-hierarchical-transformer/train.py\", line 100, in <module>\n",
            "    accelerator.backward(loss / GRADIENT_ACCUMULATE_EVERY)\n",
            "  File \"/usr/local/lib/python3.11/dist-packages/accelerate/accelerator.py\", line 2248, in backward\n",
            "    loss.backward(**kwargs)\n",
            "  File \"/usr/local/lib/python3.11/dist-packages/torch/_tensor.py\", line 581, in backward\n",
            "    torch.autograd.backward(\n",
            "  File \"/usr/local/lib/python3.11/dist-packages/torch/autograd/__init__.py\", line 347, in backward\n",
            "    _engine_run_backward(\n",
            "  File \"/usr/local/lib/python3.11/dist-packages/torch/autograd/graph.py\", line 825, in _engine_run_backward\n",
            "    return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass\n",
            "           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n",
            "KeyboardInterrupt\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# MEGABYTE-pytorch"
      ],
      "metadata": {
        "id": "OmbhcT3LBZiH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%cd /content"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Rz6VxoieBgh6",
        "outputId": "dbc4dcea-d3de-452b-96ee-ec0d0164f4bb"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!git clone https://github.com/lucidrains/MEGABYTE-pytorch.git"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IWQbXcaGBjxB",
        "outputId": "78bbfc4d-2e30-45e9-c54d-53f7fde55774"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Cloning into 'MEGABYTE-pytorch'...\n",
            "remote: Enumerating objects: 188, done.\u001b[K\n",
            "remote: Counting objects: 100% (126/126), done.\u001b[K\n",
            "remote: Compressing objects: 100% (50/50), done.\u001b[K\n",
            "remote: Total 188 (delta 98), reused 77 (delta 76), pack-reused 62 (from 1)\u001b[K\n",
            "Receiving objects: 100% (188/188), 35.31 MiB | 25.92 MiB/s, done.\n",
            "Resolving deltas: 100% (126/126), done.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "%cd /content/MEGABYTE-pytorch"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jQjv1AR4Bqpe",
        "outputId": "527135df-088c-4124-8587-ffc7e2d2ba94"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/MEGABYTE-pytorch\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install -e ."
      ],
      "metadata": {
        "id": "IAgwmLQJBtKh"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import torch\n",
        "from MEGABYTE_pytorch import MEGABYTE\n",
        "\n",
        "model = MEGABYTE(\n",
        "    num_tokens = 16000,             # number of tokens\n",
        "    dim = (512, 256),               # transformer model dimension (512 for coarsest, 256 for fine in this example)\n",
        "    max_seq_len = (1024, 4),        # sequence length for global and then local. this can be more than 2\n",
        "    depth = (6, 4),                 # number of layers for global and then local. this can be more than 2, but length must match the max_seq_len's\n",
        "    dim_head = 64,                  # dimension per head\n",
        "    heads = 8,                      # number of attention heads\n",
        "    flash_attn = True               # use flash attention\n",
        ")\n",
        "print(model)\n",
        "\n",
        "x = torch.randint(0, 16000, (1, 1024, 4))\n",
        "print(x.shape)\n",
        "\n",
        "loss = model(x, return_loss = True)\n",
        "print(loss)\n",
        "loss.backward()\n",
        "\n",
        "# then after much training\n",
        "\n",
        "logits = model(x)\n",
        "print(logits)\n",
        "\n",
        "# and sample from the logits accordingly\n",
        "# or you can use the generate function\n",
        "\n",
        "sampled = model.generate(temperature = 0.9, filter_thres = 0.9) # (1, 1024, 4)\n",
        "print(sampled)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IFZ6XGOJCAwl",
        "outputId": "3e639c34-9722-4f99-d6c5-6467144084dc"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Non-A100 GPU detected, using math or mem efficient attention if input tensor is on cuda\n",
            "MEGABYTE(\n",
            "  (start_tokens): ParameterList(\n",
            "      (0): Parameter containing: [torch.float32 of size 512]\n",
            "      (1): Parameter containing: [torch.float32 of size 256]\n",
            "  )\n",
            "  (token_embs): ModuleList(\n",
            "    (0): Embedding(16000, 256)\n",
            "    (1): Sequential(\n",
            "      (0): Embedding(16000, 256)\n",
            "      (1): Rearrange('... r d -> ... (r d)')\n",
            "      (2): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)\n",
            "      (3): Linear(in_features=1024, out_features=512, bias=True)\n",
            "      (4): LayerNorm((512,), eps=1e-05, elementwise_affine=True)\n",
            "    )\n",
            "  )\n",
            "  (transformers): ModuleList(\n",
            "    (0): Transformer(\n",
            "      (layers): ModuleList(\n",
            "        (0-5): 6 x ModuleList(\n",
            "          (0): Attention(\n",
            "            (attend): Attend(\n",
            "              (attn_dropout): Dropout(p=0.0, inplace=False)\n",
            "            )\n",
            "            (dropout): Dropout(p=0.0, inplace=False)\n",
            "            (norm): RMSNorm()\n",
            "            (to_q): Linear(in_features=512, out_features=512, bias=False)\n",
            "            (to_kv): Linear(in_features=512, out_features=1024, bias=False)\n",
            "            (to_out): Linear(in_features=512, out_features=512, bias=False)\n",
            "          )\n",
            "          (1): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=512, out_features=2048, bias=True)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Dropout(p=0.0, inplace=False)\n",
            "            (4): Linear(in_features=2048, out_features=512, bias=True)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "      (norm): RMSNorm()\n",
            "    )\n",
            "    (1): Transformer(\n",
            "      (layers): ModuleList(\n",
            "        (0-3): 4 x ModuleList(\n",
            "          (0): Attention(\n",
            "            (attend): Attend(\n",
            "              (attn_dropout): Dropout(p=0.0, inplace=False)\n",
            "            )\n",
            "            (dropout): Dropout(p=0.0, inplace=False)\n",
            "            (norm): RMSNorm()\n",
            "            (to_q): Linear(in_features=256, out_features=512, bias=False)\n",
            "            (to_kv): Linear(in_features=256, out_features=1024, bias=False)\n",
            "            (to_out): Linear(in_features=512, out_features=256, bias=False)\n",
            "          )\n",
            "          (1): Sequential(\n",
            "            (0): RMSNorm()\n",
            "            (1): Linear(in_features=256, out_features=1024, bias=True)\n",
            "            (2): GELU(approximate='none')\n",
            "            (3): Dropout(p=0.0, inplace=False)\n",
            "            (4): Linear(in_features=1024, out_features=256, bias=True)\n",
            "          )\n",
            "        )\n",
            "      )\n",
            "      (norm): RMSNorm()\n",
            "    )\n",
            "  )\n",
            "  (to_next_transformer_projections): ModuleList(\n",
            "    (0): Sequential(\n",
            "      (0): Rearrange('b ... d -> b (...) d')\n",
            "      (1): Linear(in_features=512, out_features=1024, bias=True)\n",
            "      (2): Rearrange('b m (n d) -> (b m) n d', n=4)\n",
            "    )\n",
            "    (1): Identity()\n",
            "  )\n",
            "  (to_logits): Linear(in_features=256, out_features=16000, bias=True)\n",
            ")\n",
            "torch.Size([1, 1024, 4])\n",
            "tensor(9.8497, grad_fn=<NllLoss2DBackward0>)\n",
            "tensor([[[[-0.4943,  0.1143, -0.0231,  ...,  0.1245, -0.5350,  0.5650],\n",
            "          [ 0.1847, -0.3926,  0.7861,  ..., -0.0890,  0.4597,  0.9959],\n",
            "          [ 0.3339,  0.5563, -0.3116,  ..., -0.3866,  0.5440,  0.3514],\n",
            "          [ 0.0211, -1.0455,  0.2691,  ..., -0.1794,  0.6831, -0.3927]],\n",
            "\n",
            "         [[ 0.4604,  0.3415, -0.3087,  ..., -0.2270, -0.6176,  0.4187],\n",
            "          [ 0.0401,  0.1568, -1.1114,  ..., -0.7763,  0.9537,  0.4145],\n",
            "          [ 0.7518, -0.1061, -0.0683,  ...,  0.7875, -0.0641, -0.1145],\n",
            "          [-0.0706,  0.8074, -0.2956,  ..., -1.3693,  0.0241,  0.0143]],\n",
            "\n",
            "         [[ 0.2181, -0.3898, -0.5323,  ..., -0.4099,  0.6210,  0.8403],\n",
            "          [ 1.0235, -0.3286,  0.4173,  ..., -0.3496, -0.0380,  1.2426],\n",
            "          [ 0.3527,  0.2448,  0.1678,  ..., -0.6246,  0.0925,  0.0846],\n",
            "          [ 0.0725,  0.4074, -0.1016,  ..., -1.5008,  0.6853,  0.2040]],\n",
            "\n",
            "         ...,\n",
            "\n",
            "         [[-0.4750,  0.6405,  0.2786,  ..., -0.6838,  0.2464, -1.4629],\n",
            "          [ 0.2489, -0.2560,  0.3751,  ...,  0.1836,  0.5290,  0.2579],\n",
            "          [ 0.4820,  1.2440,  0.0793,  ...,  0.1882, -0.0142, -0.2511],\n",
            "          [-0.1188,  0.6345, -0.2967,  ..., -0.7147,  0.2616,  0.0900]],\n",
            "\n",
            "         [[ 0.0642,  0.0233,  0.2822,  ..., -1.2601, -0.3762, -0.4715],\n",
            "          [-0.1478, -0.9028, -0.5440,  ...,  0.5542, -0.5728,  0.1481],\n",
            "          [ 0.2448,  0.4092,  0.6897,  ...,  0.0592,  0.5244, -0.4434],\n",
            "          [-0.3372, -0.7207, -0.2221,  ...,  0.4248,  0.1397, -0.0669]],\n",
            "\n",
            "         [[-0.0105, -0.8228, -0.2399,  ...,  0.9273,  0.2117,  1.1639],\n",
            "          [-0.7845, -0.6572, -0.1693,  ...,  0.3130, -0.4866, -0.0569],\n",
            "          [-0.6811,  0.4095,  0.1487,  ..., -0.1394, -0.6068, -0.2261],\n",
            "          [ 0.8618,  0.0835,  0.9183,  ..., -1.1080,  0.0137, -0.0765]]]],\n",
            "       grad_fn=<SliceBackward0>)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            " 61%|██████▏   | 2514/4096 [33:14<1:08:40,  2.60s/it]"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!python train.py"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "44bCxc60B5_o",
        "outputId": "a76c0607-9650-4394-e595-81e6d1438452"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Non-A100 GPU detected, using math or mem efficient attention if input tensor is on cuda\n",
            "training:   0% 0/100000 [00:00<?, ?it/s]training loss: 5.705471515655518\n",
            "validation loss: 5.229538440704346\n",
            "training loss: 5.157638072967529\n",
            "training loss: 4.804471969604492\n",
            "training:   0% 3/100000 [00:10<100:40:18,  3.62s/it]training loss: 4.644378185272217\n",
            "training loss: 4.377236366271973\n",
            "training loss: 4.3670196533203125\n",
            "training loss: 4.2122907638549805\n",
            "training:   0% 7/100000 [00:23<90:50:02,  3.27s/it] training loss: 4.318767547607422\n",
            "training loss: 4.108756065368652\n",
            "training loss: 4.0011067390441895\n",
            "training loss: 3.890439748764038\n",
            "training:   0% 11/100000 [00:35<88:44:14,  3.19s/it]training loss: 3.8987667560577393\n",
            "training loss: 3.9163074493408203\n",
            "training loss: 3.841243267059326\n",
            "training loss: 3.8642024993896484\n",
            "training:   0% 15/100000 [00:48<88:26:14,  3.18s/it]training loss: 3.62235164642334\n",
            "training loss: 3.9597721099853516\n",
            "training loss: 3.819218158721924\n",
            "training:   0% 15/100000 [01:00<88:26:14,  3.18s/it]training loss: 3.6493301391601562\n",
            "training:   0% 19/100000 [01:01<88:45:13,  3.20s/it]training loss: 3.537214756011963\n",
            "training loss: 3.650087833404541\n",
            "training loss: 3.5029683113098145\n",
            "training loss: 3.5410315990448\n",
            "training:   0% 23/100000 [01:14<89:32:06,  3.22s/it]training loss: 3.740438461303711\n",
            "training loss: 3.4931747913360596\n",
            "training loss: 3.332310676574707\n",
            "training loss: 3.3656697273254395\n",
            "training:   0% 27/100000 [01:27<90:25:32,  3.26s/it]training loss: 3.432007312774658\n",
            "training loss: 3.4658846855163574\n",
            "training loss: 3.227494716644287\n",
            "training:   0% 27/100000 [01:40<90:25:32,  3.26s/it]training loss: 3.415137767791748\n",
            "training:   0% 31/100000 [01:40<90:43:46,  3.27s/it]training loss: 3.3413329124450684\n",
            "training loss: 3.18973445892334\n",
            "training loss: 3.4701178073883057\n",
            "training loss: 3.21579647064209\n",
            "training:   0% 35/100000 [01:53<90:34:43,  3.26s/it]training loss: 3.1391103267669678\n",
            "training loss: 3.051894426345825\n",
            "training loss: 3.1173243522644043\n",
            "training loss: 3.1533401012420654\n",
            "training:   0% 39/100000 [02:06<90:19:05,  3.25s/it]training loss: 2.870779514312744\n",
            "training loss: 2.9276950359344482\n",
            "training loss: 3.1499807834625244\n",
            "training loss: 2.97934627532959\n",
            "training:   0% 43/100000 [02:19<90:13:03,  3.25s/it]training loss: 2.967946767807007\n",
            "training loss: 2.925365924835205\n",
            "training loss: 2.906187057495117\n",
            "training:   0% 43/100000 [02:30<90:13:03,  3.25s/it]training loss: 2.9607632160186768\n",
            "training:   0% 47/100000 [02:32<90:16:30,  3.25s/it]training loss: 2.921325922012329\n",
            "training loss: 2.9350528717041016\n",
            "training loss: 2.8141555786132812\n",
            "training loss: 2.8271822929382324\n",
            "training:   0% 51/100000 [02:45<90:24:52,  3.26s/it]training loss: 3.056817054748535\n",
            "training loss: 3.1984424591064453\n",
            "training loss: 2.78092622756958\n",
            "training loss: 2.730290412902832\n",
            "training:   0% 55/100000 [02:58<90:31:54,  3.26s/it]training loss: 3.029428243637085\n",
            "training loss: 2.896733283996582\n",
            "training loss: 2.701331615447998\n",
            "training:   0% 55/100000 [03:10<90:31:54,  3.26s/it]training loss: 2.7915000915527344\n",
            "training:   0% 59/100000 [03:11<90:34:19,  3.26s/it]training loss: 2.8217482566833496\n",
            "training loss: 2.5956363677978516\n",
            "training loss: 2.6837480068206787\n",
            "training loss: 3.1570229530334473\n",
            "training:   0% 63/100000 [03:24<90:29:34,  3.26s/it]training loss: 2.927358388900757\n",
            "training loss: 2.6423802375793457\n",
            "training loss: 2.7779932022094727\n",
            "training loss: 2.598036050796509\n",
            "training:   0% 67/100000 [03:37<90:25:04,  3.26s/it]training loss: 2.797236919403076\n",
            "training loss: 2.6918742656707764\n",
            "training loss: 2.708031415939331\n",
            "training:   0% 67/100000 [03:50<90:25:04,  3.26s/it]training loss: 2.4911136627197266\n",
            "training:   0% 71/100000 [03:50<90:22:45,  3.26s/it]training loss: 2.635128974914551\n",
            "training loss: 2.6644327640533447\n",
            "training loss: 2.74798583984375\n",
            "training loss: 2.4871222972869873\n",
            "training:   0% 75/100000 [04:03<90:26:25,  3.26s/it]training loss: 2.8140974044799805\n",
            "training loss: 2.5521984100341797\n",
            "training loss: 2.5661256313323975\n",
            "training loss: 2.4600393772125244\n",
            "training:   0% 79/100000 [04:17<90:26:54,  3.26s/it]training loss: 2.7193942070007324\n",
            "training loss: 2.6546356678009033\n",
            "training loss: 2.8046793937683105\n",
            "training:   0% 79/100000 [04:30<90:26:54,  3.26s/it]training loss: 2.4827957153320312\n",
            "training:   0% 83/100000 [04:30<90:28:21,  3.26s/it]training loss: 2.562002658843994\n",
            "training loss: 2.4553208351135254\n",
            "training loss: 2.5133838653564453\n",
            "training loss: 2.451014757156372\n",
            "training:   0% 87/100000 [04:43<90:29:53,  3.26s/it]training loss: 2.468503475189209\n",
            "training loss: 2.500386953353882\n",
            "training loss: 2.5075581073760986\n",
            "training loss: 2.533438205718994\n",
            "training:   0% 91/100000 [04:56<90:31:07,  3.26s/it]training loss: 2.5118744373321533\n",
            "training loss: 2.4699087142944336\n",
            "training loss: 2.4673266410827637\n",
            "training loss: 2.5327281951904297\n",
            "training:   0% 95/100000 [05:09<90:29:02,  3.26s/it]training loss: 2.360416889190674\n",
            "training loss: 2.5084779262542725\n",
            "training loss: 2.5309033393859863\n",
            "training:   0% 95/100000 [05:20<90:29:02,  3.26s/it]training loss: 2.447052001953125\n",
            "training:   0% 99/100000 [05:22<90:28:33,  3.26s/it]training loss: 2.4389681816101074\n",
            "training loss: 2.4608163833618164\n",
            "validation loss: 2.3397130966186523\n",
            "training loss: 2.3993678092956543\n",
            "training loss: 2.5534896850585938\n",
            "training:   0% 103/100000 [05:35<90:58:37,  3.28s/it]training loss: 2.5267817974090576\n",
            "training loss: 2.48968505859375\n",
            "training loss: 2.3763320446014404\n",
            "training loss: 2.340885639190674\n",
            "training:   0% 107/100000 [05:48<90:49:30,  3.27s/it]training loss: 2.3894805908203125\n",
            "training loss: 2.4589216709136963\n",
            "training loss: 2.4752840995788574\n",
            "training:   0% 107/100000 [06:00<90:49:30,  3.27s/it]training loss: 2.3724403381347656\n",
            "training:   0% 111/100000 [06:01<90:45:40,  3.27s/it]training loss: 2.421093463897705\n",
            "training loss: 2.3705921173095703\n",
            "training loss: 2.2842066287994385\n",
            "training loss: 2.3808085918426514\n",
            "training:   0% 115/100000 [06:14<90:41:46,  3.27s/it]training loss: 2.356661319732666\n",
            "training loss: 2.3537678718566895\n",
            "training loss: 2.372596263885498\n",
            "training loss: 2.3147339820861816\n",
            "training:   0% 119/100000 [06:27<90:37:32,  3.27s/it]training loss: 2.42429256439209\n",
            "training loss: 2.329951286315918\n",
            "training loss: 2.577165126800537\n",
            "training:   0% 119/100000 [06:40<90:37:32,  3.27s/it]training loss: 2.363126754760742\n",
            "training:   0% 123/100000 [06:40<90:36:35,  3.27s/it]training loss: 2.316194534301758\n",
            "training loss: 2.2781968116760254\n",
            "training loss: 2.3497562408447266\n",
            "training loss: 2.287128448486328\n",
            "training:   0% 127/100000 [06:53<90:34:15,  3.26s/it]training loss: 2.400522232055664\n",
            "training loss: 2.279531478881836\n",
            "training loss: 2.311062812805176\n",
            "training loss: 2.3214032649993896\n",
            "training:   0% 131/100000 [07:06<90:34:21,  3.26s/it]training loss: 2.3030028343200684\n",
            "training loss: 2.3242642879486084\n",
            "training loss: 2.4131879806518555\n",
            "training loss: 2.3900609016418457\n",
            "training:   0% 135/100000 [07:19<90:33:18,  3.26s/it]training loss: 2.3506698608398438\n",
            "training loss: 2.2969326972961426\n",
            "training loss: 2.241105079650879\n",
            "training:   0% 135/100000 [07:30<90:33:18,  3.26s/it]training loss: 2.3124208450317383\n",
            "training:   0% 139/100000 [07:33<90:31:05,  3.26s/it]training loss: 2.2921154499053955\n",
            "training loss: 2.2749862670898438\n",
            "training loss: 2.319890022277832\n",
            "training loss: 2.4289743900299072\n",
            "training:   0% 143/100000 [07:46<90:27:33,  3.26s/it]training loss: 2.225823402404785\n",
            "training loss: 2.3324666023254395\n",
            "training loss: 2.32633638381958\n",
            "training loss: 2.321639060974121\n",
            "training:   0% 147/100000 [07:59<90:24:27,  3.26s/it]training loss: 2.3571512699127197\n",
            "training loss: 2.26077938079834\n",
            "training loss: 2.454716205596924\n",
            "training:   0% 147/100000 [08:10<90:24:27,  3.26s/it]training loss: 2.416062831878662\n",
            "training:   0% 151/100000 [08:12<90:26:03,  3.26s/it]training loss: 2.2237913608551025\n",
            "training loss: 2.2676079273223877\n",
            "training loss: 2.262273073196411\n",
            "training loss: 2.3983020782470703\n",
            "training:   0% 155/100000 [08:25<90:27:29,  3.26s/it]training loss: 2.295137882232666\n",
            "training loss: 2.2479467391967773\n",
            "training loss: 2.1365013122558594\n",
            "training loss: 2.285226821899414\n",
            "training:   0% 159/100000 [08:38<90:24:56,  3.26s/it]training loss: 2.211986541748047\n",
            "training loss: 2.235379219055176\n",
            "training loss: 2.3624191284179688\n",
            "training:   0% 159/100000 [08:50<90:24:56,  3.26s/it]training loss: 2.210317373275757\n",
            "training:   0% 163/100000 [08:51<90:25:29,  3.26s/it]training loss: 2.3048181533813477\n",
            "training loss: 2.1654820442199707\n",
            "training loss: 2.359989643096924\n",
            "training loss: 2.307468891143799\n",
            "training:   0% 167/100000 [09:04<90:28:04,  3.26s/it]training loss: 2.2407116889953613\n",
            "training loss: 2.1870126724243164\n",
            "training loss: 2.2837634086608887\n",
            "training loss: 2.220604181289673\n",
            "training:   0% 171/100000 [09:17<90:28:44,  3.26s/it]training loss: 2.3143439292907715\n",
            "training loss: 2.2009143829345703\n",
            "training loss: 2.1612470149993896\n",
            "training:   0% 171/100000 [09:30<90:28:44,  3.26s/it]training loss: 2.194310188293457\n",
            "training:   0% 175/100000 [09:30<90:27:11,  3.26s/it]training loss: 2.1255109310150146\n",
            "training loss: 2.2572836875915527\n",
            "training loss: 2.1856489181518555\n",
            "training loss: 2.1893138885498047\n",
            "training:   0% 179/100000 [09:43<90:25:45,  3.26s/it]training loss: 2.1897940635681152\n",
            "training loss: 2.174685001373291\n",
            "training loss: 2.2123475074768066\n",
            "training loss: 2.3409547805786133\n",
            "training:   0% 183/100000 [09:56<90:24:55,  3.26s/it]training loss: 2.1104607582092285\n",
            "training loss: 2.157086133956909\n",
            "training loss: 2.1234428882598877\n",
            "training loss: 2.154426097869873\n",
            "training:   0% 187/100000 [10:09<90:27:24,  3.26s/it]training loss: 2.190891981124878\n",
            "training loss: 2.311950445175171\n",
            "training loss: 2.226283550262451\n",
            "training:   0% 187/100000 [10:20<90:27:24,  3.26s/it]training loss: 2.1747183799743652\n",
            "training:   0% 191/100000 [10:22<90:26:32,  3.26s/it]training loss: 2.1750950813293457\n",
            "training loss: 2.1138124465942383\n",
            "training loss: 2.157479763031006\n",
            "training loss: 2.2071590423583984\n",
            "training:   0% 195/100000 [10:35<90:30:14,  3.26s/it]training loss: 2.1159634590148926\n",
            "training loss: 2.18137264251709\n",
            "training loss: 2.260080337524414\n",
            "training loss: 2.2740371227264404\n",
            "training:   0% 199/100000 [10:48<90:29:50,  3.26s/it]training loss: 2.162501811981201\n",
            "training loss: 2.1222503185272217\n",
            "validation loss: 2.174881935119629\n",
            "training loss: 2.3583996295928955\n",
            "training:   0% 199/100000 [11:00<90:29:50,  3.26s/it]training loss: 2.1368801593780518\n",
            "training:   0% 203/100000 [11:02<91:00:33,  3.28s/it]training loss: 2.2296950817108154\n",
            "training loss: 2.2642550468444824\n",
            "training loss: 2.3023452758789062\n",
            "training loss: 2.304734945297241\n",
            "training:   0% 207/100000 [11:15<90:54:58,  3.28s/it]training loss: 2.197486400604248\n",
            "training loss: 2.1551175117492676\n",
            "training loss: 2.284496784210205\n",
            "training loss: 2.1277904510498047\n",
            "training:   0% 211/100000 [11:28<90:50:43,  3.28s/it]training loss: 2.175051689147949\n",
            "training loss: 2.136810302734375\n",
            "training loss: 2.0980706214904785\n",
            "training:   0% 211/100000 [11:40<90:50:43,  3.28s/it]training loss: 2.093656063079834\n",
            "training:   0% 215/100000 [11:41<90:49:09,  3.28s/it]training loss: 2.1068978309631348\n",
            "training loss: 2.1691408157348633\n",
            "training loss: 2.242001533508301\n",
            "training loss: 2.159320831298828\n",
            "training:   0% 219/100000 [11:54<90:44:51,  3.27s/it]training loss: 2.118852138519287\n",
            "training loss: 2.0936059951782227\n",
            "training loss: 2.1387815475463867\n",
            "training loss: 2.1377429962158203\n",
            "training:   0% 223/100000 [12:07<90:40:08,  3.27s/it]training loss: 2.1912214756011963\n",
            "training loss: 2.117096424102783\n",
            "training loss: 2.034675121307373\n",
            "training:   0% 223/100000 [12:20<90:40:08,  3.27s/it]training loss: 2.081725597381592\n",
            "training:   0% 227/100000 [12:20<90:34:02,  3.27s/it]training loss: 2.2155983448028564\n",
            "training loss: 2.099849224090576\n",
            "training loss: 2.108546495437622\n",
            "training loss: 2.149226665496826\n",
            "training:   0% 231/100000 [12:33<90:29:27,  3.27s/it]training loss: 2.129607915878296\n",
            "training loss: 2.1898746490478516\n",
            "training loss: 2.084137439727783\n",
            "training loss: 2.1094279289245605\n",
            "training:   0% 235/100000 [12:46<90:27:20,  3.26s/it]training loss: 2.1444849967956543\n",
            "training loss: 2.135111093521118\n",
            "training loss: 2.050504207611084\n",
            "training loss: 2.0195422172546387\n",
            "training:   0% 239/100000 [12:59<90:27:15,  3.26s/it]training loss: 2.407824993133545\n",
            "training loss: 2.1303272247314453\n",
            "training loss: 2.1106464862823486\n",
            "training:   0% 239/100000 [13:10<90:27:15,  3.26s/it]training loss: 2.0870885848999023\n",
            "training:   0% 243/100000 [13:12<90:26:18,  3.26s/it]training loss: 2.0278546810150146\n",
            "training loss: 2.149264335632324\n",
            "training loss: 2.265665054321289\n",
            "training loss: 2.132596969604492\n",
            "training:   0% 247/100000 [13:25<90:25:16,  3.26s/it]training loss: 2.093745708465576\n",
            "training loss: 2.0907363891601562\n",
            "training loss: 2.0685107707977295\n",
            "training loss: 2.13996958732605\n",
            "training:   0% 251/100000 [13:38<90:26:28,  3.26s/it]training loss: 2.1783676147460938\n",
            "training loss: 2.1466684341430664\n",
            "training loss: 2.284046173095703\n",
            "training:   0% 251/100000 [13:50<90:26:28,  3.26s/it]training loss: 2.150052070617676\n",
            "training:   0% 255/100000 [13:51<90:26:12,  3.26s/it]training loss: 2.1278111934661865\n",
            "training loss: 2.1288115978240967\n",
            "training loss: 2.160231113433838\n",
            "training loss: 1.995205044746399\n",
            "training:   0% 259/100000 [14:04<90:28:10,  3.27s/it]training loss: 2.0615243911743164\n",
            "training loss: 2.2053427696228027\n",
            "training loss: 2.1133110523223877\n",
            "training loss: 2.1095662117004395\n",
            "training:   0% 263/100000 [14:17<90:27:26,  3.27s/it]training loss: 2.091498374938965\n",
            "training loss: 2.0736961364746094\n",
            "training loss: 2.2134785652160645\n",
            "training:   0% 263/100000 [14:30<90:27:26,  3.27s/it]training loss: 2.1439366340637207\n",
            "training:   0% 267/100000 [14:31<90:27:19,  3.27s/it]training loss: 2.330440044403076\n",
            "training loss: 2.186475992202759\n",
            "training loss: 2.1551513671875\n",
            "training loss: 2.0099518299102783\n",
            "training:   0% 271/100000 [14:44<90:25:55,  3.26s/it]training loss: 2.1284337043762207\n",
            "training loss: 1.9726736545562744\n",
            "training loss: 2.152956247329712\n",
            "training loss: 2.054873466491699\n",
            "training:   0% 275/100000 [14:57<90:24:48,  3.26s/it]training loss: 2.099933624267578\n",
            "training loss: 1.91866934299469\n",
            "training loss: 2.0507242679595947\n",
            "training:   0% 275/100000 [15:10<90:24:48,  3.26s/it]training loss: 2.0557422637939453\n",
            "training:   0% 279/100000 [15:10<90:25:02,  3.26s/it]training loss: 2.0475544929504395\n",
            "training loss: 2.022981882095337\n",
            "training loss: 2.047163963317871\n",
            "training loss: 2.0633158683776855\n",
            "training:   0% 283/100000 [15:23<90:26:19,  3.27s/it]training loss: 2.1602213382720947\n",
            "training loss: 2.1383185386657715\n",
            "training loss: 2.0212559700012207\n",
            "training loss: 1.9951457977294922\n",
            "training:   0% 287/100000 [15:36<90:27:13,  3.27s/it]training loss: 2.175100564956665\n",
            "training loss: 2.05594539642334\n",
            "training loss: 1.909066915512085\n",
            "training loss: 2.1037464141845703\n",
            "training:   0% 291/100000 [15:49<90:27:23,  3.27s/it]training loss: 2.0656704902648926\n",
            "training loss: 2.1501548290252686\n",
            "training loss: 2.0980184078216553\n",
            "training:   0% 291/100000 [16:00<90:27:23,  3.27s/it]training loss: 2.150491714477539\n",
            "training:   0% 295/100000 [16:02<90:28:57,  3.27s/it]training loss: 2.0918354988098145\n",
            "training loss: 2.093489646911621\n",
            "training loss: 2.1820852756500244\n",
            "training loss: 2.0087695121765137\n",
            "training:   0% 299/100000 [16:15<90:28:51,  3.27s/it]training loss: 2.1245627403259277\n",
            "training loss: 2.0648131370544434\n",
            "validation loss: 2.0954160690307617\n",
            "training loss: 2.121487617492676\n",
            "training loss: 2.021251678466797\n",
            "training:   0% 303/100000 [16:28<91:00:50,  3.29s/it]training loss: 2.0430397987365723\n",
            "training loss: 2.0458688735961914\n",
            "training loss: 1.9582939147949219\n",
            "training:   0% 303/100000 [16:40<91:00:50,  3.29s/it]training loss: 2.1616621017456055\n",
            "training:   0% 307/100000 [16:41<90:49:47,  3.28s/it]training loss: 2.1285367012023926\n",
            "training loss: 2.1785078048706055\n",
            "training loss: 1.9937238693237305\n",
            "training loss: 1.9473259449005127\n",
            "training:   0% 311/100000 [16:54<90:42:16,  3.28s/it]training loss: 2.038374662399292\n",
            "training loss: 2.044426202774048\n",
            "training loss: 2.0588722229003906\n",
            "training loss: 2.027459144592285\n",
            "training:   0% 315/100000 [17:08<90:36:16,  3.27s/it]training loss: 2.1555709838867188\n",
            "training loss: 2.1884238719940186\n",
            "training loss: 2.2065954208374023\n",
            "training:   0% 315/100000 [17:20<90:36:16,  3.27s/it]training loss: 2.1008384227752686\n",
            "training:   0% 319/100000 [17:21<90:32:30,  3.27s/it]training loss: 1.9609731435775757\n",
            "training loss: 1.9312593936920166\n",
            "training loss: 2.041640043258667\n",
            "training loss: 2.1008613109588623\n",
            "training:   0% 323/100000 [17:34<90:29:06,  3.27s/it]training loss: 2.1529688835144043\n",
            "training loss: 1.98182213306427\n",
            "training loss: 2.0465927124023438\n",
            "training loss: 1.9170788526535034\n",
            "training:   0% 327/100000 [17:47<90:26:53,  3.27s/it]training loss: 2.022467613220215\n",
            "training loss: 2.0815937519073486\n",
            "training loss: 2.1237080097198486\n",
            "training:   0% 327/100000 [18:00<90:26:53,  3.27s/it]training loss: 1.9835810661315918\n",
            "training:   0% 331/100000 [18:00<90:26:15,  3.27s/it]training loss: 1.9679582118988037\n",
            "training loss: 2.1929965019226074\n",
            "training loss: 2.044863700866699\n",
            "training loss: 1.9641304016113281\n",
            "training:   0% 335/100000 [18:13<90:23:00,  3.26s/it]training loss: 1.9696840047836304\n",
            "training loss: 1.910114049911499\n",
            "training loss: 2.0925612449645996\n",
            "training loss: 2.061105251312256\n",
            "training:   0% 339/100000 [18:26<90:22:25,  3.26s/it]training loss: 2.0627670288085938\n",
            "training loss: 2.2113020420074463\n",
            "training loss: 2.0851824283599854\n",
            "training loss: 1.9918171167373657\n",
            "training:   0% 343/100000 [18:39<90:24:18,  3.27s/it]training loss: 1.9353467226028442\n",
            "training loss: 1.9991958141326904\n",
            "training loss: 2.107084274291992\n",
            "training:   0% 343/100000 [18:50<90:24:18,  3.27s/it]training loss: 2.099862813949585\n",
            "training:   0% 347/100000 [18:52<90:21:45,  3.26s/it]training loss: 1.9402556419372559\n",
            "training loss: 1.9988130331039429\n",
            "training loss: 2.0257632732391357\n",
            "training loss: 1.9908888339996338\n",
            "training:   0% 351/100000 [19:05<90:20:57,  3.26s/it]training loss: 1.9468879699707031\n",
            "training loss: 1.9925026893615723\n",
            "training loss: 1.8981351852416992\n",
            "training loss: 2.2019236087799072\n",
            "training:   0% 355/100000 [19:18<90:21:20,  3.26s/it]training loss: 1.983171820640564\n",
            "training loss: 2.037393093109131\n",
            "training loss: 2.0219998359680176\n",
            "training:   0% 355/100000 [19:30<90:21:20,  3.26s/it]training loss: 1.9286259412765503\n",
            "training:   0% 359/100000 [19:31<90:22:09,  3.27s/it]training loss: 2.0827741622924805\n",
            "training loss: 1.9542827606201172\n",
            "training loss: 1.8422770500183105\n",
            "training loss: 1.8993593454360962\n",
            "training:   0% 363/100000 [19:44<90:22:26,  3.27s/it]training loss: 1.886815071105957\n",
            "training loss: 1.9122939109802246\n",
            "training loss: 1.9785621166229248\n",
            "training loss: 1.9820717573165894\n",
            "training:   0% 367/100000 [19:57<90:21:01,  3.26s/it]training loss: 2.0802576541900635\n",
            "training loss: 1.8787074089050293\n",
            "training loss: 1.9319519996643066\n",
            "training:   0% 367/100000 [20:10<90:21:01,  3.26s/it]training loss: 2.0875258445739746\n",
            "training:   0% 371/100000 [20:10<90:19:57,  3.26s/it]training loss: 1.898415446281433\n",
            "training loss: 1.9300810098648071\n",
            "training loss: 1.9618260860443115\n",
            "training loss: 1.9995489120483398\n",
            "training:   0% 375/100000 [20:23<90:16:34,  3.26s/it]training loss: 2.007373332977295\n",
            "training loss: 1.9535284042358398\n",
            "training loss: 2.001633644104004\n",
            "training loss: 1.952019214630127\n",
            "training:   0% 379/100000 [20:36<90:15:33,  3.26s/it]training loss: 2.0254087448120117\n",
            "training loss: 1.988936185836792\n",
            "training loss: 2.0303516387939453\n",
            "training loss: 1.8551238775253296\n",
            "training:   0% 383/100000 [20:49<90:12:58,  3.26s/it]training loss: 2.017484664916992\n",
            "training loss: 1.9214575290679932\n",
            "training loss: 2.0142300128936768\n",
            "training:   0% 383/100000 [21:00<90:12:58,  3.26s/it]training loss: 1.9888676404953003\n",
            "training:   0% 387/100000 [21:02<90:12:59,  3.26s/it]training loss: 1.7515462636947632\n",
            "training loss: 1.9027029275894165\n",
            "training loss: 1.9608137607574463\n",
            "training loss: 1.9937357902526855\n",
            "training:   0% 391/100000 [21:16<90:13:15,  3.26s/it]training loss: 2.0419135093688965\n",
            "training loss: 1.8979442119598389\n",
            "training loss: 1.9040958881378174\n",
            "training loss: 1.837041974067688\n",
            "training:   0% 395/100000 [21:29<90:13:37,  3.26s/it]training loss: 1.9498956203460693\n",
            "training loss: 1.9745869636535645\n",
            "training loss: 1.92098867893219\n",
            "training:   0% 395/100000 [21:40<90:13:37,  3.26s/it]training loss: 2.0206847190856934\n",
            "training:   0% 399/100000 [21:42<90:10:38,  3.26s/it]training loss: 2.066779136657715\n",
            "training loss: 2.004873752593994\n",
            "validation loss: 1.8804161548614502\n",
            "training loss: 2.1251370906829834\n",
            "training loss: 1.9951331615447998\n",
            "training:   0% 403/100000 [21:55<90:41:19,  3.28s/it]training loss: 2.126898765563965\n",
            "training loss: 1.9925459623336792\n",
            "training loss: 2.0588836669921875\n",
            "training loss: 1.877345085144043\n",
            "training:   0% 407/100000 [22:08<90:34:45,  3.27s/it]training loss: 1.8718172311782837\n",
            "training loss: 1.9102493524551392\n",
            "training loss: 1.9163768291473389\n",
            "training:   0% 407/100000 [22:20<90:34:45,  3.27s/it]training loss: 1.9903910160064697\n",
            "training:   0% 411/100000 [22:21<90:29:26,  3.27s/it]training loss: 1.9313606023788452\n",
            "training loss: 1.9830200672149658\n",
            "training loss: 2.0513510704040527\n",
            "training loss: 1.957526445388794\n",
            "training:   0% 415/100000 [22:34<90:26:49,  3.27s/it]training loss: 2.060307025909424\n",
            "training loss: 1.9266624450683594\n",
            "training loss: 1.9878677129745483\n",
            "training loss: 1.9488790035247803\n",
            "training:   0% 419/100000 [22:47<90:21:31,  3.27s/it]training loss: 1.8065820932388306\n",
            "training loss: 1.8854162693023682\n",
            "training loss: 1.907813549041748\n",
            "training:   0% 419/100000 [23:00<90:21:31,  3.27s/it]training loss: 1.941481113433838\n",
            "training:   0% 423/100000 [23:00<90:22:07,  3.27s/it]training loss: 1.926876187324524\n",
            "training loss: 1.9573460817337036\n",
            "training loss: 2.028815269470215\n",
            "training loss: 1.9991081953048706\n",
            "training:   0% 427/100000 [23:13<90:21:36,  3.27s/it]training loss: 1.9503612518310547\n",
            "training loss: 1.9360003471374512\n",
            "training loss: 2.005091667175293\n",
            "training loss: 1.8655188083648682\n",
            "training:   0% 431/100000 [23:26<90:20:08,  3.27s/it]training loss: 1.8638819456100464\n",
            "training loss: 1.8813512325286865\n",
            "training loss: 1.8957138061523438\n",
            "training loss: 1.9689868688583374\n",
            "training:   0% 435/100000 [23:39<90:18:04,  3.27s/it]training loss: 1.8829187154769897\n",
            "training loss: 1.9878008365631104\n",
            "training loss: 2.0112998485565186\n",
            "training:   0% 435/100000 [23:50<90:18:04,  3.27s/it]training loss: 1.9897825717926025\n",
            "training:   0% 439/100000 [23:52<90:18:10,  3.27s/it]training loss: 1.9502654075622559\n",
            "training loss: 1.8983027935028076\n",
            "training loss: 1.8753273487091064\n",
            "training loss: 2.0354771614074707\n",
            "training:   0% 443/100000 [24:05<90:15:41,  3.26s/it]training loss: 1.9379839897155762\n",
            "training loss: 1.898531436920166\n",
            "training loss: 1.8974491357803345\n",
            "training loss: 1.9581464529037476\n",
            "training:   0% 447/100000 [24:18<90:12:52,  3.26s/it]training loss: 1.9321346282958984\n",
            "training loss: 1.8316359519958496\n",
            "training loss: 1.966077208518982\n",
            "training:   0% 447/100000 [24:30<90:12:52,  3.26s/it]training loss: 1.9429103136062622\n",
            "training:   0% 451/100000 [24:32<90:13:18,  3.26s/it]training loss: 1.8665530681610107\n",
            "training loss: 1.8845958709716797\n",
            "training loss: 1.8608287572860718\n",
            "training loss: 1.8772438764572144\n",
            "training:   0% 455/100000 [24:45<90:12:47,  3.26s/it]training loss: 1.8439702987670898\n",
            "training loss: 1.970598578453064\n",
            "training loss: 1.8644137382507324\n",
            "training loss: 1.9033112525939941\n",
            "training:   0% 459/100000 [24:58<90:13:15,  3.26s/it]training loss: 1.8539358377456665\n",
            "training loss: 1.9013819694519043\n",
            "training loss: 1.8811595439910889\n",
            "training:   0% 459/100000 [25:10<90:13:15,  3.26s/it]training loss: 1.9216077327728271\n",
            "training:   0% 463/100000 [25:11<90:14:17,  3.26s/it]training loss: 1.832273006439209\n",
            "training loss: 1.9081034660339355\n",
            "training loss: 1.8981564044952393\n",
            "training loss: 1.929281234741211\n",
            "training:   0% 467/100000 [25:24<90:13:14,  3.26s/it]training loss: 2.0039732456207275\n",
            "training loss: 1.9351766109466553\n",
            "training loss: 1.9538894891738892\n",
            "training loss: 1.9065353870391846\n",
            "training:   0% 471/100000 [25:37<90:12:50,  3.26s/it]training loss: 1.964172124862671\n",
            "training loss: 2.005817174911499\n",
            "training loss: 1.9383066892623901\n",
            "training:   0% 471/100000 [25:50<90:12:50,  3.26s/it]training loss: 1.8145049810409546\n",
            "training:   0% 475/100000 [25:50<90:12:06,  3.26s/it]training loss: 1.905462384223938\n",
            "training loss: 1.8075761795043945\n",
            "training loss: 1.9070364236831665\n",
            "training loss: 1.9345262050628662\n",
            "training:   0% 479/100000 [26:03<90:11:39,  3.26s/it]training loss: 1.8720985651016235\n",
            "training loss: 1.8440041542053223\n",
            "training loss: 1.9859025478363037\n",
            "training loss: 1.9042003154754639\n",
            "training:   0% 483/100000 [26:16<90:10:01,  3.26s/it]training loss: 1.9496756792068481\n",
            "training loss: 1.8951191902160645\n",
            "training loss: 1.8686327934265137\n",
            "training loss: 1.9480161666870117\n",
            "training:   0% 487/100000 [26:29<90:08:27,  3.26s/it]training loss: 1.9670333862304688\n",
            "training loss: 2.056797742843628\n",
            "training loss: 1.8825016021728516\n",
            "training:   0% 487/100000 [26:40<90:08:27,  3.26s/it]training loss: 1.849893569946289\n",
            "training:   0% 491/100000 [26:42<90:07:44,  3.26s/it]training loss: 1.9998801946640015\n",
            "training loss: 1.8657808303833008\n",
            "training loss: 1.9658013582229614\n",
            "training loss: 1.8547223806381226\n",
            "training:   0% 495/100000 [26:55<90:11:13,  3.26s/it]training loss: 1.8758121728897095\n",
            "training loss: 1.997986078262329\n",
            "training loss: 1.966107964515686\n",
            "training loss: 1.877305030822754\n",
            "training:   0% 499/100000 [27:08<90:09:34,  3.26s/it]training loss: 1.8662922382354736\n",
            "training loss: 1.8684093952178955\n",
            "validation loss: 2.016204357147217\n",
            "%s \n",
            "\n",
            " %s ('the Roman God [[Mercury (mythology)|Mercury]], who had many similar characteristics, such as both be', '****************************************************************************************************')\n",
            "\n",
            "  0% 0/8092 [00:00<?, ?it/s]\u001b[A\n",
            "  0% 1/8092 [00:00<40:11,  3.35it/s]\u001b[A\n",
            "  0% 8/8092 [00:00<05:28, 24.58it/s]\u001b[A\n",
            "  0% 16/8092 [00:00<03:20, 40.19it/s]\u001b[A\n",
            "  0% 23/8092 [00:00<02:44, 49.03it/s]\u001b[A\n",
            "  0% 30/8092 [00:00<02:27, 54.82it/s]\u001b[A\n",
            "  0% 38/8092 [00:00<02:15, 59.24it/s]\u001b[A\n",
            "  1% 45/8092 [00:00<02:13, 60.33it/s]\u001b[A\n",
            "  1% 52/8092 [00:01<02:08, 62.50it/s]\u001b[A\n",
            "  1% 60/8092 [00:01<02:02, 65.44it/s]\u001b[A\n",
            "  1% 67/8092 [00:01<02:01, 65.99it/s]\u001b[A\n",
            "  1% 74/8092 [00:01<01:59, 67.06it/s]\u001b[A\n",
            "  1% 81/8092 [00:01<02:12, 60.25it/s]\u001b[A\n",
            "  1% 88/8092 [00:01<02:21, 56.69it/s]\u001b[A\n",
            "  1% 94/8092 [00:01<02:23, 55.75it/s]\u001b[A\n",
            "  1% 100/8092 [00:01<02:25, 54.80it/s]\u001b[A\n",
            "  1% 106/8092 [00:02<02:33, 52.05it/s]\u001b[A\n",
            "  1% 112/8092 [00:02<02:32, 52.22it/s]\u001b[A\n",
            "  1% 118/8092 [00:02<02:38, 50.20it/s]\u001b[A\n",
            "  2% 124/8092 [00:02<02:33, 51.97it/s]\u001b[A\n",
            "  2% 130/8092 [00:02<02:28, 53.50it/s]\u001b[A\n",
            "  2% 136/8092 [00:02<02:30, 52.94it/s]\u001b[A\n",
            "  2% 142/8092 [00:02<02:31, 52.60it/s]\u001b[A\n",
            "  2% 148/8092 [00:02<02:35, 51.11it/s]\u001b[A\n",
            "  2% 154/8092 [00:02<02:37, 50.34it/s]\u001b[A\n",
            "  2% 160/8092 [00:03<02:46, 47.60it/s]\u001b[A\n",
            "  2% 165/8092 [00:03<02:49, 46.65it/s]\u001b[A\n",
            "  2% 170/8092 [00:03<02:53, 45.79it/s]\u001b[A\n",
            "  2% 175/8092 [00:03<02:57, 44.57it/s]\u001b[A\n",
            "  2% 180/8092 [00:03<02:57, 44.45it/s]\u001b[A\n",
            "  2% 187/8092 [00:03<02:40, 49.31it/s]\u001b[A\n",
            "  2% 194/8092 [00:03<02:25, 54.26it/s]\u001b[A\n",
            "  2% 201/8092 [00:03<02:16, 57.79it/s]\u001b[A\n",
            "  3% 208/8092 [00:03<02:10, 60.35it/s]\u001b[A\n",
            "  3% 215/8092 [00:04<02:09, 60.97it/s]\u001b[A\n",
            "  3% 222/8092 [00:04<02:04, 63.41it/s]\u001b[A\n",
            "  3% 229/8092 [00:04<02:00, 65.12it/s]\u001b[A\n",
            "  3% 236/8092 [00:04<01:59, 65.87it/s]\u001b[A\n",
            "  3% 243/8092 [00:04<01:58, 66.05it/s]\u001b[A\n",
            "training:   0% 499/100000 [27:20<90:09:34,  3.26s/it]\n",
            "  3% 257/8092 [00:04<01:56, 67.01it/s]\u001b[A\n",
            "  3% 265/8092 [00:04<01:54, 68.28it/s]\u001b[A\n",
            "  3% 272/8092 [00:04<01:57, 66.77it/s]\u001b[A\n",
            "  3% 279/8092 [00:05<01:56, 67.20it/s]\u001b[A\n",
            "  4% 286/8092 [00:05<01:58, 65.90it/s]\u001b[A\n",
            "  4% 293/8092 [00:05<01:57, 66.38it/s]\u001b[A\n",
            "  4% 300/8092 [00:05<01:57, 66.57it/s]\u001b[A\n",
            "  4% 308/8092 [00:05<01:54, 67.77it/s]\u001b[A\n",
            "  4% 315/8092 [00:05<01:53, 68.31it/s]\u001b[A\n",
            "  4% 322/8092 [00:05<01:53, 68.28it/s]\u001b[A\n",
            "  4% 330/8092 [00:05<01:52, 69.04it/s]\u001b[A\n",
            "  4% 337/8092 [00:05<01:53, 68.22it/s]\u001b[A\n",
            "  4% 344/8092 [00:05<01:52, 68.58it/s]\u001b[A\n",
            "  4% 351/8092 [00:06<01:56, 66.44it/s]\u001b[A\n",
            "  4% 358/8092 [00:06<01:56, 66.41it/s]\u001b[A\n",
            "  5% 365/8092 [00:06<01:58, 65.47it/s]\u001b[A\n",
            "  5% 372/8092 [00:06<01:55, 66.58it/s]\u001b[A\n",
            "  5% 379/8092 [00:06<01:54, 67.23it/s]\u001b[A\n",
            "  5% 386/8092 [00:06<01:53, 67.66it/s]\u001b[A\n",
            "  5% 394/8092 [00:06<01:52, 68.62it/s]\u001b[A\n",
            "  5% 401/8092 [00:06<01:52, 68.14it/s]\u001b[A\n",
            "  5% 408/8092 [00:06<01:52, 68.01it/s]\u001b[A\n",
            "  5% 415/8092 [00:07<01:55, 66.63it/s]\u001b[A\n",
            "  5% 422/8092 [00:07<01:57, 65.07it/s]\u001b[A\n",
            "  5% 429/8092 [00:07<01:56, 65.98it/s]\u001b[A\n",
            "  5% 436/8092 [00:07<01:56, 65.67it/s]\u001b[A\n",
            "  5% 443/8092 [00:07<01:55, 66.04it/s]\u001b[A\n",
            "  6% 450/8092 [00:07<01:54, 66.47it/s]\u001b[A\n",
            "  6% 457/8092 [00:07<01:53, 67.04it/s]\u001b[A\n",
            "  6% 464/8092 [00:07<01:55, 66.11it/s]\u001b[A\n",
            "  6% 471/8092 [00:07<01:54, 66.31it/s]\u001b[A\n",
            "  6% 478/8092 [00:07<01:53, 67.25it/s]\u001b[A\n",
            "  6% 485/8092 [00:08<01:53, 67.30it/s]\u001b[A\n",
            "  6% 492/8092 [00:08<01:55, 65.71it/s]\u001b[A\n",
            "  6% 499/8092 [00:08<01:54, 66.34it/s]\u001b[A\n",
            "  6% 506/8092 [00:08<01:52, 67.32it/s]\u001b[A\n",
            "  6% 513/8092 [00:08<01:52, 67.66it/s]\u001b[A\n",
            "  6% 520/8092 [00:08<01:52, 67.10it/s]\u001b[A\n",
            "  7% 527/8092 [00:08<01:52, 67.18it/s]\u001b[A\n",
            "  7% 534/8092 [00:08<01:52, 67.42it/s]\u001b[A\n",
            "  7% 541/8092 [00:08<01:53, 66.50it/s]\u001b[A\n",
            "  7% 548/8092 [00:09<01:53, 66.43it/s]\u001b[A\n",
            "  7% 555/8092 [00:09<01:52, 67.23it/s]\u001b[A\n",
            "  7% 562/8092 [00:09<01:55, 65.05it/s]\u001b[A\n",
            "  7% 569/8092 [00:09<01:56, 64.79it/s]\u001b[A\n",
            "  7% 576/8092 [00:09<01:54, 65.60it/s]\u001b[A\n",
            "  7% 583/8092 [00:09<01:53, 65.89it/s]\u001b[A\n",
            "  7% 590/8092 [00:09<01:52, 66.83it/s]\u001b[A\n",
            "  7% 597/8092 [00:09<01:51, 67.34it/s]\u001b[A\n",
            "  7% 604/8092 [00:09<01:52, 66.54it/s]\u001b[A\n",
            "  8% 611/8092 [00:09<01:51, 67.11it/s]\u001b[A\n",
            "  8% 618/8092 [00:10<01:50, 67.52it/s]\u001b[A\n",
            "  8% 625/8092 [00:10<01:55, 64.72it/s]\u001b[A\n",
            "  8% 632/8092 [00:10<01:52, 66.04it/s]\u001b[A\n",
            "  8% 639/8092 [00:10<01:52, 66.23it/s]\u001b[A\n",
            "  8% 646/8092 [00:10<01:51, 66.88it/s]\u001b[A\n",
            "  8% 653/8092 [00:10<01:50, 67.08it/s]\u001b[A\n",
            "  8% 660/8092 [00:10<01:52, 66.20it/s]\u001b[A\n",
            "  8% 667/8092 [00:10<01:52, 66.29it/s]\u001b[A\n",
            "  8% 674/8092 [00:10<01:50, 66.83it/s]\u001b[A\n",
            "  8% 681/8092 [00:11<01:52, 66.13it/s]\u001b[A\n",
            "  9% 688/8092 [00:11<01:52, 65.98it/s]\u001b[A\n",
            "  9% 695/8092 [00:11<01:55, 64.24it/s]\u001b[A\n",
            "  9% 702/8092 [00:11<01:54, 64.26it/s]\u001b[A\n",
            "  9% 709/8092 [00:11<01:52, 65.62it/s]\u001b[A\n",
            "  9% 716/8092 [00:11<01:51, 66.07it/s]\u001b[A\n",
            "  9% 723/8092 [00:11<01:50, 66.81it/s]\u001b[A\n",
            "  9% 731/8092 [00:11<01:48, 68.06it/s]\u001b[A\n",
            "  9% 738/8092 [00:11<01:48, 68.01it/s]\u001b[A\n",
            "  9% 745/8092 [00:12<01:47, 68.04it/s]\u001b[A\n",
            "  9% 752/8092 [00:12<01:48, 67.61it/s]\u001b[A\n",
            "  9% 760/8092 [00:12<01:47, 68.31it/s]\u001b[A\n",
            "  9% 767/8092 [00:12<01:50, 66.39it/s]\u001b[A\n",
            " 10% 775/8092 [00:12<01:48, 67.74it/s]\u001b[A\n",
            " 10% 782/8092 [00:12<01:48, 67.31it/s]\u001b[A\n",
            " 10% 790/8092 [00:12<01:46, 68.51it/s]\u001b[A\n",
            " 10% 798/8092 [00:12<01:45, 69.24it/s]\u001b[A\n",
            " 10% 805/8092 [00:12<01:46, 68.25it/s]\u001b[A\n",
            " 10% 812/8092 [00:12<01:45, 68.69it/s]\u001b[A\n",
            " 10% 819/8092 [00:13<01:46, 68.27it/s]\u001b[A\n",
            " 10% 826/8092 [00:13<01:45, 68.60it/s]\u001b[A\n",
            " 10% 833/8092 [00:13<01:48, 66.98it/s]\u001b[A\n",
            " 10% 840/8092 [00:13<01:46, 67.83it/s]\u001b[A\n",
            " 10% 847/8092 [00:13<01:46, 68.00it/s]\u001b[A\n",
            " 11% 854/8092 [00:13<01:52, 64.40it/s]\u001b[A\n",
            " 11% 861/8092 [00:13<02:06, 57.32it/s]\u001b[A\n",
            " 11% 867/8092 [00:13<02:17, 52.46it/s]\u001b[A\n",
            " 11% 873/8092 [00:14<02:18, 52.02it/s]\u001b[A\n",
            " 11% 879/8092 [00:14<02:18, 52.22it/s]\u001b[A\n",
            " 11% 885/8092 [00:14<02:17, 52.26it/s]\u001b[A\n",
            " 11% 891/8092 [00:14<02:26, 49.19it/s]\u001b[A\n",
            " 11% 897/8092 [00:14<02:23, 50.00it/s]\u001b[A\n",
            " 11% 903/8092 [00:14<02:19, 51.61it/s]\u001b[A\n",
            " 11% 909/8092 [00:14<02:14, 53.22it/s]\u001b[A\n",
            " 11% 915/8092 [00:14<02:13, 53.72it/s]\u001b[A\n",
            " 11% 921/8092 [00:14<02:16, 52.58it/s]\u001b[A\n",
            " 11% 927/8092 [00:15<02:21, 50.64it/s]\u001b[A\n",
            " 12% 933/8092 [00:15<02:20, 51.12it/s]\u001b[A\n",
            " 12% 939/8092 [00:15<02:29, 47.92it/s]\u001b[A\n",
            " 12% 944/8092 [00:15<02:39, 44.81it/s]\u001b[A\n",
            " 12% 949/8092 [00:15<02:39, 44.68it/s]\u001b[A\n",
            " 12% 954/8092 [00:15<02:40, 44.48it/s]\u001b[A\n",
            " 12% 959/8092 [00:15<02:41, 44.12it/s]\u001b[A\n",
            " 12% 965/8092 [00:15<02:29, 47.63it/s]\u001b[A\n",
            " 12% 972/8092 [00:16<02:14, 53.13it/s]\u001b[A\n",
            " 12% 980/8092 [00:16<02:01, 58.43it/s]\u001b[A\n",
            " 12% 987/8092 [00:16<01:55, 61.41it/s]\u001b[A\n",
            " 12% 994/8092 [00:16<01:52, 63.23it/s]\u001b[A\n",
            " 12% 1001/8092 [00:16<01:52, 62.98it/s]\u001b[A\n",
            " 12% 1008/8092 [00:16<01:49, 64.79it/s]\u001b[A\n",
            " 13% 1016/8092 [00:16<01:45, 66.76it/s]\u001b[A\n",
            " 13% 1024/8092 [00:16<01:43, 68.21it/s]\u001b[A\n",
            " 13% 1031/8092 [00:16<01:44, 67.36it/s]\u001b[A\n",
            " 13% 1038/8092 [00:17<01:43, 67.85it/s]\u001b[A\n",
            " 13% 1046/8092 [00:17<01:42, 68.73it/s]\u001b[A\n",
            " 13% 1053/8092 [00:17<01:41, 69.08it/s]\u001b[A\n",
            " 13% 1061/8092 [00:17<01:41, 69.58it/s]\u001b[A\n",
            " 13% 1068/8092 [00:17<01:42, 68.50it/s]\u001b[A\n",
            " 13% 1075/8092 [00:17<01:46, 65.71it/s]\u001b[A\n",
            " 13% 1082/8092 [00:17<01:45, 66.57it/s]\u001b[A\n",
            " 13% 1089/8092 [00:17<01:44, 67.22it/s]\u001b[A\n",
            " 14% 1096/8092 [00:17<01:44, 66.89it/s]\u001b[A\n",
            " 14% 1103/8092 [00:17<01:45, 66.33it/s]\u001b[A\n",
            " 14% 1110/8092 [00:18<01:45, 66.23it/s]\u001b[A\n",
            " 14% 1117/8092 [00:18<01:45, 66.12it/s]\u001b[A\n",
            " 14% 1124/8092 [00:18<01:45, 66.15it/s]\u001b[A\n",
            " 14% 1131/8092 [00:18<01:45, 66.27it/s]\u001b[A\n",
            " 14% 1138/8092 [00:18<01:47, 64.57it/s]\u001b[A\n",
            " 14% 1145/8092 [00:18<01:46, 65.35it/s]\u001b[A\n",
            " 14% 1152/8092 [00:18<01:46, 65.22it/s]\u001b[A\n",
            " 14% 1159/8092 [00:18<01:46, 65.27it/s]\u001b[A\n",
            " 14% 1166/8092 [00:18<01:46, 65.12it/s]\u001b[A\n",
            " 14% 1173/8092 [00:19<01:44, 66.42it/s]\u001b[A\n",
            " 15% 1180/8092 [00:19<01:43, 66.56it/s]\u001b[A\n",
            " 15% 1187/8092 [00:19<01:43, 66.42it/s]\u001b[A\n",
            " 15% 1194/8092 [00:19<01:42, 67.18it/s]\u001b[A\n",
            " 15% 1201/8092 [00:19<01:43, 66.88it/s]\u001b[A\n",
            " 15% 1208/8092 [00:19<01:44, 65.65it/s]\u001b[A\n",
            " 15% 1215/8092 [00:19<01:44, 65.89it/s]\u001b[A\n",
            " 15% 1223/8092 [00:19<01:41, 67.66it/s]\u001b[A\n",
            " 15% 1230/8092 [00:19<01:41, 67.69it/s]\u001b[A\n",
            " 15% 1237/8092 [00:19<01:41, 67.70it/s]\u001b[A\n",
            " 15% 1244/8092 [00:20<01:40, 68.13it/s]\u001b[A\n",
            " 15% 1251/8092 [00:20<01:40, 68.34it/s]\u001b[A\n",
            " 16% 1259/8092 [00:20<01:38, 69.32it/s]\u001b[A\n",
            " 16% 1266/8092 [00:20<01:38, 69.12it/s]\u001b[A\n",
            " 16% 1273/8092 [00:20<01:39, 68.54it/s]\u001b[A\n",
            " 16% 1280/8092 [00:20<01:42, 66.70it/s]\u001b[A\n",
            " 16% 1288/8092 [00:20<01:40, 67.88it/s]\u001b[A\n",
            " 16% 1295/8092 [00:20<01:41, 66.82it/s]\u001b[A\n",
            " 16% 1302/8092 [00:20<01:40, 67.25it/s]\u001b[A\n",
            " 16% 1309/8092 [00:21<01:41, 66.58it/s]\u001b[A\n",
            " 16% 1316/8092 [00:21<01:43, 65.68it/s]\u001b[A\n",
            " 16% 1323/8092 [00:21<01:42, 65.89it/s]\u001b[A\n",
            " 16% 1331/8092 [00:21<01:41, 66.51it/s]\u001b[A\n",
            " 17% 1338/8092 [00:21<01:40, 66.99it/s]\u001b[A\n",
            " 17% 1345/8092 [00:21<01:42, 65.96it/s]\u001b[A\n",
            " 17% 1352/8092 [00:21<01:42, 65.63it/s]\u001b[A\n",
            " 17% 1359/8092 [00:21<01:40, 66.78it/s]\u001b[A\n",
            " 17% 1366/8092 [00:21<01:40, 67.17it/s]\u001b[A\n",
            " 17% 1373/8092 [00:22<01:39, 67.78it/s]\u001b[A\n",
            " 17% 1380/8092 [00:22<01:38, 68.19it/s]\u001b[A\n",
            " 17% 1388/8092 [00:22<01:36, 69.38it/s]\u001b[A\n",
            " 17% 1395/8092 [00:22<01:37, 68.82it/s]\u001b[A\n",
            " 17% 1402/8092 [00:22<01:37, 68.80it/s]\u001b[A\n",
            " 17% 1409/8092 [00:22<01:37, 68.81it/s]\u001b[A\n",
            " 17% 1416/8092 [00:22<01:40, 66.55it/s]\u001b[A\n",
            " 18% 1423/8092 [00:22<01:40, 66.22it/s]\u001b[A\n",
            " 18% 1430/8092 [00:22<01:41, 65.63it/s]\u001b[A\n",
            " 18% 1437/8092 [00:22<01:39, 66.64it/s]\u001b[A\n",
            " 18% 1444/8092 [00:23<01:39, 66.72it/s]\u001b[A\n",
            " 18% 1452/8092 [00:23<01:37, 68.32it/s]\u001b[A\n",
            " 18% 1459/8092 [00:23<01:36, 68.38it/s]\u001b[A\n",
            " 18% 1467/8092 [00:23<01:35, 69.70it/s]\u001b[A\n",
            " 18% 1474/8092 [00:23<01:36, 68.31it/s]\u001b[A\n",
            " 18% 1481/8092 [00:23<01:36, 68.26it/s]\u001b[A\n",
            " 18% 1488/8092 [00:23<01:39, 66.34it/s]\u001b[A\n",
            " 18% 1495/8092 [00:23<01:39, 66.11it/s]\u001b[A\n",
            " 19% 1502/8092 [00:23<01:40, 65.87it/s]\u001b[A\n",
            " 19% 1509/8092 [00:24<01:38, 66.93it/s]\u001b[A\n",
            " 19% 1516/8092 [00:24<01:38, 66.59it/s]\u001b[A\n",
            " 19% 1523/8092 [00:24<01:37, 67.35it/s]\u001b[A\n",
            " 19% 1530/8092 [00:24<01:37, 66.99it/s]\u001b[A\n",
            " 19% 1537/8092 [00:24<01:42, 63.92it/s]\u001b[A\n",
            " 19% 1544/8092 [00:24<01:39, 65.57it/s]\u001b[A\n",
            " 19% 1551/8092 [00:24<01:41, 64.76it/s]\u001b[A\n",
            " 19% 1558/8092 [00:24<01:39, 65.72it/s]\u001b[A\n",
            " 19% 1565/8092 [00:24<01:39, 65.91it/s]\u001b[A\n",
            " 19% 1572/8092 [00:24<01:38, 65.99it/s]\u001b[A\n",
            " 20% 1579/8092 [00:25<01:40, 64.88it/s]\u001b[A\n",
            " 20% 1586/8092 [00:25<01:39, 65.18it/s]\u001b[A\n",
            " 20% 1593/8092 [00:25<01:40, 64.71it/s]\u001b[A\n",
            " 20% 1600/8092 [00:25<01:41, 63.91it/s]\u001b[A\n",
            " 20% 1607/8092 [00:25<01:40, 64.36it/s]\u001b[A\n",
            " 20% 1614/8092 [00:25<01:41, 64.02it/s]\u001b[A\n",
            " 20% 1621/8092 [00:25<01:42, 63.30it/s]\u001b[A\n",
            " 20% 1628/8092 [00:25<01:41, 63.45it/s]\u001b[A\n",
            " 20% 1635/8092 [00:26<01:56, 55.47it/s]\u001b[A\n",
            " 20% 1641/8092 [00:26<02:00, 53.63it/s]\u001b[A\n",
            " 20% 1647/8092 [00:26<02:00, 53.36it/s]\u001b[A\n",
            " 20% 1653/8092 [00:26<02:01, 52.79it/s]\u001b[A\n",
            " 21% 1659/8092 [00:26<02:01, 53.15it/s]\u001b[A\n",
            " 21% 1665/8092 [00:26<02:02, 52.41it/s]\u001b[A\n",
            " 21% 1671/8092 [00:26<02:05, 51.01it/s]\u001b[A\n",
            " 21% 1677/8092 [00:26<02:04, 51.53it/s]\u001b[A\n",
            " 21% 1683/8092 [00:26<02:05, 51.16it/s]\u001b[A\n",
            " 21% 1689/8092 [00:27<02:02, 52.40it/s]\u001b[A\n",
            " 21% 1695/8092 [00:27<02:06, 50.69it/s]\u001b[A\n",
            " 21% 1701/8092 [00:27<02:12, 48.18it/s]\u001b[A\n",
            " 21% 1706/8092 [00:27<02:12, 48.20it/s]\u001b[A\n",
            " 21% 1711/8092 [00:27<02:21, 45.16it/s]\u001b[A\n",
            " 21% 1716/8092 [00:27<02:19, 45.72it/s]\u001b[A\n",
            " 21% 1721/8092 [00:27<02:22, 44.61it/s]\u001b[A\n",
            " 21% 1726/8092 [00:27<02:27, 43.17it/s]\u001b[A\n",
            " 21% 1731/8092 [00:28<02:28, 42.91it/s]\u001b[A\n",
            " 21% 1737/8092 [00:28<02:14, 47.22it/s]\u001b[A\n",
            " 22% 1744/8092 [00:28<02:01, 52.23it/s]\u001b[A\n",
            " 22% 1751/8092 [00:28<01:51, 56.72it/s]\u001b[A\n",
            " 22% 1758/8092 [00:28<01:47, 59.10it/s]\u001b[A\n",
            " 22% 1765/8092 [00:28<01:44, 60.31it/s]\u001b[A\n",
            " 22% 1772/8092 [00:28<01:44, 60.36it/s]\u001b[A\n",
            " 22% 1779/8092 [00:28<01:49, 57.73it/s]\u001b[A\n",
            " 22% 1785/8092 [00:28<01:49, 57.78it/s]\u001b[A\n",
            " 22% 1792/8092 [00:29<01:44, 60.34it/s]\u001b[A\n",
            " 22% 1799/8092 [00:29<01:42, 61.67it/s]\u001b[A\n",
            " 22% 1806/8092 [00:29<01:41, 61.84it/s]\u001b[A\n",
            " 22% 1813/8092 [00:29<01:42, 61.41it/s]\u001b[A\n",
            " 22% 1820/8092 [00:29<01:42, 60.94it/s]\u001b[A\n",
            " 23% 1827/8092 [00:29<01:43, 60.35it/s]\u001b[A\n",
            " 23% 1834/8092 [00:29<01:44, 60.06it/s]\u001b[A\n",
            " 23% 1841/8092 [00:29<01:44, 59.87it/s]\u001b[A\n",
            " 23% 1847/8092 [00:29<01:44, 59.82it/s]\u001b[A\n",
            " 23% 1853/8092 [00:30<01:45, 59.28it/s]\u001b[A\n",
            " 23% 1859/8092 [00:30<01:44, 59.45it/s]\u001b[A\n",
            " 23% 1866/8092 [00:30<01:43, 60.02it/s]\u001b[A\n",
            " 23% 1872/8092 [00:30<01:43, 59.82it/s]\u001b[A\n",
            " 23% 1878/8092 [00:30<01:43, 59.82it/s]\u001b[A\n",
            " 23% 1884/8092 [00:30<01:44, 59.58it/s]\u001b[A\n",
            " 23% 1890/8092 [00:30<01:45, 59.04it/s]\u001b[A\n",
            " 23% 1896/8092 [00:30<01:44, 59.15it/s]\u001b[A\n",
            " 24% 1902/8092 [00:30<01:47, 57.82it/s]\u001b[A\n",
            " 24% 1909/8092 [00:30<01:44, 59.10it/s]\u001b[A\n",
            " 24% 1915/8092 [00:31<01:44, 58.84it/s]\u001b[A\n",
            " 24% 1921/8092 [00:31<01:45, 58.70it/s]\u001b[A\n",
            " 24% 1927/8092 [00:31<01:45, 58.68it/s]\u001b[A\n",
            " 24% 1933/8092 [00:31<01:44, 58.66it/s]\u001b[A\n",
            " 24% 1939/8092 [00:31<01:46, 57.78it/s]\u001b[A\n",
            " 24% 1945/8092 [00:31<01:48, 56.52it/s]\u001b[A\n",
            " 24% 1951/8092 [00:31<01:49, 56.09it/s]\u001b[A\n",
            " 24% 1957/8092 [00:31<01:53, 53.91it/s]\u001b[A\n",
            " 24% 1963/8092 [00:31<01:56, 52.58it/s]\u001b[A\n",
            " 24% 1969/8092 [00:32<01:58, 51.61it/s]\u001b[A\n",
            " 24% 1975/8092 [00:32<02:00, 50.91it/s]\u001b[A\n",
            " 24% 1981/8092 [00:32<02:01, 50.46it/s]\u001b[A\n",
            " 25% 1987/8092 [00:32<02:02, 49.64it/s]\u001b[A\n",
            " 25% 1993/8092 [00:32<02:02, 49.68it/s]\u001b[A\n",
            " 25% 1999/8092 [00:32<02:02, 49.93it/s]\u001b[A\n",
            " 25% 2004/8092 [00:32<02:02, 49.69it/s]\u001b[A\n",
            " 25% 2009/8092 [00:32<02:02, 49.64it/s]\u001b[A\n",
            " 25% 2014/8092 [00:33<02:02, 49.49it/s]\u001b[A\n",
            " 25% 2019/8092 [00:33<02:02, 49.48it/s]\u001b[A\n",
            " 25% 2024/8092 [00:33<02:02, 49.50it/s]\u001b[A\n",
            " 25% 2029/8092 [00:33<02:02, 49.41it/s]\u001b[A\n",
            " 25% 2034/8092 [00:33<02:02, 49.30it/s]\u001b[A\n",
            " 25% 2039/8092 [00:33<02:02, 49.28it/s]\u001b[A\n",
            " 25% 2044/8092 [00:33<02:03, 49.06it/s]\u001b[A\n",
            " 25% 2049/8092 [00:33<02:03, 49.12it/s]\u001b[A\n",
            " 25% 2054/8092 [00:33<02:02, 49.20it/s]\u001b[A\n",
            " 25% 2059/8092 [00:33<02:02, 49.27it/s]\u001b[A\n",
            " 26% 2064/8092 [00:34<02:02, 49.20it/s]\u001b[A\n",
            " 26% 2070/8092 [00:34<02:01, 49.46it/s]\u001b[A\n",
            " 26% 2075/8092 [00:34<02:01, 49.47it/s]\u001b[A\n",
            " 26% 2080/8092 [00:34<02:01, 49.56it/s]\u001b[A\n",
            " 26% 2085/8092 [00:34<02:01, 49.46it/s]\u001b[A\n",
            " 26% 2090/8092 [00:34<02:02, 49.10it/s]\u001b[A\n",
            " 26% 2095/8092 [00:34<02:03, 48.63it/s]\u001b[A\n",
            " 26% 2100/8092 [00:34<02:03, 48.36it/s]\u001b[A\n",
            " 26% 2105/8092 [00:34<02:05, 47.81it/s]\u001b[A\n",
            " 26% 2110/8092 [00:34<02:04, 48.21it/s]\u001b[A\n",
            " 26% 2115/8092 [00:35<02:03, 48.42it/s]\u001b[A\n",
            " 26% 2120/8092 [00:35<02:03, 48.38it/s]\u001b[A\n",
            " 26% 2125/8092 [00:35<02:02, 48.54it/s]\u001b[A\n",
            " 26% 2130/8092 [00:35<02:03, 48.20it/s]\u001b[A\n",
            " 26% 2135/8092 [00:35<02:04, 47.82it/s]\u001b[A\n",
            " 26% 2140/8092 [00:35<02:04, 47.86it/s]\u001b[A\n",
            " 27% 2145/8092 [00:35<02:04, 47.90it/s]\u001b[A\n",
            " 27% 2150/8092 [00:35<02:03, 48.05it/s]\u001b[A\n",
            " 27% 2155/8092 [00:35<02:03, 48.03it/s]\u001b[A\n",
            " 27% 2160/8092 [00:36<02:03, 47.89it/s]\u001b[A\n",
            " 27% 2165/8092 [00:36<02:03, 47.91it/s]\u001b[A\n",
            " 27% 2170/8092 [00:36<02:02, 48.15it/s]\u001b[A\n",
            " 27% 2175/8092 [00:36<02:02, 48.22it/s]\u001b[A\n",
            " 27% 2180/8092 [00:36<02:02, 48.14it/s]\u001b[A\n",
            " 27% 2185/8092 [00:36<02:03, 47.65it/s]\u001b[A\n",
            " 27% 2190/8092 [00:36<02:04, 47.43it/s]\u001b[A\n",
            " 27% 2195/8092 [00:36<02:03, 47.57it/s]\u001b[A\n",
            " 27% 2200/8092 [00:36<02:03, 47.72it/s]\u001b[A\n",
            " 27% 2205/8092 [00:36<02:04, 47.47it/s]\u001b[A\n",
            " 27% 2210/8092 [00:37<02:04, 47.19it/s]\u001b[A\n",
            " 27% 2215/8092 [00:37<02:03, 47.44it/s]\u001b[A\n",
            " 27% 2220/8092 [00:37<02:03, 47.68it/s]\u001b[A\n",
            " 27% 2225/8092 [00:37<02:03, 47.55it/s]\u001b[A\n",
            " 28% 2230/8092 [00:37<02:04, 46.95it/s]\u001b[A\n",
            " 28% 2235/8092 [00:37<02:04, 46.97it/s]\u001b[A\n",
            " 28% 2240/8092 [00:37<02:03, 47.26it/s]\u001b[A\n",
            " 28% 2245/8092 [00:37<02:04, 46.96it/s]\u001b[A\n",
            " 28% 2250/8092 [00:37<02:05, 46.53it/s]\u001b[A\n",
            " 28% 2255/8092 [00:38<02:04, 46.86it/s]\u001b[A\n",
            " 28% 2260/8092 [00:38<02:11, 44.30it/s]\u001b[A\n",
            " 28% 2265/8092 [00:38<02:10, 44.61it/s]\u001b[A\n",
            " 28% 2270/8092 [00:38<02:08, 45.23it/s]\u001b[A\n",
            " 28% 2275/8092 [00:38<02:07, 45.77it/s]\u001b[A\n",
            " 28% 2280/8092 [00:38<02:04, 46.85it/s]\u001b[A\n",
            " 28% 2286/8092 [00:38<02:00, 48.20it/s]\u001b[A\n",
            " 28% 2291/8092 [00:38<02:00, 48.22it/s]\u001b[A\n",
            " 28% 2296/8092 [00:38<02:03, 47.00it/s]\u001b[A\n",
            " 28% 2301/8092 [00:39<02:01, 47.78it/s]\u001b[A\n",
            " 28% 2306/8092 [00:39<02:01, 47.61it/s]\u001b[A\n",
            " 29% 2311/8092 [00:39<02:02, 47.10it/s]\u001b[A\n",
            " 29% 2316/8092 [00:39<02:02, 47.04it/s]\u001b[A\n",
            " 29% 2321/8092 [00:39<02:04, 46.21it/s]\u001b[A\n",
            " 29% 2326/8092 [00:39<02:07, 45.05it/s]\u001b[A\n",
            " 29% 2331/8092 [00:39<02:06, 45.52it/s]\u001b[A\n",
            " 29% 2336/8092 [00:39<02:05, 45.79it/s]\u001b[A\n",
            " 29% 2341/8092 [00:39<02:08, 44.74it/s]\u001b[A\n",
            " 29% 2346/8092 [00:40<02:11, 43.80it/s]\u001b[A\n",
            " 29% 2351/8092 [00:40<02:12, 43.22it/s]\u001b[A\n",
            " 29% 2356/8092 [00:40<02:13, 42.85it/s]\u001b[A\n",
            " 29% 2361/8092 [00:40<02:12, 43.27it/s]\u001b[A\n",
            " 29% 2368/8092 [00:40<01:54, 49.82it/s]\u001b[A\n",
            " 29% 2374/8092 [00:40<01:58, 48.08it/s]\u001b[A\n",
            " 29% 2379/8092 [00:40<02:02, 46.82it/s]\u001b[A\n",
            " 29% 2384/8092 [00:40<02:03, 46.12it/s]\u001b[A\n",
            " 30% 2389/8092 [00:40<02:04, 45.71it/s]\u001b[A\n",
            " 30% 2394/8092 [00:41<02:05, 45.40it/s]\u001b[A\n",
            " 30% 2399/8092 [00:41<02:06, 44.97it/s]\u001b[A\n",
            " 30% 2404/8092 [00:41<02:06, 44.80it/s]\u001b[A\n",
            " 30% 2409/8092 [00:41<02:06, 44.95it/s]\u001b[A\n",
            " 30% 2414/8092 [00:41<02:06, 44.82it/s]\u001b[A\n",
            " 30% 2419/8092 [00:41<02:06, 44.91it/s]\u001b[A\n",
            " 30% 2424/8092 [00:41<02:07, 44.54it/s]\u001b[A\n",
            " 30% 2429/8092 [00:41<02:07, 44.46it/s]\u001b[A\n",
            " 30% 2434/8092 [00:41<02:07, 44.53it/s]\u001b[A\n",
            " 30% 2439/8092 [00:42<02:06, 44.57it/s]\u001b[A\n",
            " 30% 2444/8092 [00:42<02:06, 44.59it/s]\u001b[A\n",
            " 30% 2449/8092 [00:42<02:09, 43.66it/s]\u001b[A\n",
            " 30% 2454/8092 [00:42<02:06, 44.41it/s]\u001b[A\n",
            " 30% 2459/8092 [00:42<02:07, 44.33it/s]\u001b[A\n",
            " 30% 2464/8092 [00:42<02:07, 44.28it/s]\u001b[A\n",
            " 31% 2469/8092 [00:42<02:08, 43.90it/s]\u001b[A\n",
            " 31% 2474/8092 [00:42<02:09, 43.50it/s]\u001b[A\n",
            " 31% 2479/8092 [00:42<02:09, 43.27it/s]\u001b[A\n",
            " 31% 2484/8092 [00:43<02:09, 43.18it/s]\u001b[A\n",
            " 31% 2489/8092 [00:43<02:10, 43.03it/s]\u001b[A\n",
            " 31% 2494/8092 [00:43<02:10, 42.98it/s]\u001b[A\n",
            " 31% 2499/8092 [00:43<02:10, 43.00it/s]\u001b[A\n",
            " 31% 2504/8092 [00:43<02:10, 42.96it/s]\u001b[A\n",
            " 31% 2509/8092 [00:43<02:09, 42.95it/s]\u001b[A\n",
            " 31% 2514/8092 [00:43<02:10, 42.85it/s]\u001b[A\n",
            " 31% 2519/8092 [00:43<02:09, 42.90it/s]\u001b[A\n",
            " 31% 2524/8092 [00:44<02:10, 42.52it/s]\u001b[A\n",
            " 31% 2529/8092 [00:44<02:10, 42.74it/s]\u001b[A\n",
            " 31% 2534/8092 [00:44<02:10, 42.68it/s]\u001b[A\n",
            " 31% 2539/8092 [00:44<02:09, 42.76it/s]\u001b[A\n",
            " 31% 2544/8092 [00:44<02:09, 42.98it/s]\u001b[A\n",
            " 32% 2549/8092 [00:44<02:09, 42.73it/s]\u001b[A\n",
            " 32% 2554/8092 [00:44<02:10, 42.46it/s]\u001b[A\n",
            " 32% 2559/8092 [00:44<02:10, 42.35it/s]\u001b[A\n",
            " 32% 2564/8092 [00:44<02:09, 42.75it/s]\u001b[A\n",
            " 32% 2569/8092 [00:45<02:09, 42.68it/s]\u001b[A\n",
            " 32% 2574/8092 [00:45<02:09, 42.50it/s]\u001b[A\n",
            " 32% 2579/8092 [00:45<02:09, 42.50it/s]\u001b[A\n",
            " 32% 2584/8092 [00:45<02:09, 42.63it/s]\u001b[A\n",
            " 32% 2589/8092 [00:45<02:09, 42.37it/s]\u001b[A\n",
            " 32% 2594/8092 [00:45<02:09, 42.55it/s]\u001b[A\n",
            " 32% 2599/8092 [00:45<02:09, 42.57it/s]\u001b[A\n",
            " 32% 2604/8092 [00:45<02:09, 42.38it/s]\u001b[A\n",
            " 32% 2609/8092 [00:46<02:09, 42.38it/s]\u001b[A\n",
            " 32% 2614/8092 [00:46<02:08, 42.66it/s]\u001b[A\n",
            " 32% 2619/8092 [00:46<02:08, 42.51it/s]\u001b[A\n",
            " 32% 2624/8092 [00:46<02:09, 42.14it/s]\u001b[A\n",
            " 32% 2629/8092 [00:46<02:08, 42.53it/s]\u001b[A\n",
            " 33% 2634/8092 [00:46<02:08, 42.42it/s]\u001b[A\n",
            " 33% 2639/8092 [00:46<02:08, 42.54it/s]\u001b[A\n",
            " 33% 2644/8092 [00:46<02:08, 42.43it/s]\u001b[A\n",
            " 33% 2649/8092 [00:46<02:09, 42.18it/s]\u001b[A\n",
            " 33% 2654/8092 [00:47<02:08, 42.26it/s]\u001b[A\n",
            " 33% 2659/8092 [00:47<02:08, 42.37it/s]\u001b[A\n",
            " 33% 2664/8092 [00:47<02:09, 41.90it/s]\u001b[A\n",
            " 33% 2669/8092 [00:47<02:09, 41.93it/s]\u001b[A\n",
            " 33% 2674/8092 [00:47<02:09, 41.81it/s]\u001b[A\n",
            " 33% 2679/8092 [00:47<02:08, 42.00it/s]\u001b[A\n",
            " 33% 2684/8092 [00:47<02:08, 42.11it/s]\u001b[A\n",
            " 33% 2689/8092 [00:47<02:09, 41.87it/s]\u001b[A\n",
            " 33% 2694/8092 [00:48<02:09, 41.60it/s]\u001b[A\n",
            " 33% 2699/8092 [00:48<02:08, 42.09it/s]\u001b[A\n",
            " 33% 2704/8092 [00:48<02:08, 41.84it/s]\u001b[A\n",
            " 33% 2709/8092 [00:48<02:08, 41.81it/s]\u001b[A\n",
            " 34% 2714/8092 [00:48<02:08, 41.74it/s]\u001b[A\n",
            " 34% 2719/8092 [00:48<02:09, 41.51it/s]\u001b[A\n",
            " 34% 2724/8092 [00:48<02:09, 41.59it/s]\u001b[A\n",
            " 34% 2729/8092 [00:48<02:09, 41.34it/s]\u001b[A\n",
            " 34% 2734/8092 [00:49<02:09, 41.52it/s]\u001b[A\n",
            " 34% 2739/8092 [00:49<02:08, 41.55it/s]\u001b[A\n",
            " 34% 2744/8092 [00:49<02:10, 41.10it/s]\u001b[A\n",
            " 34% 2749/8092 [00:49<02:08, 41.45it/s]\u001b[A\n",
            " 34% 2754/8092 [00:49<02:08, 41.44it/s]\u001b[A\n",
            " 34% 2759/8092 [00:49<02:08, 41.37it/s]\u001b[A\n",
            " 34% 2764/8092 [00:49<02:09, 41.18it/s]\u001b[A\n",
            " 34% 2769/8092 [00:49<02:09, 40.99it/s]\u001b[A\n",
            " 34% 2774/8092 [00:49<02:09, 41.20it/s]\u001b[A\n",
            " 34% 2779/8092 [00:50<02:09, 41.03it/s]\u001b[A\n",
            " 34% 2784/8092 [00:50<02:09, 40.84it/s]\u001b[A\n",
            " 34% 2789/8092 [00:50<02:11, 40.43it/s]\u001b[A\n",
            " 35% 2794/8092 [00:50<02:12, 39.92it/s]\u001b[A\n",
            " 35% 2799/8092 [00:50<02:10, 40.68it/s]\u001b[A\n",
            " 35% 2804/8092 [00:50<02:10, 40.40it/s]\u001b[A\n",
            " 35% 2809/8092 [00:50<02:10, 40.61it/s]\u001b[A\n",
            " 35% 2814/8092 [00:50<02:10, 40.39it/s]\u001b[A\n",
            " 35% 2819/8092 [00:51<02:09, 40.68it/s]\u001b[A\n",
            " 35% 2824/8092 [00:51<02:10, 40.49it/s]\u001b[A\n",
            " 35% 2829/8092 [00:51<02:09, 40.64it/s]\u001b[A\n",
            " 35% 2834/8092 [00:51<02:09, 40.46it/s]\u001b[A\n",
            " 35% 2839/8092 [00:51<02:11, 40.05it/s]\u001b[A\n",
            " 35% 2844/8092 [00:51<02:10, 40.13it/s]\u001b[A\n",
            " 35% 2849/8092 [00:51<02:10, 40.02it/s]\u001b[A\n",
            " 35% 2854/8092 [00:51<02:09, 40.52it/s]\u001b[A\n",
            " 35% 2859/8092 [00:52<02:09, 40.34it/s]\u001b[A\n",
            " 35% 2864/8092 [00:52<02:09, 40.23it/s]\u001b[A\n",
            " 35% 2869/8092 [00:52<02:10, 40.03it/s]\u001b[A\n",
            " 36% 2874/8092 [00:52<02:10, 39.98it/s]\u001b[A\n",
            " 36% 2878/8092 [00:52<02:13, 39.16it/s]\u001b[A\n",
            " 36% 2883/8092 [00:52<02:07, 41.01it/s]\u001b[A\n",
            " 36% 2888/8092 [00:52<02:07, 40.72it/s]\u001b[A\n",
            " 36% 2893/8092 [00:52<02:08, 40.59it/s]\u001b[A\n",
            " 36% 2898/8092 [00:53<02:08, 40.31it/s]\u001b[A\n",
            " 36% 2903/8092 [00:53<02:09, 40.06it/s]\u001b[A\n",
            " 36% 2908/8092 [00:53<02:09, 40.18it/s]\u001b[A\n",
            " 36% 2913/8092 [00:53<02:09, 40.08it/s]\u001b[A\n",
            " 36% 2918/8092 [00:53<02:09, 40.10it/s]\u001b[A\n",
            " 36% 2923/8092 [00:53<02:09, 39.88it/s]\u001b[A\n",
            " 36% 2928/8092 [00:53<02:09, 39.85it/s]\u001b[A\n",
            " 36% 2932/8092 [00:53<02:09, 39.77it/s]\u001b[A\n",
            " 36% 2936/8092 [00:54<02:10, 39.59it/s]\u001b[A\n",
            " 36% 2941/8092 [00:54<02:09, 39.86it/s]\u001b[A\n",
            " 36% 2945/8092 [00:54<02:09, 39.71it/s]\u001b[A\n",
            " 36% 2949/8092 [00:54<02:09, 39.77it/s]\u001b[A\n",
            " 36% 2953/8092 [00:54<02:09, 39.82it/s]\u001b[A\n",
            " 37% 2957/8092 [00:54<02:09, 39.71it/s]\u001b[A\n",
            " 37% 2961/8092 [00:54<02:09, 39.71it/s]\u001b[A\n",
            " 37% 2965/8092 [00:54<02:10, 39.41it/s]\u001b[A\n",
            " 37% 2969/8092 [00:54<02:11, 38.91it/s]\u001b[A\n",
            " 37% 2973/8092 [00:54<02:12, 38.76it/s]\u001b[A\n",
            " 37% 2977/8092 [00:55<02:14, 38.14it/s]\u001b[A\n",
            " 37% 2981/8092 [00:55<02:16, 37.46it/s]\u001b[A\n",
            " 37% 2985/8092 [00:55<02:17, 37.10it/s]\u001b[A\n",
            " 37% 2989/8092 [00:55<02:21, 36.15it/s]\u001b[A\n",
            " 37% 2993/8092 [00:55<02:19, 36.64it/s]\u001b[A\n",
            " 37% 2997/8092 [00:55<02:19, 36.40it/s]\u001b[A\n",
            " 37% 3001/8092 [00:55<02:21, 35.95it/s]\u001b[A\n",
            " 37% 3005/8092 [00:55<02:21, 35.96it/s]\u001b[A\n",
            " 37% 3009/8092 [00:55<02:21, 36.03it/s]\u001b[A\n",
            " 37% 3013/8092 [00:56<02:20, 36.08it/s]\u001b[A\n",
            " 37% 3017/8092 [00:56<02:21, 35.96it/s]\u001b[A\n",
            " 37% 3021/8092 [00:56<02:21, 35.88it/s]\u001b[A\n",
            " 37% 3025/8092 [00:56<02:20, 35.97it/s]\u001b[A\n",
            " 37% 3029/8092 [00:56<02:21, 35.76it/s]\u001b[A\n",
            " 37% 3033/8092 [00:56<02:20, 35.94it/s]\u001b[A\n",
            " 38% 3037/8092 [00:56<02:23, 35.33it/s]\u001b[A\n",
            " 38% 3041/8092 [00:56<02:20, 35.87it/s]\u001b[A\n",
            " 38% 3045/8092 [00:56<02:20, 35.82it/s]\u001b[A\n",
            " 38% 3049/8092 [00:57<02:21, 35.76it/s]\u001b[A\n",
            " 38% 3053/8092 [00:57<02:20, 35.79it/s]\u001b[A\n",
            " 38% 3057/8092 [00:57<02:20, 35.76it/s]\u001b[A\n",
            " 38% 3061/8092 [00:57<02:20, 35.69it/s]\u001b[A\n",
            " 38% 3065/8092 [00:57<02:20, 35.74it/s]\u001b[A\n",
            " 38% 3069/8092 [00:57<02:20, 35.81it/s]\u001b[A\n",
            " 38% 3073/8092 [00:57<02:22, 35.17it/s]\u001b[A\n",
            " 38% 3077/8092 [00:57<02:20, 35.68it/s]\u001b[A\n",
            " 38% 3081/8092 [00:57<02:20, 35.59it/s]\u001b[A\n",
            " 38% 3085/8092 [00:58<02:21, 35.51it/s]\u001b[A\n",
            " 38% 3089/8092 [00:58<02:20, 35.63it/s]\u001b[A\n",
            " 38% 3093/8092 [00:58<02:20, 35.61it/s]\u001b[A\n",
            " 38% 3097/8092 [00:58<02:22, 34.96it/s]\u001b[A\n",
            " 38% 3101/8092 [00:58<02:18, 35.92it/s]\u001b[A\n",
            " 38% 3105/8092 [00:58<02:19, 35.87it/s]\u001b[A\n",
            " 38% 3109/8092 [00:58<02:19, 35.67it/s]\u001b[A\n",
            " 38% 3113/8092 [00:58<02:23, 34.80it/s]\u001b[A\n",
            " 39% 3117/8092 [00:58<02:19, 35.71it/s]\u001b[A\n",
            " 39% 3121/8092 [00:59<02:20, 35.44it/s]\u001b[A\n",
            " 39% 3125/8092 [00:59<02:19, 35.57it/s]\u001b[A\n",
            " 39% 3129/8092 [00:59<02:20, 35.32it/s]\u001b[A\n",
            " 39% 3133/8092 [00:59<02:19, 35.49it/s]\u001b[A\n",
            " 39% 3137/8092 [00:59<02:19, 35.53it/s]\u001b[A\n",
            " 39% 3141/8092 [00:59<02:19, 35.58it/s]\u001b[A\n",
            " 39% 3145/8092 [00:59<02:19, 35.40it/s]\u001b[A\n",
            " 39% 3149/8092 [00:59<02:19, 35.45it/s]\u001b[A\n",
            " 39% 3153/8092 [01:00<02:19, 35.40it/s]\u001b[A\n",
            " 39% 3157/8092 [01:00<02:19, 35.45it/s]\u001b[A\n",
            " 39% 3161/8092 [01:00<02:20, 35.08it/s]\u001b[A\n",
            " 39% 3165/8092 [01:00<02:19, 35.24it/s]\u001b[A\n",
            " 39% 3169/8092 [01:00<02:19, 35.31it/s]\u001b[A\n",
            " 39% 3173/8092 [01:00<02:20, 35.12it/s]\u001b[A\n",
            " 39% 3177/8092 [01:00<02:20, 34.92it/s]\u001b[A\n",
            " 39% 3181/8092 [01:00<02:21, 34.70it/s]\u001b[A\n",
            " 39% 3185/8092 [01:00<02:21, 34.61it/s]\u001b[A\n",
            " 39% 3189/8092 [01:01<02:22, 34.33it/s]\u001b[A\n",
            " 39% 3193/8092 [01:01<02:21, 34.58it/s]\u001b[A\n",
            " 40% 3197/8092 [01:01<02:21, 34.51it/s]\u001b[A\n",
            " 40% 3201/8092 [01:01<02:22, 34.44it/s]\u001b[A\n",
            " 40% 3205/8092 [01:01<02:21, 34.53it/s]\u001b[A\n",
            " 40% 3209/8092 [01:01<02:21, 34.49it/s]\u001b[A\n",
            " 40% 3213/8092 [01:01<02:21, 34.51it/s]\u001b[A\n",
            " 40% 3217/8092 [01:01<02:23, 34.02it/s]\u001b[A\n",
            " 40% 3221/8092 [01:01<02:21, 34.32it/s]\u001b[A\n",
            " 40% 3225/8092 [01:02<02:21, 34.32it/s]\u001b[A\n",
            " 40% 3229/8092 [01:02<02:21, 34.30it/s]\u001b[A\n",
            " 40% 3233/8092 [01:02<02:21, 34.29it/s]\u001b[A\n",
            " 40% 3237/8092 [01:02<02:21, 34.31it/s]\u001b[A\n",
            " 40% 3241/8092 [01:02<02:21, 34.20it/s]\u001b[A\n",
            " 40% 3245/8092 [01:02<02:25, 33.38it/s]\u001b[A\n",
            " 40% 3249/8092 [01:02<02:21, 34.11it/s]\u001b[A\n",
            " 40% 3253/8092 [01:02<02:23, 33.69it/s]\u001b[A\n",
            " 40% 3257/8092 [01:03<02:21, 34.24it/s]\u001b[A\n",
            " 40% 3261/8092 [01:03<02:21, 34.17it/s]\u001b[A\n",
            " 40% 3265/8092 [01:03<02:23, 33.61it/s]\u001b[A\n",
            " 40% 3269/8092 [01:03<02:20, 34.29it/s]\u001b[A\n",
            " 40% 3273/8092 [01:03<02:21, 34.04it/s]\u001b[A\n",
            " 40% 3277/8092 [01:03<02:20, 34.31it/s]\u001b[A\n",
            " 41% 3281/8092 [01:03<02:20, 34.15it/s]\u001b[A\n",
            " 41% 3285/8092 [01:03<02:21, 33.94it/s]\u001b[A\n",
            " 41% 3289/8092 [01:03<02:22, 33.63it/s]\u001b[A\n",
            " 41% 3293/8092 [01:04<02:21, 34.03it/s]\u001b[A\n",
            " 41% 3297/8092 [01:04<02:21, 33.96it/s]\u001b[A\n",
            " 41% 3301/8092 [01:04<02:21, 33.83it/s]\u001b[A\n",
            " 41% 3305/8092 [01:04<02:20, 33.95it/s]\u001b[A\n",
            " 41% 3309/8092 [01:04<02:21, 33.77it/s]\u001b[A\n",
            " 41% 3313/8092 [01:04<02:21, 33.70it/s]\u001b[A\n",
            " 41% 3317/8092 [01:04<02:20, 33.92it/s]\u001b[A\n",
            " 41% 3321/8092 [01:04<02:19, 34.22it/s]\u001b[A\n",
            " 41% 3325/8092 [01:05<02:20, 34.02it/s]\u001b[A\n",
            " 41% 3329/8092 [01:05<02:19, 34.06it/s]\u001b[A\n",
            " 41% 3333/8092 [01:05<02:21, 33.58it/s]\u001b[A\n",
            " 41% 3337/8092 [01:05<02:19, 34.01it/s]\u001b[A\n",
            " 41% 3341/8092 [01:05<02:19, 34.00it/s]\u001b[A\n",
            " 41% 3345/8092 [01:05<02:20, 33.88it/s]\u001b[A\n",
            " 41% 3349/8092 [01:05<02:20, 33.71it/s]\u001b[A\n",
            " 41% 3353/8092 [01:05<02:21, 33.57it/s]\u001b[A\n",
            " 41% 3357/8092 [01:05<02:21, 33.50it/s]\u001b[A\n",
            " 42% 3361/8092 [01:06<02:20, 33.56it/s]\u001b[A\n",
            " 42% 3365/8092 [01:06<02:21, 33.40it/s]\u001b[A\n",
            " 42% 3369/8092 [01:06<02:20, 33.54it/s]\u001b[A\n",
            " 42% 3373/8092 [01:06<02:20, 33.58it/s]\u001b[A\n",
            " 42% 3377/8092 [01:06<02:20, 33.53it/s]\u001b[A\n",
            " 42% 3381/8092 [01:06<02:21, 33.41it/s]\u001b[A\n",
            " 42% 3385/8092 [01:06<02:20, 33.53it/s]\u001b[A\n",
            " 42% 3389/8092 [01:06<02:20, 33.49it/s]\u001b[A\n",
            " 42% 3393/8092 [01:07<02:20, 33.35it/s]\u001b[A\n",
            " 42% 3397/8092 [01:07<02:20, 33.36it/s]\u001b[A\n",
            " 42% 3401/8092 [01:07<02:19, 33.52it/s]\u001b[A\n",
            " 42% 3405/8092 [01:07<02:20, 33.40it/s]\u001b[A\n",
            " 42% 3409/8092 [01:07<02:20, 33.31it/s]\u001b[A\n",
            " 42% 3413/8092 [01:07<02:20, 33.29it/s]\u001b[A\n",
            " 42% 3417/8092 [01:07<02:20, 33.32it/s]\u001b[A\n",
            " 42% 3421/8092 [01:07<02:21, 32.98it/s]\u001b[A\n",
            " 42% 3425/8092 [01:08<02:19, 33.38it/s]\u001b[A\n",
            " 42% 3429/8092 [01:08<02:19, 33.33it/s]\u001b[A\n",
            " 42% 3433/8092 [01:08<02:19, 33.37it/s]\u001b[A\n",
            " 42% 3437/8092 [01:08<02:19, 33.37it/s]\u001b[A\n",
            " 43% 3441/8092 [01:08<02:19, 33.28it/s]\u001b[A\n",
            " 43% 3445/8092 [01:08<02:20, 33.16it/s]\u001b[A\n",
            " 43% 3449/8092 [01:08<02:20, 33.08it/s]\u001b[A\n",
            " 43% 3453/8092 [01:08<02:20, 33.09it/s]\u001b[A\n",
            " 43% 3457/8092 [01:08<02:20, 32.94it/s]\u001b[A\n",
            " 43% 3461/8092 [01:09<02:20, 32.93it/s]\u001b[A\n",
            " 43% 3465/8092 [01:09<02:20, 32.96it/s]\u001b[A\n",
            " 43% 3469/8092 [01:09<02:20, 32.93it/s]\u001b[A\n",
            " 43% 3473/8092 [01:09<02:20, 32.93it/s]\u001b[A\n",
            " 43% 3477/8092 [01:09<02:19, 33.03it/s]\u001b[A\n",
            " 43% 3481/8092 [01:09<02:19, 33.00it/s]\u001b[A\n",
            " 43% 3485/8092 [01:09<02:20, 32.85it/s]\u001b[A\n",
            " 43% 3489/8092 [01:09<02:19, 32.96it/s]\u001b[A\n",
            " 43% 3493/8092 [01:10<02:19, 32.87it/s]\u001b[A\n",
            " 43% 3497/8092 [01:10<02:20, 32.81it/s]\u001b[A\n",
            " 43% 3501/8092 [01:10<02:19, 32.83it/s]\u001b[A\n",
            " 43% 3505/8092 [01:10<02:20, 32.76it/s]\u001b[A\n",
            " 43% 3509/8092 [01:10<02:20, 32.73it/s]\u001b[A\n",
            " 43% 3513/8092 [01:10<02:19, 32.76it/s]\u001b[A\n",
            " 43% 3517/8092 [01:10<02:19, 32.87it/s]\u001b[A\n",
            " 44% 3521/8092 [01:10<02:19, 32.81it/s]\u001b[A\n",
            " 44% 3525/8092 [01:11<02:19, 32.71it/s]\u001b[A\n",
            " 44% 3529/8092 [01:11<02:19, 32.67it/s]\u001b[A\n",
            " 44% 3533/8092 [01:11<02:19, 32.74it/s]\u001b[A\n",
            " 44% 3537/8092 [01:11<02:18, 32.77it/s]\u001b[A\n",
            " 44% 3541/8092 [01:11<02:19, 32.55it/s]\u001b[A\n",
            " 44% 3545/8092 [01:11<02:19, 32.59it/s]\u001b[A\n",
            " 44% 3549/8092 [01:11<02:19, 32.59it/s]\u001b[A\n",
            " 44% 3553/8092 [01:11<02:19, 32.58it/s]\u001b[A\n",
            " 44% 3557/8092 [01:12<02:20, 32.24it/s]\u001b[A\n",
            " 44% 3561/8092 [01:12<02:20, 32.21it/s]\u001b[A\n",
            " 44% 3565/8092 [01:12<02:19, 32.37it/s]\u001b[A\n",
            " 44% 3569/8092 [01:12<02:19, 32.52it/s]\u001b[A\n",
            " 44% 3573/8092 [01:12<02:18, 32.60it/s]\u001b[A\n",
            " 44% 3577/8092 [01:12<02:19, 32.46it/s]\u001b[A\n",
            " 44% 3581/8092 [01:12<02:18, 32.50it/s]\u001b[A\n",
            " 44% 3585/8092 [01:12<02:19, 32.20it/s]\u001b[A\n",
            " 44% 3589/8092 [01:13<02:19, 32.28it/s]\u001b[A\n",
            " 44% 3593/8092 [01:13<02:18, 32.37it/s]\u001b[A\n",
            " 44% 3597/8092 [01:13<02:17, 32.57it/s]\u001b[A\n",
            " 45% 3601/8092 [01:13<02:18, 32.53it/s]\u001b[A\n",
            " 45% 3605/8092 [01:13<02:18, 32.50it/s]\u001b[A\n",
            " 45% 3609/8092 [01:13<02:18, 32.44it/s]\u001b[A\n",
            " 45% 3613/8092 [01:13<02:18, 32.45it/s]\u001b[A\n",
            " 45% 3617/8092 [01:13<02:18, 32.36it/s]\u001b[A\n",
            " 45% 3621/8092 [01:14<02:18, 32.32it/s]\u001b[A\n",
            " 45% 3625/8092 [01:14<02:18, 32.34it/s]\u001b[A\n",
            " 45% 3629/8092 [01:14<02:17, 32.40it/s]\u001b[A\n",
            " 45% 3633/8092 [01:14<02:17, 32.54it/s]\u001b[A\n",
            " 45% 3637/8092 [01:14<02:17, 32.47it/s]\u001b[A\n",
            " 45% 3641/8092 [01:14<02:17, 32.31it/s]\u001b[A\n",
            " 45% 3645/8092 [01:14<02:18, 32.16it/s]\u001b[A\n",
            " 45% 3649/8092 [01:14<02:19, 31.83it/s]\u001b[A\n",
            " 45% 3653/8092 [01:15<02:18, 31.98it/s]\u001b[A\n",
            " 45% 3657/8092 [01:15<02:18, 31.95it/s]\u001b[A\n",
            " 45% 3661/8092 [01:15<02:17, 32.21it/s]\u001b[A\n",
            " 45% 3665/8092 [01:15<02:17, 32.10it/s]\u001b[A\n",
            " 45% 3669/8092 [01:15<02:18, 31.93it/s]\u001b[A\n",
            " 45% 3673/8092 [01:15<02:16, 32.41it/s]\u001b[A\n",
            " 45% 3677/8092 [01:15<02:16, 32.27it/s]\u001b[A\n",
            " 45% 3681/8092 [01:15<02:19, 31.67it/s]\u001b[A\n",
            " 46% 3685/8092 [01:16<02:16, 32.40it/s]\u001b[A\n",
            " 46% 3689/8092 [01:16<02:16, 32.17it/s]\u001b[A\n",
            " 46% 3693/8092 [01:16<02:18, 31.74it/s]\u001b[A\n",
            " 46% 3697/8092 [01:16<02:17, 32.01it/s]\u001b[A\n",
            " 46% 3701/8092 [01:16<02:16, 32.17it/s]\u001b[A\n",
            " 46% 3705/8092 [01:16<02:17, 31.89it/s]\u001b[A\n",
            " 46% 3709/8092 [01:16<02:18, 31.67it/s]\u001b[A\n",
            " 46% 3713/8092 [01:16<02:18, 31.64it/s]\u001b[A\n",
            " 46% 3717/8092 [01:17<02:17, 31.71it/s]\u001b[A\n",
            " 46% 3721/8092 [01:17<02:16, 32.09it/s]\u001b[A\n",
            " 46% 3725/8092 [01:17<02:16, 31.98it/s]\u001b[A\n",
            " 46% 3729/8092 [01:17<02:16, 32.02it/s]\u001b[A\n",
            " 46% 3733/8092 [01:17<02:16, 31.90it/s]\u001b[A\n",
            " 46% 3737/8092 [01:17<02:16, 31.88it/s]\u001b[A\n",
            " 46% 3741/8092 [01:17<02:16, 31.84it/s]\u001b[A\n",
            " 46% 3745/8092 [01:17<02:17, 31.66it/s]\u001b[A\n",
            " 46% 3749/8092 [01:18<02:17, 31.62it/s]\u001b[A\n",
            " 46% 3753/8092 [01:18<02:17, 31.53it/s]\u001b[A\n",
            " 46% 3757/8092 [01:18<02:17, 31.53it/s]\u001b[A\n",
            " 46% 3761/8092 [01:18<02:17, 31.53it/s]\u001b[A\n",
            " 47% 3765/8092 [01:18<02:17, 31.55it/s]\u001b[A\n",
            " 47% 3769/8092 [01:18<02:17, 31.43it/s]\u001b[A\n",
            " 47% 3773/8092 [01:18<02:17, 31.30it/s]\u001b[A\n",
            " 47% 3777/8092 [01:18<02:17, 31.27it/s]\u001b[A\n",
            " 47% 3781/8092 [01:19<02:17, 31.39it/s]\u001b[A\n",
            " 47% 3785/8092 [01:19<02:17, 31.41it/s]\u001b[A\n",
            " 47% 3789/8092 [01:19<02:17, 31.36it/s]\u001b[A\n",
            " 47% 3793/8092 [01:19<02:17, 31.32it/s]\u001b[A\n",
            " 47% 3797/8092 [01:19<02:16, 31.35it/s]\u001b[A\n",
            " 47% 3801/8092 [01:19<02:17, 31.21it/s]\u001b[A\n",
            " 47% 3805/8092 [01:19<02:17, 31.26it/s]\u001b[A\n",
            " 47% 3809/8092 [01:19<02:16, 31.28it/s]\u001b[A\n",
            " 47% 3813/8092 [01:20<02:18, 31.00it/s]\u001b[A\n",
            " 47% 3817/8092 [01:20<02:17, 31.13it/s]\u001b[A\n",
            " 47% 3821/8092 [01:20<02:17, 31.11it/s]\u001b[A\n",
            " 47% 3825/8092 [01:20<02:16, 31.15it/s]\u001b[A\n",
            " 47% 3829/8092 [01:20<02:16, 31.13it/s]\u001b[A\n",
            " 47% 3833/8092 [01:20<02:16, 31.17it/s]\u001b[A\n",
            " 47% 3837/8092 [01:20<02:16, 31.08it/s]\u001b[A\n",
            " 47% 3841/8092 [01:20<02:16, 31.11it/s]\u001b[A\n",
            " 48% 3845/8092 [01:21<02:16, 31.03it/s]\u001b[A\n",
            " 48% 3849/8092 [01:21<02:15, 31.32it/s]\u001b[A\n",
            " 48% 3853/8092 [01:21<02:15, 31.33it/s]\u001b[A\n",
            " 48% 3857/8092 [01:21<02:15, 31.28it/s]\u001b[A\n",
            " 48% 3861/8092 [01:21<02:15, 31.22it/s]\u001b[A\n",
            " 48% 3865/8092 [01:21<02:15, 31.14it/s]\u001b[A\n",
            " 48% 3869/8092 [01:21<02:15, 31.27it/s]\u001b[A\n",
            " 48% 3873/8092 [01:21<02:14, 31.26it/s]\u001b[A\n",
            " 48% 3877/8092 [01:22<02:15, 31.10it/s]\u001b[A\n",
            " 48% 3881/8092 [01:22<02:15, 31.14it/s]\u001b[A\n",
            " 48% 3885/8092 [01:22<02:14, 31.16it/s]\u001b[A\n",
            " 48% 3889/8092 [01:22<02:15, 31.00it/s]\u001b[A\n",
            " 48% 3893/8092 [01:22<02:15, 30.92it/s]\u001b[A\n",
            " 48% 3897/8092 [01:22<02:15, 30.88it/s]\u001b[A\n",
            " 48% 3901/8092 [01:22<02:15, 31.00it/s]\u001b[A\n",
            " 48% 3905/8092 [01:23<02:15, 30.95it/s]\u001b[A\n",
            " 48% 3909/8092 [01:23<02:15, 30.96it/s]\u001b[A\n",
            " 48% 3913/8092 [01:23<02:14, 30.96it/s]\u001b[A\n",
            " 48% 3917/8092 [01:23<02:15, 30.74it/s]\u001b[A\n",
            " 48% 3921/8092 [01:23<02:15, 30.88it/s]\u001b[A\n",
            " 49% 3925/8092 [01:23<02:15, 30.79it/s]\u001b[A\n",
            " 49% 3929/8092 [01:23<02:15, 30.76it/s]\u001b[A\n",
            " 49% 3933/8092 [01:23<02:15, 30.73it/s]\u001b[A\n",
            " 49% 3937/8092 [01:24<02:14, 30.79it/s]\u001b[A\n",
            " 49% 3941/8092 [01:24<02:15, 30.68it/s]\u001b[A\n",
            " 49% 3945/8092 [01:24<02:14, 30.73it/s]\u001b[A\n",
            " 49% 3949/8092 [01:24<02:14, 30.74it/s]\u001b[A\n",
            " 49% 3953/8092 [01:24<02:15, 30.58it/s]\u001b[A\n",
            " 49% 3957/8092 [01:24<02:14, 30.64it/s]\u001b[A\n",
            " 49% 3961/8092 [01:24<02:14, 30.76it/s]\u001b[A\n",
            " 49% 3965/8092 [01:24<02:14, 30.60it/s]\u001b[A\n",
            " 49% 3969/8092 [01:25<02:15, 30.44it/s]\u001b[A\n",
            " 49% 3973/8092 [01:25<02:14, 30.66it/s]\u001b[A\n",
            " 49% 3977/8092 [01:25<02:14, 30.68it/s]\u001b[A\n",
            " 49% 3981/8092 [01:25<02:14, 30.58it/s]\u001b[A\n",
            " 49% 3985/8092 [01:25<02:14, 30.43it/s]\u001b[A\n",
            " 49% 3989/8092 [01:25<02:15, 30.20it/s]\u001b[A\n",
            " 49% 3993/8092 [01:25<02:15, 30.15it/s]\u001b[A\n",
            " 49% 3997/8092 [01:26<02:16, 30.04it/s]\u001b[A\n",
            " 49% 4001/8092 [01:26<02:17, 29.71it/s]\u001b[A\n",
            " 49% 4004/8092 [01:26<02:20, 29.13it/s]\u001b[A\n",
            " 50% 4007/8092 [01:26<02:22, 28.68it/s]\u001b[A\n",
            " 50% 4010/8092 [01:26<02:23, 28.37it/s]\u001b[A\n",
            " 50% 4013/8092 [01:26<02:24, 28.25it/s]\u001b[A\n",
            " 50% 4016/8092 [01:26<02:24, 28.19it/s]\u001b[A\n",
            " 50% 4019/8092 [01:26<02:25, 27.98it/s]\u001b[A\n",
            " 50% 4022/8092 [01:26<02:26, 27.80it/s]\u001b[A\n",
            " 50% 4025/8092 [01:27<02:24, 28.20it/s]\u001b[A\n",
            " 50% 4028/8092 [01:27<02:26, 27.81it/s]\u001b[A\n",
            " 50% 4031/8092 [01:27<02:26, 27.81it/s]\u001b[A\n",
            " 50% 4034/8092 [01:27<02:24, 28.17it/s]\u001b[A\n",
            " 50% 4037/8092 [01:27<02:24, 28.02it/s]\u001b[A\n",
            " 50% 4040/8092 [01:27<02:26, 27.73it/s]\u001b[A\n",
            " 50% 4043/8092 [01:27<02:25, 27.76it/s]\u001b[A\n",
            " 50% 4046/8092 [01:27<02:25, 27.75it/s]\u001b[A\n",
            " 50% 4049/8092 [01:27<02:25, 27.76it/s]\u001b[A\n",
            " 50% 4052/8092 [01:28<02:24, 27.87it/s]\u001b[A\n",
            " 50% 4055/8092 [01:28<02:24, 27.92it/s]\u001b[A\n",
            " 50% 4058/8092 [01:28<02:26, 27.61it/s]\u001b[A\n",
            " 50% 4061/8092 [01:28<02:24, 27.86it/s]\u001b[A\n",
            " 50% 4064/8092 [01:28<02:26, 27.52it/s]\u001b[A\n",
            " 50% 4067/8092 [01:28<02:26, 27.38it/s]\u001b[A\n",
            " 50% 4070/8092 [01:28<02:28, 27.17it/s]\u001b[A\n",
            " 50% 4073/8092 [01:28<02:27, 27.28it/s]\u001b[A\n",
            " 50% 4076/8092 [01:28<02:26, 27.49it/s]\u001b[A\n",
            " 50% 4079/8092 [01:28<02:25, 27.58it/s]\u001b[A\n",
            " 50% 4082/8092 [01:29<02:25, 27.51it/s]\u001b[A\n",
            " 50% 4085/8092 [01:29<02:25, 27.61it/s]\u001b[A\n",
            " 51% 4088/8092 [01:29<02:23, 27.85it/s]\u001b[A\n",
            " 51% 4091/8092 [01:29<02:25, 27.56it/s]\u001b[A\n",
            " 51% 4094/8092 [01:29<02:25, 27.57it/s]\u001b[A\n",
            " 51% 4097/8092 [01:29<02:25, 27.37it/s]\u001b[A\n",
            " 51% 4100/8092 [01:29<02:25, 27.38it/s]\u001b[A\n",
            " 51% 4103/8092 [01:29<02:25, 27.50it/s]\u001b[A\n",
            " 51% 4106/8092 [01:29<02:26, 27.30it/s]\u001b[A\n",
            " 51% 4109/8092 [01:30<02:26, 27.27it/s]\u001b[A\n",
            " 51% 4112/8092 [01:30<02:26, 27.15it/s]\u001b[A\n",
            " 51% 4115/8092 [01:30<02:26, 27.23it/s]\u001b[A\n",
            " 51% 4118/8092 [01:30<02:24, 27.41it/s]\u001b[A\n",
            " 51% 4121/8092 [01:30<02:24, 27.53it/s]\u001b[A\n",
            " 51% 4124/8092 [01:30<02:24, 27.50it/s]\u001b[A\n",
            " 51% 4127/8092 [01:30<02:26, 27.06it/s]\u001b[A\n",
            " 51% 4130/8092 [01:30<02:25, 27.31it/s]\u001b[A\n",
            " 51% 4133/8092 [01:30<02:25, 27.18it/s]\u001b[A\n",
            " 51% 4136/8092 [01:31<02:24, 27.30it/s]\u001b[A\n",
            " 51% 4139/8092 [01:31<02:24, 27.26it/s]\u001b[A\n",
            " 51% 4142/8092 [01:31<02:24, 27.38it/s]\u001b[A\n",
            " 51% 4145/8092 [01:31<02:23, 27.51it/s]\u001b[A\n",
            " 51% 4148/8092 [01:31<02:23, 27.41it/s]\u001b[A\n",
            " 51% 4151/8092 [01:31<02:23, 27.42it/s]\u001b[A\n",
            " 51% 4154/8092 [01:31<02:22, 27.55it/s]\u001b[A\n",
            " 51% 4157/8092 [01:31<02:23, 27.37it/s]\u001b[A\n",
            " 51% 4160/8092 [01:31<02:24, 27.18it/s]\u001b[A\n",
            " 51% 4163/8092 [01:32<02:24, 27.20it/s]\u001b[A\n",
            " 51% 4166/8092 [01:32<02:23, 27.28it/s]\u001b[A\n",
            " 52% 4169/8092 [01:32<02:24, 27.15it/s]\u001b[A\n",
            " 52% 4172/8092 [01:32<02:23, 27.22it/s]\u001b[A\n",
            " 52% 4175/8092 [01:32<02:23, 27.27it/s]\u001b[A\n",
            " 52% 4178/8092 [01:32<02:24, 27.12it/s]\u001b[A\n",
            " 52% 4181/8092 [01:32<02:24, 27.10it/s]\u001b[A\n",
            " 52% 4184/8092 [01:32<02:25, 26.80it/s]\u001b[A\n",
            " 52% 4187/8092 [01:32<02:23, 27.16it/s]\u001b[A\n",
            " 52% 4190/8092 [01:33<02:24, 26.98it/s]\u001b[A\n",
            " 52% 4193/8092 [01:33<02:23, 27.10it/s]\u001b[A\n",
            " 52% 4196/8092 [01:33<02:23, 27.14it/s]\u001b[A\n",
            " 52% 4199/8092 [01:33<02:23, 27.21it/s]\u001b[A\n",
            " 52% 4202/8092 [01:33<02:24, 26.96it/s]\u001b[A\n",
            " 52% 4205/8092 [01:33<02:23, 27.04it/s]\u001b[A\n",
            " 52% 4208/8092 [01:33<02:24, 26.89it/s]\u001b[A\n",
            " 52% 4211/8092 [01:33<02:26, 26.47it/s]\u001b[A\n",
            " 52% 4214/8092 [01:33<02:23, 26.96it/s]\u001b[A\n",
            " 52% 4217/8092 [01:34<02:22, 27.11it/s]\u001b[A\n",
            " 52% 4220/8092 [01:34<02:24, 26.87it/s]\u001b[A\n",
            " 52% 4223/8092 [01:34<02:23, 26.92it/s]\u001b[A\n",
            " 52% 4226/8092 [01:34<02:22, 27.10it/s]\u001b[A\n",
            " 52% 4229/8092 [01:34<02:22, 27.02it/s]\u001b[A\n",
            " 52% 4232/8092 [01:34<02:23, 26.95it/s]\u001b[A\n",
            " 52% 4235/8092 [01:34<02:23, 26.93it/s]\u001b[A\n",
            " 52% 4238/8092 [01:34<02:23, 26.87it/s]\u001b[A\n",
            " 52% 4241/8092 [01:34<02:22, 26.95it/s]\u001b[A\n",
            " 52% 4244/8092 [01:35<02:22, 27.02it/s]\u001b[A\n",
            " 52% 4247/8092 [01:35<02:21, 27.18it/s]\u001b[A\n",
            " 53% 4250/8092 [01:35<02:21, 27.15it/s]\u001b[A\n",
            " 53% 4253/8092 [01:35<02:20, 27.26it/s]\u001b[A\n",
            " 53% 4256/8092 [01:35<02:21, 27.08it/s]\u001b[A\n",
            " 53% 4259/8092 [01:35<02:22, 26.96it/s]\u001b[A\n",
            " 53% 4262/8092 [01:35<02:22, 26.87it/s]\u001b[A\n",
            " 53% 4265/8092 [01:35<02:23, 26.76it/s]\u001b[A\n",
            " 53% 4268/8092 [01:35<02:22, 26.79it/s]\u001b[A\n",
            " 53% 4271/8092 [01:36<02:22, 26.83it/s]\u001b[A\n",
            " 53% 4274/8092 [01:36<02:21, 27.00it/s]\u001b[A\n",
            " 53% 4277/8092 [01:36<02:21, 27.00it/s]\u001b[A\n",
            " 53% 4280/8092 [01:36<02:20, 27.06it/s]\u001b[A\n",
            " 53% 4283/8092 [01:36<02:21, 26.86it/s]\u001b[A\n",
            " 53% 4286/8092 [01:36<02:21, 26.89it/s]\u001b[A\n",
            " 53% 4289/8092 [01:36<02:21, 26.95it/s]\u001b[A\n",
            " 53% 4292/8092 [01:36<02:22, 26.63it/s]\u001b[A\n",
            " 53% 4295/8092 [01:36<02:22, 26.71it/s]\u001b[A\n",
            " 53% 4298/8092 [01:37<02:21, 26.78it/s]\u001b[A\n",
            " 53% 4301/8092 [01:37<02:21, 26.78it/s]\u001b[A\n",
            " 53% 4304/8092 [01:37<02:20, 26.97it/s]\u001b[A\n",
            " 53% 4307/8092 [01:37<02:19, 27.11it/s]\u001b[A\n",
            " 53% 4310/8092 [01:37<02:20, 26.95it/s]\u001b[A\n",
            " 53% 4313/8092 [01:37<02:20, 26.90it/s]\u001b[A\n",
            " 53% 4316/8092 [01:37<02:21, 26.72it/s]\u001b[A\n",
            " 53% 4319/8092 [01:37<02:20, 26.87it/s]\u001b[A\n",
            " 53% 4322/8092 [01:37<02:21, 26.59it/s]\u001b[A\n",
            " 53% 4325/8092 [01:38<02:19, 26.93it/s]\u001b[A\n",
            " 53% 4328/8092 [01:38<02:20, 26.80it/s]\u001b[A\n",
            " 54% 4331/8092 [01:38<02:20, 26.72it/s]\u001b[A\n",
            " 54% 4334/8092 [01:38<02:20, 26.79it/s]\u001b[A\n",
            " 54% 4337/8092 [01:38<02:19, 26.85it/s]\u001b[A\n",
            " 54% 4340/8092 [01:38<02:19, 26.96it/s]\u001b[A\n",
            " 54% 4343/8092 [01:38<02:19, 26.83it/s]\u001b[A\n",
            " 54% 4346/8092 [01:38<02:19, 26.82it/s]\u001b[A\n",
            " 54% 4349/8092 [01:38<02:20, 26.70it/s]\u001b[A\n",
            " 54% 4352/8092 [01:39<02:20, 26.69it/s]\u001b[A\n",
            " 54% 4355/8092 [01:39<02:20, 26.66it/s]\u001b[A\n",
            " 54% 4358/8092 [01:39<02:19, 26.67it/s]\u001b[A\n",
            " 54% 4361/8092 [01:39<02:19, 26.83it/s]\u001b[A\n",
            " 54% 4364/8092 [01:39<02:19, 26.69it/s]\u001b[A\n",
            " 54% 4367/8092 [01:39<02:20, 26.60it/s]\u001b[A\n",
            " 54% 4370/8092 [01:39<02:19, 26.66it/s]\u001b[A\n",
            " 54% 4373/8092 [01:39<02:18, 26.76it/s]\u001b[A\n",
            " 54% 4376/8092 [01:39<02:19, 26.66it/s]\u001b[A\n",
            " 54% 4379/8092 [01:40<02:18, 26.72it/s]\u001b[A\n",
            " 54% 4382/8092 [01:40<02:19, 26.52it/s]\u001b[A\n",
            " 54% 4385/8092 [01:40<02:19, 26.61it/s]\u001b[A\n",
            " 54% 4388/8092 [01:40<02:19, 26.58it/s]\u001b[A\n",
            " 54% 4391/8092 [01:40<02:21, 26.22it/s]\u001b[A\n",
            " 54% 4394/8092 [01:40<02:18, 26.76it/s]\u001b[A\n",
            " 54% 4397/8092 [01:40<02:17, 26.86it/s]\u001b[A\n",
            " 54% 4400/8092 [01:40<02:20, 26.20it/s]\u001b[A\n",
            " 54% 4403/8092 [01:40<02:17, 26.79it/s]\u001b[A\n",
            " 54% 4406/8092 [01:41<02:17, 26.74it/s]\u001b[A\n",
            " 54% 4409/8092 [01:41<02:18, 26.57it/s]\u001b[A\n",
            " 55% 4412/8092 [01:41<02:17, 26.67it/s]\u001b[A\n",
            " 55% 4415/8092 [01:41<02:18, 26.52it/s]\u001b[A\n",
            " 55% 4418/8092 [01:41<02:19, 26.38it/s]\u001b[A\n",
            " 55% 4421/8092 [01:41<02:20, 26.19it/s]\u001b[A\n",
            " 55% 4424/8092 [01:41<02:19, 26.29it/s]\u001b[A\n",
            " 55% 4427/8092 [01:41<02:18, 26.45it/s]\u001b[A\n",
            " 55% 4430/8092 [01:42<02:19, 26.34it/s]\u001b[A\n",
            " 55% 4433/8092 [01:42<02:18, 26.35it/s]\u001b[A\n",
            " 55% 4436/8092 [01:42<02:18, 26.35it/s]\u001b[A\n",
            " 55% 4439/8092 [01:42<02:19, 26.24it/s]\u001b[A\n",
            " 55% 4442/8092 [01:42<02:19, 26.17it/s]\u001b[A\n",
            " 55% 4445/8092 [01:42<02:18, 26.36it/s]\u001b[A\n",
            " 55% 4448/8092 [01:42<02:18, 26.25it/s]\u001b[A\n",
            " 55% 4451/8092 [01:42<02:18, 26.28it/s]\u001b[A\n",
            " 55% 4454/8092 [01:42<02:18, 26.23it/s]\u001b[A\n",
            " 55% 4457/8092 [01:43<02:18, 26.29it/s]\u001b[A\n",
            " 55% 4460/8092 [01:43<02:19, 26.12it/s]\u001b[A\n",
            " 55% 4463/8092 [01:43<02:18, 26.21it/s]\u001b[A\n",
            " 55% 4466/8092 [01:43<02:17, 26.28it/s]\u001b[A\n",
            " 55% 4469/8092 [01:43<02:17, 26.40it/s]\u001b[A\n",
            " 55% 4472/8092 [01:43<02:17, 26.27it/s]\u001b[A\n",
            " 55% 4475/8092 [01:43<02:17, 26.33it/s]\u001b[A\n",
            " 55% 4478/8092 [01:43<02:17, 26.19it/s]\u001b[A\n",
            " 55% 4481/8092 [01:43<02:17, 26.17it/s]\u001b[A\n",
            " 55% 4484/8092 [01:44<02:17, 26.21it/s]\u001b[A\n",
            " 55% 4487/8092 [01:44<02:16, 26.31it/s]\u001b[A\n",
            " 55% 4490/8092 [01:44<02:17, 26.22it/s]\u001b[A\n",
            " 56% 4493/8092 [01:44<02:16, 26.29it/s]\u001b[A\n",
            " 56% 4496/8092 [01:44<02:17, 26.19it/s]\u001b[A\n",
            " 56% 4499/8092 [01:44<02:17, 26.14it/s]\u001b[A\n",
            " 56% 4502/8092 [01:44<02:17, 26.13it/s]\u001b[A\n",
            " 56% 4505/8092 [01:44<02:17, 26.10it/s]\u001b[A\n",
            " 56% 4508/8092 [01:45<02:17, 26.08it/s]\u001b[A\n",
            " 56% 4511/8092 [01:45<02:17, 26.12it/s]\u001b[A\n",
            " 56% 4514/8092 [01:45<02:16, 26.13it/s]\u001b[A\n",
            " 56% 4517/8092 [01:45<02:18, 25.90it/s]\u001b[A\n",
            " 56% 4520/8092 [01:45<02:18, 25.88it/s]\u001b[A\n",
            " 56% 4523/8092 [01:45<02:17, 26.03it/s]\u001b[A\n",
            " 56% 4526/8092 [01:45<02:17, 25.99it/s]\u001b[A\n",
            " 56% 4529/8092 [01:45<02:17, 25.94it/s]\u001b[A\n",
            " 56% 4532/8092 [01:45<02:17, 25.90it/s]\u001b[A\n",
            " 56% 4535/8092 [01:46<02:17, 25.83it/s]\u001b[A\n",
            " 56% 4538/8092 [01:46<02:17, 25.90it/s]\u001b[A\n",
            " 56% 4541/8092 [01:46<02:17, 25.82it/s]\u001b[A\n",
            " 56% 4544/8092 [01:46<02:17, 25.86it/s]\u001b[A\n",
            " 56% 4547/8092 [01:46<02:17, 25.81it/s]\u001b[A\n",
            " 56% 4550/8092 [01:46<02:16, 25.91it/s]\u001b[A\n",
            " 56% 4553/8092 [01:46<02:16, 25.85it/s]\u001b[A\n",
            " 56% 4556/8092 [01:46<02:16, 25.84it/s]\u001b[A\n",
            " 56% 4559/8092 [01:46<02:17, 25.73it/s]\u001b[A\n",
            " 56% 4562/8092 [01:47<02:16, 25.78it/s]\u001b[A\n",
            " 56% 4565/8092 [01:47<02:17, 25.73it/s]\u001b[A\n",
            " 56% 4568/8092 [01:47<02:17, 25.71it/s]\u001b[A\n",
            " 56% 4571/8092 [01:47<02:17, 25.68it/s]\u001b[A\n",
            " 57% 4574/8092 [01:47<02:16, 25.74it/s]\u001b[A\n",
            " 57% 4577/8092 [01:47<02:16, 25.73it/s]\u001b[A\n",
            " 57% 4580/8092 [01:47<02:17, 25.55it/s]\u001b[A\n",
            " 57% 4583/8092 [01:47<02:17, 25.59it/s]\u001b[A\n",
            " 57% 4586/8092 [01:48<02:16, 25.62it/s]\u001b[A\n",
            " 57% 4589/8092 [01:48<02:15, 25.81it/s]\u001b[A\n",
            " 57% 4592/8092 [01:48<02:17, 25.55it/s]\u001b[A\n",
            " 57% 4595/8092 [01:48<02:15, 25.73it/s]\u001b[A\n",
            " 57% 4598/8092 [01:48<02:15, 25.72it/s]\u001b[A\n",
            " 57% 4601/8092 [01:48<02:16, 25.63it/s]\u001b[A\n",
            " 57% 4604/8092 [01:48<02:15, 25.75it/s]\u001b[A\n",
            " 57% 4607/8092 [01:48<02:15, 25.74it/s]\u001b[A\n",
            " 57% 4610/8092 [01:48<02:15, 25.61it/s]\u001b[A\n",
            " 57% 4613/8092 [01:49<02:16, 25.45it/s]\u001b[A\n",
            " 57% 4616/8092 [01:49<02:16, 25.44it/s]\u001b[A\n",
            " 57% 4619/8092 [01:49<02:16, 25.43it/s]\u001b[A\n",
            " 57% 4622/8092 [01:49<02:15, 25.63it/s]\u001b[A\n",
            " 57% 4625/8092 [01:49<02:15, 25.51it/s]\u001b[A\n",
            " 57% 4628/8092 [01:49<02:16, 25.40it/s]\u001b[A\n",
            " 57% 4631/8092 [01:49<02:16, 25.39it/s]\u001b[A\n",
            " 57% 4634/8092 [01:49<02:16, 25.28it/s]\u001b[A\n",
            " 57% 4637/8092 [01:50<02:15, 25.44it/s]\u001b[A\n",
            " 57% 4640/8092 [01:50<02:15, 25.41it/s]\u001b[A\n",
            " 57% 4643/8092 [01:50<02:15, 25.47it/s]\u001b[A\n",
            " 57% 4646/8092 [01:50<02:15, 25.50it/s]\u001b[A\n",
            " 57% 4649/8092 [01:50<02:15, 25.38it/s]\u001b[A\n",
            " 57% 4652/8092 [01:50<02:15, 25.35it/s]\u001b[A\n",
            " 58% 4655/8092 [01:50<02:14, 25.48it/s]\u001b[A\n",
            " 58% 4658/8092 [01:50<02:14, 25.47it/s]\u001b[A\n",
            " 58% 4661/8092 [01:50<02:14, 25.42it/s]\u001b[A\n",
            " 58% 4664/8092 [01:51<02:14, 25.44it/s]\u001b[A\n",
            " 58% 4667/8092 [01:51<02:14, 25.38it/s]\u001b[A\n",
            " 58% 4670/8092 [01:51<02:16, 25.14it/s]\u001b[A\n",
            " 58% 4673/8092 [01:51<02:15, 25.32it/s]\u001b[A\n",
            " 58% 4676/8092 [01:51<02:15, 25.29it/s]\u001b[A\n",
            " 58% 4679/8092 [01:51<02:14, 25.29it/s]\u001b[A\n",
            " 58% 4682/8092 [01:51<02:14, 25.32it/s]\u001b[A\n",
            " 58% 4685/8092 [01:51<02:15, 25.21it/s]\u001b[A\n",
            " 58% 4688/8092 [01:52<02:14, 25.32it/s]\u001b[A\n",
            " 58% 4691/8092 [01:52<02:15, 25.13it/s]\u001b[A\n",
            " 58% 4694/8092 [01:52<02:14, 25.28it/s]\u001b[A\n",
            " 58% 4697/8092 [01:52<02:14, 25.17it/s]\u001b[A\n",
            " 58% 4700/8092 [01:52<02:15, 24.97it/s]\u001b[A\n",
            " 58% 4703/8092 [01:52<02:14, 25.26it/s]\u001b[A\n",
            " 58% 4706/8092 [01:52<02:13, 25.37it/s]\u001b[A\n",
            " 58% 4709/8092 [01:52<02:13, 25.30it/s]\u001b[A\n",
            " 58% 4712/8092 [01:52<02:13, 25.30it/s]\u001b[A\n",
            " 58% 4715/8092 [01:53<02:13, 25.38it/s]\u001b[A\n",
            " 58% 4718/8092 [01:53<02:13, 25.25it/s]\u001b[A\n",
            " 58% 4721/8092 [01:53<02:14, 25.15it/s]\u001b[A\n",
            " 58% 4724/8092 [01:53<02:14, 25.10it/s]\u001b[A\n",
            " 58% 4727/8092 [01:53<02:15, 24.82it/s]\u001b[A\n",
            " 58% 4730/8092 [01:53<02:15, 24.87it/s]\u001b[A\n",
            " 58% 4733/8092 [01:53<02:17, 24.39it/s]\u001b[A\n",
            " 59% 4736/8092 [01:53<02:15, 24.69it/s]\u001b[A\n",
            " 59% 4739/8092 [01:54<02:14, 24.88it/s]\u001b[A\n",
            " 59% 4742/8092 [01:54<02:13, 25.16it/s]\u001b[A\n",
            " 59% 4745/8092 [01:54<02:13, 25.14it/s]\u001b[A\n",
            " 59% 4748/8092 [01:54<02:13, 25.01it/s]\u001b[A\n",
            " 59% 4751/8092 [01:54<02:12, 25.22it/s]\u001b[A\n",
            " 59% 4754/8092 [01:54<02:13, 25.10it/s]\u001b[A\n",
            " 59% 4757/8092 [01:54<02:13, 25.00it/s]\u001b[A\n",
            " 59% 4760/8092 [01:54<02:13, 24.94it/s]\u001b[A\n",
            " 59% 4763/8092 [01:55<02:13, 24.86it/s]\u001b[A\n",
            " 59% 4766/8092 [01:55<02:12, 25.17it/s]\u001b[A\n",
            " 59% 4769/8092 [01:55<02:12, 25.09it/s]\u001b[A\n",
            " 59% 4772/8092 [01:55<02:11, 25.20it/s]\u001b[A\n",
            " 59% 4775/8092 [01:55<02:12, 25.09it/s]\u001b[A\n",
            " 59% 4778/8092 [01:55<02:12, 24.97it/s]\u001b[A\n",
            " 59% 4781/8092 [01:55<02:13, 24.85it/s]\u001b[A\n",
            " 59% 4784/8092 [01:55<02:13, 24.79it/s]\u001b[A\n",
            " 59% 4787/8092 [01:55<02:13, 24.77it/s]\u001b[A\n",
            " 59% 4790/8092 [01:56<02:12, 24.86it/s]\u001b[A\n",
            " 59% 4793/8092 [01:56<02:12, 24.85it/s]\u001b[A\n",
            " 59% 4796/8092 [01:56<02:12, 24.83it/s]\u001b[A\n",
            " 59% 4799/8092 [01:56<02:12, 24.80it/s]\u001b[A\n",
            " 59% 4802/8092 [01:56<02:12, 24.75it/s]\u001b[A\n",
            " 59% 4805/8092 [01:56<02:12, 24.72it/s]\u001b[A\n",
            " 59% 4808/8092 [01:56<02:13, 24.64it/s]\u001b[A\n",
            " 59% 4811/8092 [01:56<02:13, 24.63it/s]\u001b[A\n",
            " 59% 4814/8092 [01:57<02:12, 24.73it/s]\u001b[A\n",
            " 60% 4817/8092 [01:57<02:12, 24.67it/s]\u001b[A\n",
            " 60% 4820/8092 [01:57<02:11, 24.82it/s]\u001b[A\n",
            " 60% 4823/8092 [01:57<02:11, 24.78it/s]\u001b[A\n",
            " 60% 4826/8092 [01:57<02:12, 24.74it/s]\u001b[A\n",
            " 60% 4829/8092 [01:57<02:11, 24.77it/s]\u001b[A\n",
            " 60% 4832/8092 [01:57<02:11, 24.79it/s]\u001b[A\n",
            " 60% 4835/8092 [01:57<02:11, 24.80it/s]\u001b[A\n",
            " 60% 4838/8092 [01:58<02:11, 24.76it/s]\u001b[A\n",
            " 60% 4841/8092 [01:58<02:11, 24.70it/s]\u001b[A\n",
            " 60% 4844/8092 [01:58<02:11, 24.74it/s]\u001b[A\n",
            " 60% 4847/8092 [01:58<02:10, 24.78it/s]\u001b[A\n",
            " 60% 4850/8092 [01:58<02:11, 24.70it/s]\u001b[A\n",
            " 60% 4853/8092 [01:58<02:11, 24.71it/s]\u001b[A\n",
            " 60% 4856/8092 [01:58<02:10, 24.75it/s]\u001b[A\n",
            " 60% 4859/8092 [01:58<02:10, 24.79it/s]\u001b[A\n",
            " 60% 4862/8092 [01:59<02:10, 24.73it/s]\u001b[A\n",
            " 60% 4865/8092 [01:59<02:10, 24.76it/s]\u001b[A\n",
            " 60% 4868/8092 [01:59<02:10, 24.76it/s]\u001b[A\n",
            " 60% 4871/8092 [01:59<02:09, 24.79it/s]\u001b[A\n",
            " 60% 4874/8092 [01:59<02:09, 24.77it/s]\u001b[A\n",
            " 60% 4877/8092 [01:59<02:10, 24.68it/s]\u001b[A\n",
            " 60% 4880/8092 [01:59<02:09, 24.78it/s]\u001b[A\n",
            " 60% 4883/8092 [01:59<02:09, 24.76it/s]\u001b[A\n",
            " 60% 4886/8092 [01:59<02:09, 24.75it/s]\u001b[A\n",
            " 60% 4889/8092 [02:00<02:09, 24.78it/s]\u001b[A\n",
            " 60% 4892/8092 [02:00<02:09, 24.75it/s]\u001b[A\n",
            " 60% 4895/8092 [02:00<02:09, 24.75it/s]\u001b[A\n",
            " 61% 4898/8092 [02:00<02:09, 24.69it/s]\u001b[A\n",
            " 61% 4901/8092 [02:00<02:10, 24.54it/s]\u001b[A\n",
            " 61% 4904/8092 [02:00<02:08, 24.85it/s]\u001b[A\n",
            " 61% 4907/8092 [02:00<02:08, 24.78it/s]\u001b[A\n",
            " 61% 4910/8092 [02:00<02:09, 24.55it/s]\u001b[A\n",
            " 61% 4913/8092 [02:01<02:08, 24.80it/s]\u001b[A\n",
            " 61% 4916/8092 [02:01<02:08, 24.72it/s]\u001b[A\n",
            " 61% 4919/8092 [02:01<02:08, 24.73it/s]\u001b[A\n",
            " 61% 4922/8092 [02:01<02:08, 24.71it/s]\u001b[A\n",
            " 61% 4925/8092 [02:01<02:09, 24.55it/s]\u001b[A\n",
            " 61% 4928/8092 [02:01<02:08, 24.69it/s]\u001b[A\n",
            " 61% 4931/8092 [02:01<02:08, 24.64it/s]\u001b[A\n",
            " 61% 4934/8092 [02:01<02:07, 24.72it/s]\u001b[A\n",
            " 61% 4937/8092 [02:02<02:07, 24.69it/s]\u001b[A\n",
            " 61% 4940/8092 [02:02<02:08, 24.61it/s]\u001b[A\n",
            " 61% 4943/8092 [02:02<02:07, 24.70it/s]\u001b[A\n",
            " 61% 4946/8092 [02:02<02:07, 24.68it/s]\u001b[A\n",
            " 61% 4949/8092 [02:02<02:07, 24.65it/s]\u001b[A\n",
            " 61% 4952/8092 [02:02<02:07, 24.56it/s]\u001b[A\n",
            " 61% 4955/8092 [02:02<02:07, 24.64it/s]\u001b[A\n",
            " 61% 4958/8092 [02:02<02:06, 24.75it/s]\u001b[A\n",
            " 61% 4961/8092 [02:03<02:06, 24.68it/s]\u001b[A\n",
            " 61% 4964/8092 [02:03<02:06, 24.75it/s]\u001b[A\n",
            " 61% 4967/8092 [02:03<02:06, 24.74it/s]\u001b[A\n",
            " 61% 4970/8092 [02:03<02:06, 24.70it/s]\u001b[A\n",
            " 61% 4973/8092 [02:03<02:06, 24.74it/s]\u001b[A\n",
            " 61% 4976/8092 [02:03<02:05, 24.78it/s]\u001b[A\n",
            " 62% 4979/8092 [02:03<02:06, 24.65it/s]\u001b[A\n",
            " 62% 4982/8092 [02:03<02:06, 24.55it/s]\u001b[A\n",
            " 62% 4985/8092 [02:04<02:05, 24.78it/s]\u001b[A\n",
            " 62% 4988/8092 [02:04<02:06, 24.62it/s]\u001b[A\n",
            " 62% 4991/8092 [02:04<02:05, 24.66it/s]\u001b[A\n",
            " 62% 4994/8092 [02:04<02:05, 24.76it/s]\u001b[A\n",
            " 62% 4997/8092 [02:04<02:06, 24.45it/s]\u001b[A\n",
            " 62% 5000/8092 [02:04<02:04, 24.75it/s]\u001b[A\n",
            " 62% 5003/8092 [02:04<02:04, 24.74it/s]\u001b[A\n",
            " 62% 5006/8092 [02:04<02:05, 24.68it/s]\u001b[A\n",
            " 62% 5009/8092 [02:04<02:05, 24.64it/s]\u001b[A\n",
            " 62% 5012/8092 [02:05<02:05, 24.51it/s]\u001b[A\n",
            " 62% 5015/8092 [02:05<02:06, 24.37it/s]\u001b[A\n",
            " 62% 5018/8092 [02:05<02:06, 24.35it/s]\u001b[A\n",
            " 62% 5021/8092 [02:05<02:06, 24.21it/s]\u001b[A\n",
            " 62% 5024/8092 [02:05<02:08, 23.95it/s]\u001b[A\n",
            " 62% 5027/8092 [02:05<02:10, 23.50it/s]\u001b[A\n",
            " 62% 5030/8092 [02:05<02:10, 23.45it/s]\u001b[A\n",
            " 62% 5033/8092 [02:06<02:13, 22.83it/s]\u001b[A\n",
            " 62% 5036/8092 [02:06<02:11, 23.30it/s]\u001b[A\n",
            " 62% 5039/8092 [02:06<02:11, 23.29it/s]\u001b[A\n",
            " 62% 5042/8092 [02:06<02:11, 23.14it/s]\u001b[A\n",
            " 62% 5045/8092 [02:06<02:12, 23.08it/s]\u001b[A\n",
            " 62% 5048/8092 [02:06<02:12, 22.92it/s]\u001b[A\n",
            " 62% 5051/8092 [02:06<02:12, 23.03it/s]\u001b[A\n",
            " 62% 5054/8092 [02:06<02:12, 22.86it/s]\u001b[A\n",
            " 62% 5057/8092 [02:07<02:12, 22.94it/s]\u001b[A\n",
            " 63% 5060/8092 [02:07<02:12, 22.81it/s]\u001b[A\n",
            " 63% 5063/8092 [02:07<02:13, 22.75it/s]\u001b[A\n",
            " 63% 5066/8092 [02:07<02:13, 22.75it/s]\u001b[A\n",
            " 63% 5069/8092 [02:07<02:11, 22.90it/s]\u001b[A\n",
            " 63% 5072/8092 [02:07<02:12, 22.77it/s]\u001b[A\n",
            " 63% 5075/8092 [02:07<02:12, 22.76it/s]\u001b[A\n",
            " 63% 5078/8092 [02:07<02:12, 22.81it/s]\u001b[A\n",
            " 63% 5081/8092 [02:08<02:11, 22.93it/s]\u001b[A\n",
            " 63% 5084/8092 [02:08<02:10, 22.97it/s]\u001b[A\n",
            " 63% 5087/8092 [02:08<02:11, 22.82it/s]\u001b[A\n",
            " 63% 5090/8092 [02:08<02:12, 22.72it/s]\u001b[A\n",
            " 63% 5093/8092 [02:08<02:12, 22.68it/s]\u001b[A\n",
            " 63% 5096/8092 [02:08<02:11, 22.73it/s]\u001b[A\n",
            " 63% 5099/8092 [02:08<02:11, 22.77it/s]\u001b[A\n",
            " 63% 5102/8092 [02:09<02:11, 22.79it/s]\u001b[A\n",
            " 63% 5105/8092 [02:09<02:10, 22.90it/s]\u001b[A\n",
            " 63% 5108/8092 [02:09<02:10, 22.82it/s]\u001b[A\n",
            " 63% 5111/8092 [02:09<02:11, 22.74it/s]\u001b[A\n",
            " 63% 5114/8092 [02:09<02:11, 22.66it/s]\u001b[A\n",
            " 63% 5117/8092 [02:09<02:12, 22.54it/s]\u001b[A\n",
            " 63% 5120/8092 [02:09<02:11, 22.62it/s]\u001b[A\n",
            " 63% 5123/8092 [02:09<02:11, 22.55it/s]\u001b[A\n",
            " 63% 5126/8092 [02:10<02:11, 22.63it/s]\u001b[A\n",
            " 63% 5129/8092 [02:10<02:11, 22.54it/s]\u001b[A\n",
            " 63% 5132/8092 [02:10<02:11, 22.53it/s]\u001b[A\n",
            " 63% 5135/8092 [02:10<02:11, 22.55it/s]\u001b[A\n",
            " 63% 5138/8092 [02:10<02:11, 22.43it/s]\u001b[A\n",
            " 64% 5141/8092 [02:10<02:11, 22.47it/s]\u001b[A\n",
            " 64% 5144/8092 [02:10<02:10, 22.58it/s]\u001b[A\n",
            " 64% 5147/8092 [02:11<02:10, 22.65it/s]\u001b[A\n",
            " 64% 5150/8092 [02:11<02:09, 22.72it/s]\u001b[A\n",
            " 64% 5153/8092 [02:11<02:09, 22.71it/s]\u001b[A\n",
            " 64% 5156/8092 [02:11<02:08, 22.81it/s]\u001b[A\n",
            " 64% 5159/8092 [02:11<02:09, 22.69it/s]\u001b[A\n",
            " 64% 5162/8092 [02:11<02:09, 22.65it/s]\u001b[A\n",
            " 64% 5165/8092 [02:11<02:09, 22.61it/s]\u001b[A\n",
            " 64% 5168/8092 [02:11<02:09, 22.60it/s]\u001b[A\n",
            " 64% 5171/8092 [02:12<02:09, 22.62it/s]\u001b[A\n",
            " 64% 5174/8092 [02:12<02:09, 22.59it/s]\u001b[A\n",
            " 64% 5177/8092 [02:12<02:08, 22.61it/s]\u001b[A\n",
            " 64% 5180/8092 [02:12<02:09, 22.48it/s]\u001b[A\n",
            " 64% 5183/8092 [02:12<02:09, 22.50it/s]\u001b[A\n",
            " 64% 5186/8092 [02:12<02:09, 22.49it/s]\u001b[A\n",
            " 64% 5189/8092 [02:12<02:08, 22.51it/s]\u001b[A\n",
            " 64% 5192/8092 [02:13<02:09, 22.43it/s]\u001b[A\n",
            " 64% 5195/8092 [02:13<02:09, 22.45it/s]\u001b[A\n",
            " 64% 5198/8092 [02:13<02:08, 22.52it/s]\u001b[A\n",
            " 64% 5201/8092 [02:13<02:08, 22.44it/s]\u001b[A\n",
            " 64% 5204/8092 [02:13<02:08, 22.40it/s]\u001b[A\n",
            " 64% 5207/8092 [02:13<02:09, 22.30it/s]\u001b[A\n",
            " 64% 5210/8092 [02:13<02:09, 22.32it/s]\u001b[A\n",
            " 64% 5213/8092 [02:13<02:09, 22.30it/s]\u001b[A\n",
            " 64% 5216/8092 [02:14<02:09, 22.28it/s]\u001b[A\n",
            " 64% 5219/8092 [02:14<02:08, 22.34it/s]\u001b[A\n",
            " 65% 5222/8092 [02:14<02:09, 22.12it/s]\u001b[A\n",
            " 65% 5225/8092 [02:14<02:09, 22.14it/s]\u001b[A\n",
            " 65% 5228/8092 [02:14<02:09, 22.11it/s]\u001b[A\n",
            " 65% 5231/8092 [02:14<02:09, 22.15it/s]\u001b[A\n",
            " 65% 5234/8092 [02:14<02:09, 22.08it/s]\u001b[A\n",
            " 65% 5237/8092 [02:15<02:09, 22.03it/s]\u001b[A\n",
            " 65% 5240/8092 [02:15<02:08, 22.28it/s]\u001b[A\n",
            " 65% 5243/8092 [02:15<02:08, 22.14it/s]\u001b[A\n",
            " 65% 5246/8092 [02:15<02:08, 22.11it/s]\u001b[A\n",
            " 65% 5249/8092 [02:15<02:08, 22.18it/s]\u001b[A\n",
            " 65% 5252/8092 [02:15<02:07, 22.28it/s]\u001b[A\n",
            " 65% 5255/8092 [02:15<02:07, 22.17it/s]\u001b[A\n",
            " 65% 5258/8092 [02:15<02:07, 22.22it/s]\u001b[A\n",
            " 65% 5261/8092 [02:16<02:06, 22.34it/s]\u001b[A\n",
            " 65% 5264/8092 [02:16<02:07, 22.23it/s]\u001b[A\n",
            " 65% 5267/8092 [02:16<02:07, 22.12it/s]\u001b[A\n",
            " 65% 5270/8092 [02:16<02:08, 21.93it/s]\u001b[A\n",
            " 65% 5273/8092 [02:16<02:08, 21.92it/s]\u001b[A\n",
            " 65% 5276/8092 [02:16<02:08, 21.86it/s]\u001b[A\n",
            " 65% 5279/8092 [02:16<02:09, 21.66it/s]\u001b[A\n",
            " 65% 5282/8092 [02:17<02:07, 22.08it/s]\u001b[A\n",
            " 65% 5285/8092 [02:17<02:06, 22.18it/s]\u001b[A\n",
            " 65% 5288/8092 [02:17<02:06, 22.23it/s]\u001b[A\n",
            " 65% 5291/8092 [02:17<02:06, 22.20it/s]\u001b[A\n",
            " 65% 5294/8092 [02:17<02:06, 22.15it/s]\u001b[A\n",
            " 65% 5297/8092 [02:17<02:06, 22.09it/s]\u001b[A\n",
            " 65% 5300/8092 [02:17<02:07, 21.87it/s]\u001b[A\n",
            " 66% 5303/8092 [02:18<02:06, 22.03it/s]\u001b[A\n",
            " 66% 5306/8092 [02:18<02:06, 21.95it/s]\u001b[A\n",
            " 66% 5309/8092 [02:18<02:06, 22.03it/s]\u001b[A\n",
            " 66% 5312/8092 [02:18<02:06, 22.06it/s]\u001b[A\n",
            " 66% 5315/8092 [02:18<02:06, 22.01it/s]\u001b[A\n",
            " 66% 5318/8092 [02:18<02:05, 22.05it/s]\u001b[A\n",
            " 66% 5321/8092 [02:18<02:05, 22.06it/s]\u001b[A\n",
            " 66% 5324/8092 [02:18<02:05, 22.00it/s]\u001b[A\n",
            " 66% 5327/8092 [02:19<02:05, 22.10it/s]\u001b[A\n",
            " 66% 5330/8092 [02:19<02:05, 22.08it/s]\u001b[A\n",
            " 66% 5333/8092 [02:19<02:04, 22.11it/s]\u001b[A\n",
            " 66% 5336/8092 [02:19<02:05, 22.03it/s]\u001b[A\n",
            " 66% 5339/8092 [02:19<02:05, 21.94it/s]\u001b[A\n",
            " 66% 5342/8092 [02:19<02:05, 21.92it/s]\u001b[A\n",
            " 66% 5345/8092 [02:19<02:05, 21.92it/s]\u001b[A\n",
            " 66% 5348/8092 [02:20<02:05, 21.90it/s]\u001b[A\n",
            " 66% 5351/8092 [02:20<02:05, 21.91it/s]\u001b[A\n",
            " 66% 5354/8092 [02:20<02:05, 21.87it/s]\u001b[A\n",
            " 66% 5357/8092 [02:20<02:04, 21.96it/s]\u001b[A\n",
            " 66% 5360/8092 [02:20<02:05, 21.83it/s]\u001b[A\n",
            " 66% 5363/8092 [02:20<02:04, 21.89it/s]\u001b[A\n",
            " 66% 5366/8092 [02:20<02:05, 21.80it/s]\u001b[A\n",
            " 66% 5369/8092 [02:21<02:04, 21.80it/s]\u001b[A\n",
            " 66% 5372/8092 [02:21<02:04, 21.82it/s]\u001b[A\n",
            " 66% 5375/8092 [02:21<02:04, 21.81it/s]\u001b[A\n",
            " 66% 5378/8092 [02:21<02:04, 21.84it/s]\u001b[A\n",
            " 66% 5381/8092 [02:21<02:04, 21.82it/s]\u001b[A\n",
            " 67% 5384/8092 [02:21<02:03, 22.01it/s]\u001b[A\n",
            " 67% 5387/8092 [02:21<02:02, 21.99it/s]\u001b[A\n",
            " 67% 5390/8092 [02:21<02:03, 21.90it/s]\u001b[A\n",
            " 67% 5393/8092 [02:22<02:03, 21.80it/s]\u001b[A\n",
            " 67% 5396/8092 [02:22<02:04, 21.71it/s]\u001b[A\n",
            " 67% 5399/8092 [02:22<02:03, 21.87it/s]\u001b[A\n",
            " 67% 5402/8092 [02:22<02:02, 21.93it/s]\u001b[A\n",
            " 67% 5405/8092 [02:22<02:02, 21.92it/s]\u001b[A\n",
            " 67% 5408/8092 [02:22<02:02, 21.98it/s]\u001b[A\n",
            " 67% 5411/8092 [02:22<02:02, 21.88it/s]\u001b[A\n",
            " 67% 5414/8092 [02:23<02:02, 21.83it/s]\u001b[A\n",
            " 67% 5417/8092 [02:23<02:02, 21.84it/s]\u001b[A\n",
            " 67% 5420/8092 [02:23<02:02, 21.73it/s]\u001b[A\n",
            " 67% 5423/8092 [02:23<02:02, 21.79it/s]\u001b[A\n",
            " 67% 5426/8092 [02:23<02:01, 21.86it/s]\u001b[A\n",
            " 67% 5429/8092 [02:23<02:01, 21.93it/s]\u001b[A\n",
            " 67% 5432/8092 [02:23<02:01, 21.93it/s]\u001b[A\n",
            " 67% 5435/8092 [02:24<02:00, 21.98it/s]\u001b[A\n",
            " 67% 5438/8092 [02:24<02:01, 21.92it/s]\u001b[A\n",
            " 67% 5441/8092 [02:24<02:01, 21.85it/s]\u001b[A\n",
            " 67% 5444/8092 [02:24<02:00, 21.89it/s]\u001b[A\n",
            " 67% 5447/8092 [02:24<02:00, 21.88it/s]\u001b[A\n",
            " 67% 5450/8092 [02:24<02:00, 21.92it/s]\u001b[A\n",
            " 67% 5453/8092 [02:24<02:00, 21.91it/s]\u001b[A\n",
            " 67% 5456/8092 [02:25<02:00, 21.86it/s]\u001b[A\n",
            " 67% 5459/8092 [02:25<01:59, 21.99it/s]\u001b[A\n",
            " 67% 5462/8092 [02:25<02:00, 21.87it/s]\u001b[A\n",
            " 68% 5465/8092 [02:25<02:00, 21.88it/s]\u001b[A\n",
            " 68% 5468/8092 [02:25<01:59, 22.01it/s]\u001b[A\n",
            " 68% 5471/8092 [02:25<01:58, 22.03it/s]\u001b[A\n",
            " 68% 5474/8092 [02:25<01:58, 22.02it/s]\u001b[A\n",
            " 68% 5477/8092 [02:25<01:58, 22.07it/s]\u001b[A\n",
            " 68% 5480/8092 [02:26<01:58, 21.97it/s]\u001b[A\n",
            " 68% 5483/8092 [02:26<01:58, 21.94it/s]\u001b[A\n",
            " 68% 5486/8092 [02:26<01:59, 21.90it/s]\u001b[A\n",
            " 68% 5489/8092 [02:26<01:58, 21.89it/s]\u001b[A\n",
            " 68% 5492/8092 [02:26<01:58, 21.89it/s]\u001b[A\n",
            " 68% 5495/8092 [02:26<01:59, 21.77it/s]\u001b[A\n",
            " 68% 5498/8092 [02:26<01:59, 21.78it/s]\u001b[A\n",
            " 68% 5501/8092 [02:27<01:59, 21.76it/s]\u001b[A\n",
            " 68% 5504/8092 [02:27<01:59, 21.65it/s]\u001b[A\n",
            " 68% 5507/8092 [02:27<01:58, 21.79it/s]\u001b[A\n",
            " 68% 5510/8092 [02:27<01:58, 21.85it/s]\u001b[A\n",
            " 68% 5513/8092 [02:27<01:57, 21.89it/s]\u001b[A\n",
            " 68% 5516/8092 [02:27<01:58, 21.82it/s]\u001b[A\n",
            " 68% 5519/8092 [02:27<01:58, 21.72it/s]\u001b[A\n",
            " 68% 5522/8092 [02:28<01:57, 21.83it/s]\u001b[A\n",
            " 68% 5525/8092 [02:28<01:58, 21.64it/s]\u001b[A\n",
            " 68% 5528/8092 [02:28<01:58, 21.57it/s]\u001b[A\n",
            " 68% 5531/8092 [02:28<01:58, 21.66it/s]\u001b[A\n",
            " 68% 5534/8092 [02:28<01:57, 21.72it/s]\u001b[A\n",
            " 68% 5537/8092 [02:28<01:58, 21.63it/s]\u001b[A\n",
            " 68% 5540/8092 [02:28<01:58, 21.50it/s]\u001b[A\n",
            " 68% 5543/8092 [02:28<01:58, 21.55it/s]\u001b[A\n",
            " 69% 5546/8092 [02:29<01:58, 21.54it/s]\u001b[A\n",
            " 69% 5549/8092 [02:29<01:58, 21.52it/s]\u001b[A\n",
            " 69% 5552/8092 [02:29<01:58, 21.51it/s]\u001b[A\n",
            " 69% 5555/8092 [02:29<01:58, 21.50it/s]\u001b[A\n",
            " 69% 5558/8092 [02:29<01:57, 21.63it/s]\u001b[A\n",
            " 69% 5561/8092 [02:29<01:57, 21.53it/s]\u001b[A\n",
            " 69% 5564/8092 [02:29<01:57, 21.53it/s]\u001b[A\n",
            " 69% 5567/8092 [02:30<01:58, 21.29it/s]\u001b[A\n",
            " 69% 5570/8092 [02:30<01:57, 21.48it/s]\u001b[A\n",
            " 69% 5573/8092 [02:30<01:57, 21.50it/s]\u001b[A\n",
            " 69% 5576/8092 [02:30<01:56, 21.51it/s]\u001b[A\n",
            " 69% 5579/8092 [02:30<01:57, 21.46it/s]\u001b[A\n",
            " 69% 5582/8092 [02:30<01:57, 21.44it/s]\u001b[A\n",
            " 69% 5585/8092 [02:30<01:56, 21.58it/s]\u001b[A\n",
            " 69% 5588/8092 [02:31<01:57, 21.38it/s]\u001b[A\n",
            " 69% 5591/8092 [02:31<01:57, 21.37it/s]\u001b[A\n",
            " 69% 5594/8092 [02:31<01:56, 21.52it/s]\u001b[A\n",
            " 69% 5597/8092 [02:31<01:56, 21.48it/s]\u001b[A\n",
            " 69% 5600/8092 [02:31<01:56, 21.48it/s]\u001b[A\n",
            " 69% 5603/8092 [02:31<01:56, 21.45it/s]\u001b[A\n",
            " 69% 5606/8092 [02:31<01:55, 21.47it/s]\u001b[A\n",
            " 69% 5609/8092 [02:32<01:55, 21.42it/s]\u001b[A\n",
            " 69% 5612/8092 [02:32<01:56, 21.34it/s]\u001b[A\n",
            " 69% 5615/8092 [02:32<01:55, 21.39it/s]\u001b[A\n",
            " 69% 5618/8092 [02:32<01:55, 21.42it/s]\u001b[A\n",
            " 69% 5621/8092 [02:32<01:55, 21.40it/s]\u001b[A\n",
            " 70% 5624/8092 [02:32<01:55, 21.37it/s]\u001b[A\n",
            " 70% 5627/8092 [02:32<01:54, 21.50it/s]\u001b[A\n",
            " 70% 5630/8092 [02:33<01:54, 21.49it/s]\u001b[A\n",
            " 70% 5633/8092 [02:33<01:55, 21.29it/s]\u001b[A\n",
            " 70% 5636/8092 [02:33<01:55, 21.21it/s]\u001b[A\n",
            " 70% 5639/8092 [02:33<01:55, 21.19it/s]\u001b[A\n",
            " 70% 5642/8092 [02:33<01:55, 21.24it/s]\u001b[A\n",
            " 70% 5645/8092 [02:33<01:55, 21.22it/s]\u001b[A\n",
            " 70% 5648/8092 [02:33<01:55, 21.24it/s]\u001b[A\n",
            " 70% 5651/8092 [02:34<01:54, 21.23it/s]\u001b[A\n",
            " 70% 5654/8092 [02:34<01:55, 21.15it/s]\u001b[A\n",
            " 70% 5657/8092 [02:34<01:55, 21.15it/s]\u001b[A\n",
            " 70% 5660/8092 [02:34<01:55, 21.11it/s]\u001b[A\n",
            " 70% 5663/8092 [02:34<01:55, 21.10it/s]\u001b[A\n",
            " 70% 5666/8092 [02:34<01:54, 21.21it/s]\u001b[A\n",
            " 70% 5669/8092 [02:34<01:53, 21.31it/s]\u001b[A\n",
            " 70% 5672/8092 [02:35<01:54, 21.18it/s]\u001b[A\n",
            " 70% 5675/8092 [02:35<01:53, 21.22it/s]\u001b[A\n",
            " 70% 5678/8092 [02:35<01:54, 21.16it/s]\u001b[A\n",
            " 70% 5681/8092 [02:35<01:53, 21.15it/s]\u001b[A\n",
            " 70% 5684/8092 [02:35<01:54, 21.05it/s]\u001b[A\n",
            " 70% 5687/8092 [02:35<01:54, 21.10it/s]\u001b[A\n",
            " 70% 5690/8092 [02:35<01:53, 21.15it/s]\u001b[A\n",
            " 70% 5693/8092 [02:36<01:53, 21.12it/s]\u001b[A\n",
            " 70% 5696/8092 [02:36<01:53, 21.16it/s]\u001b[A\n",
            " 70% 5699/8092 [02:36<01:53, 21.15it/s]\u001b[A\n",
            " 70% 5702/8092 [02:36<01:53, 21.04it/s]\u001b[A\n",
            " 71% 5705/8092 [02:36<01:53, 21.05it/s]\u001b[A\n",
            " 71% 5708/8092 [02:36<01:52, 21.11it/s]\u001b[A\n",
            " 71% 5711/8092 [02:36<01:53, 21.04it/s]\u001b[A\n",
            " 71% 5714/8092 [02:37<01:53, 21.03it/s]\u001b[A\n",
            " 71% 5717/8092 [02:37<01:53, 20.99it/s]\u001b[A\n",
            " 71% 5720/8092 [02:37<01:52, 21.09it/s]\u001b[A\n",
            " 71% 5723/8092 [02:37<01:52, 21.12it/s]\u001b[A\n",
            " 71% 5726/8092 [02:37<01:51, 21.13it/s]\u001b[A\n",
            " 71% 5729/8092 [02:37<01:51, 21.14it/s]\u001b[A\n",
            " 71% 5732/8092 [02:37<01:52, 21.05it/s]\u001b[A\n",
            " 71% 5735/8092 [02:38<01:52, 21.02it/s]\u001b[A\n",
            " 71% 5738/8092 [02:38<01:51, 21.05it/s]\u001b[A\n",
            " 71% 5741/8092 [02:38<01:52, 20.97it/s]\u001b[A\n",
            " 71% 5744/8092 [02:38<01:51, 20.99it/s]\u001b[A\n",
            " 71% 5747/8092 [02:38<01:51, 21.04it/s]\u001b[A\n",
            " 71% 5750/8092 [02:38<01:51, 21.01it/s]\u001b[A\n",
            " 71% 5753/8092 [02:38<01:51, 20.95it/s]\u001b[A\n",
            " 71% 5756/8092 [02:39<01:51, 20.88it/s]\u001b[A\n",
            " 71% 5759/8092 [02:39<01:51, 20.90it/s]\u001b[A\n",
            " 71% 5762/8092 [02:39<01:51, 20.94it/s]\u001b[A\n",
            " 71% 5765/8092 [02:39<01:51, 20.93it/s]\u001b[A\n",
            " 71% 5768/8092 [02:39<01:50, 20.95it/s]\u001b[A\n",
            " 71% 5771/8092 [02:39<01:51, 20.90it/s]\u001b[A\n",
            " 71% 5774/8092 [02:39<01:50, 20.97it/s]\u001b[A\n",
            " 71% 5777/8092 [02:40<01:50, 21.02it/s]\u001b[A\n",
            " 71% 5780/8092 [02:40<01:49, 21.06it/s]\u001b[A\n",
            " 71% 5783/8092 [02:40<01:49, 21.05it/s]\u001b[A\n",
            " 72% 5786/8092 [02:40<01:49, 20.98it/s]\u001b[A\n",
            " 72% 5789/8092 [02:40<01:49, 20.96it/s]\u001b[A\n",
            " 72% 5792/8092 [02:40<01:49, 20.96it/s]\u001b[A\n",
            " 72% 5795/8092 [02:40<01:49, 20.94it/s]\u001b[A\n",
            " 72% 5798/8092 [02:41<01:49, 20.92it/s]\u001b[A\n",
            " 72% 5801/8092 [02:41<01:49, 20.89it/s]\u001b[A\n",
            " 72% 5804/8092 [02:41<01:49, 20.83it/s]\u001b[A\n",
            " 72% 5807/8092 [02:41<01:49, 20.82it/s]\u001b[A\n",
            " 72% 5810/8092 [02:41<01:49, 20.81it/s]\u001b[A\n",
            " 72% 5813/8092 [02:41<01:48, 21.00it/s]\u001b[A\n",
            " 72% 5816/8092 [02:41<01:49, 20.71it/s]\u001b[A\n",
            " 72% 5819/8092 [02:42<01:48, 21.00it/s]\u001b[A\n",
            " 72% 5822/8092 [02:42<01:48, 20.93it/s]\u001b[A\n",
            " 72% 5825/8092 [02:42<01:47, 21.04it/s]\u001b[A\n",
            " 72% 5828/8092 [02:42<01:47, 21.01it/s]\u001b[A\n",
            " 72% 5831/8092 [02:42<01:47, 20.94it/s]\u001b[A\n",
            " 72% 5834/8092 [02:42<01:47, 20.96it/s]\u001b[A\n",
            " 72% 5837/8092 [02:42<01:48, 20.80it/s]\u001b[A\n",
            " 72% 5840/8092 [02:43<01:47, 21.02it/s]\u001b[A\n",
            " 72% 5843/8092 [02:43<01:47, 20.91it/s]\u001b[A\n",
            " 72% 5846/8092 [02:43<01:47, 20.93it/s]\u001b[A\n",
            " 72% 5849/8092 [02:43<01:47, 20.93it/s]\u001b[A\n",
            " 72% 5852/8092 [02:43<01:47, 20.91it/s]\u001b[A\n",
            " 72% 5855/8092 [02:43<01:46, 21.02it/s]\u001b[A\n",
            " 72% 5858/8092 [02:43<01:46, 20.91it/s]\u001b[A\n",
            " 72% 5861/8092 [02:44<01:46, 20.99it/s]\u001b[A\n",
            " 72% 5864/8092 [02:44<01:46, 20.96it/s]\u001b[A\n",
            " 73% 5867/8092 [02:44<01:46, 20.98it/s]\u001b[A\n",
            " 73% 5870/8092 [02:44<01:46, 20.94it/s]\u001b[A\n",
            " 73% 5873/8092 [02:44<01:45, 20.94it/s]\u001b[A\n",
            " 73% 5876/8092 [02:44<01:45, 21.00it/s]\u001b[A\n",
            " 73% 5879/8092 [02:44<01:46, 20.87it/s]\u001b[A\n",
            " 73% 5882/8092 [02:45<01:45, 20.89it/s]\u001b[A\n",
            " 73% 5885/8092 [02:45<01:45, 20.84it/s]\u001b[A\n",
            " 73% 5888/8092 [02:45<01:45, 20.97it/s]\u001b[A\n",
            " 73% 5891/8092 [02:45<01:44, 20.98it/s]\u001b[A\n",
            " 73% 5894/8092 [02:45<01:44, 20.98it/s]\u001b[A\n",
            " 73% 5897/8092 [02:45<01:44, 21.05it/s]\u001b[A\n",
            " 73% 5900/8092 [02:45<01:44, 20.93it/s]\u001b[A\n",
            " 73% 5903/8092 [02:46<01:44, 20.91it/s]\u001b[A\n",
            " 73% 5906/8092 [02:46<01:44, 20.96it/s]\u001b[A\n",
            " 73% 5909/8092 [02:46<01:44, 20.97it/s]\u001b[A\n",
            " 73% 5912/8092 [02:46<01:44, 20.91it/s]\u001b[A\n",
            " 73% 5915/8092 [02:46<01:44, 20.93it/s]\u001b[A\n",
            " 73% 5918/8092 [02:46<01:43, 21.08it/s]\u001b[A\n",
            " 73% 5921/8092 [02:46<01:43, 20.98it/s]\u001b[A\n",
            " 73% 5924/8092 [02:47<01:44, 20.83it/s]\u001b[A\n",
            " 73% 5927/8092 [02:47<01:43, 20.95it/s]\u001b[A\n",
            " 73% 5930/8092 [02:47<01:42, 21.03it/s]\u001b[A\n",
            " 73% 5933/8092 [02:47<01:42, 21.00it/s]\u001b[A\n",
            " 73% 5936/8092 [02:47<01:43, 20.87it/s]\u001b[A\n",
            " 73% 5939/8092 [02:47<01:42, 20.93it/s]\u001b[A\n",
            " 73% 5942/8092 [02:47<01:42, 20.91it/s]\u001b[A\n",
            " 73% 5945/8092 [02:48<01:42, 20.87it/s]\u001b[A\n",
            " 74% 5948/8092 [02:48<01:42, 20.88it/s]\u001b[A\n",
            " 74% 5951/8092 [02:48<01:42, 20.86it/s]\u001b[A\n",
            " 74% 5954/8092 [02:48<01:42, 20.77it/s]\u001b[A\n",
            " 74% 5957/8092 [02:48<01:43, 20.73it/s]\u001b[A\n",
            " 74% 5960/8092 [02:48<01:42, 20.83it/s]\u001b[A\n",
            " 74% 5963/8092 [02:48<01:42, 20.78it/s]\u001b[A\n",
            " 74% 5966/8092 [02:49<01:41, 20.91it/s]\u001b[A\n",
            " 74% 5969/8092 [02:49<01:41, 20.83it/s]\u001b[A\n",
            " 74% 5972/8092 [02:49<01:41, 20.81it/s]\u001b[A\n",
            " 74% 5975/8092 [02:49<01:41, 20.78it/s]\u001b[A\n",
            " 74% 5978/8092 [02:49<01:41, 20.74it/s]\u001b[A\n",
            " 74% 5981/8092 [02:49<01:42, 20.67it/s]\u001b[A\n",
            " 74% 5984/8092 [02:49<01:42, 20.66it/s]\u001b[A\n",
            " 74% 5987/8092 [02:50<01:41, 20.72it/s]\u001b[A\n",
            " 74% 5990/8092 [02:50<01:41, 20.73it/s]\u001b[A\n",
            " 74% 5993/8092 [02:50<01:41, 20.74it/s]\u001b[A\n",
            " 74% 5996/8092 [02:50<01:40, 20.83it/s]\u001b[A\n",
            " 74% 5999/8092 [02:50<01:40, 20.84it/s]\u001b[A\n",
            " 74% 6002/8092 [02:50<01:40, 20.81it/s]\u001b[A\n",
            " 74% 6005/8092 [02:50<01:40, 20.77it/s]\u001b[A\n",
            " 74% 6008/8092 [02:51<01:40, 20.76it/s]\u001b[A\n",
            " 74% 6011/8092 [02:51<01:40, 20.74it/s]\u001b[A\n",
            " 74% 6014/8092 [02:51<01:40, 20.69it/s]\u001b[A\n",
            " 74% 6017/8092 [02:51<01:39, 20.78it/s]\u001b[A\n",
            " 74% 6020/8092 [02:51<01:39, 20.74it/s]\u001b[A\n",
            " 74% 6023/8092 [02:51<01:39, 20.79it/s]\u001b[A\n",
            " 74% 6026/8092 [02:51<01:39, 20.70it/s]\u001b[A\n",
            " 75% 6029/8092 [02:52<01:39, 20.76it/s]\u001b[A\n",
            " 75% 6032/8092 [02:52<01:39, 20.66it/s]\u001b[A\n",
            " 75% 6035/8092 [02:52<01:40, 20.46it/s]\u001b[A\n",
            " 75% 6038/8092 [02:52<01:41, 20.29it/s]\u001b[A\n",
            " 75% 6041/8092 [02:52<01:41, 20.23it/s]\u001b[A\n",
            " 75% 6044/8092 [02:52<01:42, 20.06it/s]\u001b[A\n",
            " 75% 6047/8092 [02:52<01:42, 19.96it/s]\u001b[A\n",
            " 75% 6049/8092 [02:53<01:43, 19.79it/s]\u001b[A\n",
            " 75% 6051/8092 [02:53<01:43, 19.64it/s]\u001b[A\n",
            " 75% 6053/8092 [02:53<01:44, 19.52it/s]\u001b[A\n",
            " 75% 6055/8092 [02:53<01:45, 19.35it/s]\u001b[A\n",
            " 75% 6057/8092 [02:53<01:45, 19.24it/s]\u001b[A\n",
            " 75% 6059/8092 [02:53<01:45, 19.28it/s]\u001b[A\n",
            " 75% 6061/8092 [02:53<01:45, 19.23it/s]\u001b[A\n",
            " 75% 6063/8092 [02:53<01:46, 19.06it/s]\u001b[A\n",
            " 75% 6065/8092 [02:53<01:45, 19.30it/s]\u001b[A\n",
            " 75% 6067/8092 [02:54<01:45, 19.14it/s]\u001b[A\n",
            " 75% 6069/8092 [02:54<01:47, 18.83it/s]\u001b[A\n",
            " 75% 6072/8092 [02:54<01:44, 19.24it/s]\u001b[A\n",
            " 75% 6074/8092 [02:54<01:45, 19.21it/s]\u001b[A\n",
            " 75% 6076/8092 [02:54<01:44, 19.24it/s]\u001b[A\n",
            " 75% 6078/8092 [02:54<01:45, 19.17it/s]\u001b[A\n",
            " 75% 6080/8092 [02:54<01:44, 19.25it/s]\u001b[A\n",
            " 75% 6082/8092 [02:54<01:44, 19.24it/s]\u001b[A\n",
            " 75% 6084/8092 [02:54<01:44, 19.20it/s]\u001b[A\n",
            " 75% 6086/8092 [02:55<01:45, 19.03it/s]\u001b[A\n",
            " 75% 6088/8092 [02:55<01:44, 19.17it/s]\u001b[A\n",
            " 75% 6090/8092 [02:55<01:44, 19.09it/s]\u001b[A\n",
            " 75% 6092/8092 [02:55<01:45, 18.90it/s]\u001b[A\n",
            " 75% 6094/8092 [02:55<01:44, 19.08it/s]\u001b[A\n",
            " 75% 6096/8092 [02:55<01:44, 19.18it/s]\u001b[A\n",
            " 75% 6098/8092 [02:55<01:44, 19.13it/s]\u001b[A\n",
            " 75% 6100/8092 [02:55<01:44, 19.00it/s]\u001b[A\n",
            " 75% 6102/8092 [02:55<01:43, 19.16it/s]\u001b[A\n",
            " 75% 6104/8092 [02:55<01:44, 19.05it/s]\u001b[A\n",
            " 75% 6106/8092 [02:56<01:43, 19.14it/s]\u001b[A\n",
            " 75% 6108/8092 [02:56<01:43, 19.18it/s]\u001b[A\n",
            " 76% 6110/8092 [02:56<01:44, 19.03it/s]\u001b[A\n",
            " 76% 6112/8092 [02:56<01:44, 19.02it/s]\u001b[A\n",
            " 76% 6114/8092 [02:56<01:43, 19.06it/s]\u001b[A\n",
            " 76% 6116/8092 [02:56<01:43, 19.16it/s]\u001b[A\n",
            " 76% 6118/8092 [02:56<01:43, 19.15it/s]\u001b[A\n",
            " 76% 6120/8092 [02:56<01:43, 19.09it/s]\u001b[A\n",
            " 76% 6122/8092 [02:56<01:43, 19.04it/s]\u001b[A\n",
            " 76% 6124/8092 [02:57<01:43, 18.98it/s]\u001b[A\n",
            " 76% 6126/8092 [02:57<01:42, 19.10it/s]\u001b[A\n",
            " 76% 6128/8092 [02:57<01:43, 18.98it/s]\u001b[A\n",
            " 76% 6130/8092 [02:57<01:43, 18.88it/s]\u001b[A\n",
            " 76% 6132/8092 [02:57<01:43, 18.89it/s]\u001b[A\n",
            " 76% 6134/8092 [02:57<01:44, 18.81it/s]\u001b[A\n",
            " 76% 6136/8092 [02:57<01:43, 18.84it/s]\u001b[A\n",
            " 76% 6138/8092 [02:57<01:43, 18.91it/s]\u001b[A\n",
            " 76% 6140/8092 [02:57<01:43, 18.94it/s]\u001b[A\n",
            " 76% 6142/8092 [02:57<01:43, 18.90it/s]\u001b[A\n",
            " 76% 6144/8092 [02:58<01:42, 19.03it/s]\u001b[A\n",
            " 76% 6146/8092 [02:58<01:42, 19.00it/s]\u001b[A\n",
            " 76% 6148/8092 [02:58<01:42, 19.03it/s]\u001b[A\n",
            " 76% 6150/8092 [02:58<01:42, 18.99it/s]\u001b[A\n",
            " 76% 6152/8092 [02:58<01:41, 19.05it/s]\u001b[A\n",
            " 76% 6154/8092 [02:58<01:41, 19.06it/s]\u001b[A\n",
            " 76% 6156/8092 [02:58<01:42, 18.87it/s]\u001b[A\n",
            " 76% 6158/8092 [02:58<01:42, 18.94it/s]\u001b[A\n",
            " 76% 6160/8092 [02:58<01:42, 18.90it/s]\u001b[A\n",
            " 76% 6162/8092 [02:59<01:41, 18.94it/s]\u001b[A\n",
            " 76% 6164/8092 [02:59<01:41, 18.94it/s]\u001b[A\n",
            " 76% 6166/8092 [02:59<01:41, 18.96it/s]\u001b[A\n",
            " 76% 6168/8092 [02:59<01:41, 18.93it/s]\u001b[A\n",
            " 76% 6170/8092 [02:59<01:41, 19.01it/s]\u001b[A\n",
            " 76% 6172/8092 [02:59<01:40, 19.07it/s]\u001b[A\n",
            " 76% 6174/8092 [02:59<01:41, 18.95it/s]\u001b[A\n",
            " 76% 6176/8092 [02:59<01:40, 19.01it/s]\u001b[A\n",
            " 76% 6178/8092 [02:59<01:41, 18.90it/s]\u001b[A\n",
            " 76% 6180/8092 [02:59<01:41, 18.82it/s]\u001b[A\n",
            " 76% 6182/8092 [03:00<01:41, 18.82it/s]\u001b[A\n",
            " 76% 6184/8092 [03:00<01:41, 18.80it/s]\u001b[A\n",
            " 76% 6186/8092 [03:00<01:40, 18.89it/s]\u001b[A\n",
            " 76% 6188/8092 [03:00<01:40, 18.95it/s]\u001b[A\n",
            " 76% 6190/8092 [03:00<01:41, 18.83it/s]\u001b[A\n",
            " 77% 6192/8092 [03:00<01:40, 18.93it/s]\u001b[A\n",
            " 77% 6194/8092 [03:00<01:40, 18.88it/s]\u001b[A\n",
            " 77% 6196/8092 [03:00<01:40, 18.79it/s]\u001b[A\n",
            " 77% 6198/8092 [03:00<01:40, 18.81it/s]\u001b[A\n",
            " 77% 6200/8092 [03:01<01:41, 18.72it/s]\u001b[A\n",
            " 77% 6202/8092 [03:01<01:40, 18.81it/s]\u001b[A\n",
            " 77% 6204/8092 [03:01<01:40, 18.77it/s]\u001b[A\n",
            " 77% 6206/8092 [03:01<01:40, 18.80it/s]\u001b[A\n",
            " 77% 6208/8092 [03:01<01:40, 18.76it/s]\u001b[A\n",
            " 77% 6210/8092 [03:01<01:39, 18.95it/s]\u001b[A\n",
            " 77% 6212/8092 [03:01<01:39, 18.86it/s]\u001b[A\n",
            " 77% 6214/8092 [03:01<01:39, 18.87it/s]\u001b[A\n",
            " 77% 6216/8092 [03:01<01:39, 18.85it/s]\u001b[A\n",
            " 77% 6218/8092 [03:02<01:39, 18.84it/s]\u001b[A\n",
            " 77% 6220/8092 [03:02<01:38, 18.96it/s]\u001b[A\n",
            " 77% 6222/8092 [03:02<01:39, 18.71it/s]\u001b[A\n",
            " 77% 6224/8092 [03:02<01:39, 18.70it/s]\u001b[A\n",
            " 77% 6226/8092 [03:02<01:39, 18.78it/s]\u001b[A\n",
            " 77% 6228/8092 [03:02<01:39, 18.71it/s]\u001b[A\n",
            " 77% 6230/8092 [03:02<01:38, 18.85it/s]\u001b[A\n",
            " 77% 6232/8092 [03:02<01:38, 18.81it/s]\u001b[A\n",
            " 77% 6234/8092 [03:02<01:38, 18.93it/s]\u001b[A\n",
            " 77% 6236/8092 [03:02<01:38, 18.78it/s]\u001b[A\n",
            " 77% 6238/8092 [03:03<01:39, 18.71it/s]\u001b[A\n",
            " 77% 6240/8092 [03:03<01:38, 18.74it/s]\u001b[A\n",
            " 77% 6242/8092 [03:03<01:38, 18.75it/s]\u001b[A\n",
            " 77% 6244/8092 [03:03<01:38, 18.79it/s]\u001b[A\n",
            " 77% 6246/8092 [03:03<01:38, 18.65it/s]\u001b[A\n",
            " 77% 6248/8092 [03:03<01:38, 18.71it/s]\u001b[A\n",
            " 77% 6250/8092 [03:03<01:38, 18.71it/s]\u001b[A\n",
            " 77% 6252/8092 [03:03<01:38, 18.73it/s]\u001b[A\n",
            " 77% 6254/8092 [03:03<01:38, 18.73it/s]\u001b[A\n",
            " 77% 6256/8092 [03:04<01:37, 18.78it/s]\u001b[A\n",
            " 77% 6258/8092 [03:04<01:37, 18.79it/s]\u001b[A\n",
            " 77% 6260/8092 [03:04<01:37, 18.77it/s]\u001b[A\n",
            " 77% 6262/8092 [03:04<01:37, 18.76it/s]\u001b[A\n",
            " 77% 6264/8092 [03:04<01:38, 18.65it/s]\u001b[A\n",
            " 77% 6266/8092 [03:04<01:37, 18.72it/s]\u001b[A\n",
            " 77% 6268/8092 [03:04<01:37, 18.76it/s]\u001b[A\n",
            " 77% 6270/8092 [03:04<01:36, 18.85it/s]\u001b[A\n",
            " 78% 6272/8092 [03:04<01:36, 18.78it/s]\u001b[A\n",
            " 78% 6274/8092 [03:04<01:36, 18.76it/s]\u001b[A\n",
            " 78% 6276/8092 [03:05<01:36, 18.83it/s]\u001b[A\n",
            " 78% 6278/8092 [03:05<01:36, 18.85it/s]\u001b[A\n",
            " 78% 6280/8092 [03:05<01:36, 18.72it/s]\u001b[A\n",
            " 78% 6282/8092 [03:05<01:37, 18.63it/s]\u001b[A\n",
            " 78% 6284/8092 [03:05<01:36, 18.75it/s]\u001b[A\n",
            " 78% 6286/8092 [03:05<01:36, 18.69it/s]\u001b[A\n",
            " 78% 6288/8092 [03:05<01:36, 18.71it/s]\u001b[A\n",
            " 78% 6290/8092 [03:05<01:36, 18.70it/s]\u001b[A\n",
            " 78% 6292/8092 [03:05<01:36, 18.59it/s]\u001b[A\n",
            " 78% 6294/8092 [03:06<01:36, 18.57it/s]\u001b[A\n",
            " 78% 6296/8092 [03:06<01:36, 18.61it/s]\u001b[A\n",
            " 78% 6298/8092 [03:06<01:36, 18.61it/s]\u001b[A\n",
            " 78% 6300/8092 [03:06<01:36, 18.59it/s]\u001b[A\n",
            " 78% 6302/8092 [03:06<01:36, 18.62it/s]\u001b[A\n",
            " 78% 6304/8092 [03:06<01:35, 18.66it/s]\u001b[A\n",
            " 78% 6306/8092 [03:06<01:35, 18.68it/s]\u001b[A\n",
            " 78% 6308/8092 [03:06<01:35, 18.60it/s]\u001b[A\n",
            " 78% 6310/8092 [03:06<01:35, 18.72it/s]\u001b[A\n",
            " 78% 6312/8092 [03:07<01:36, 18.53it/s]\u001b[A\n",
            " 78% 6314/8092 [03:07<01:35, 18.62it/s]\u001b[A\n",
            " 78% 6316/8092 [03:07<01:34, 18.71it/s]\u001b[A\n",
            " 78% 6318/8092 [03:07<01:35, 18.51it/s]\u001b[A\n",
            " 78% 6320/8092 [03:07<01:35, 18.56it/s]\u001b[A\n",
            " 78% 6322/8092 [03:07<01:34, 18.73it/s]\u001b[A\n",
            " 78% 6324/8092 [03:07<01:35, 18.61it/s]\u001b[A\n",
            " 78% 6326/8092 [03:07<01:35, 18.54it/s]\u001b[A\n",
            " 78% 6328/8092 [03:07<01:34, 18.65it/s]\u001b[A\n",
            " 78% 6330/8092 [03:07<01:34, 18.59it/s]\u001b[A\n",
            " 78% 6332/8092 [03:08<01:34, 18.67it/s]\u001b[A\n",
            " 78% 6334/8092 [03:08<01:33, 18.73it/s]\u001b[A\n",
            " 78% 6336/8092 [03:08<01:33, 18.81it/s]\u001b[A\n",
            " 78% 6338/8092 [03:08<01:33, 18.67it/s]\u001b[A\n",
            " 78% 6340/8092 [03:08<01:34, 18.51it/s]\u001b[A\n",
            " 78% 6342/8092 [03:08<01:34, 18.53it/s]\u001b[A\n",
            " 78% 6344/8092 [03:08<01:34, 18.57it/s]\u001b[A\n",
            " 78% 6346/8092 [03:08<01:34, 18.56it/s]\u001b[A\n",
            " 78% 6348/8092 [03:08<01:34, 18.52it/s]\u001b[A\n",
            " 78% 6350/8092 [03:09<01:34, 18.53it/s]\u001b[A\n",
            " 78% 6352/8092 [03:09<01:33, 18.57it/s]\u001b[A\n",
            " 79% 6354/8092 [03:09<01:33, 18.55it/s]\u001b[A\n",
            " 79% 6356/8092 [03:09<01:33, 18.48it/s]\u001b[A\n",
            " 79% 6358/8092 [03:09<01:33, 18.56it/s]\u001b[A\n",
            " 79% 6360/8092 [03:09<01:33, 18.57it/s]\u001b[A\n",
            " 79% 6362/8092 [03:09<01:32, 18.65it/s]\u001b[A\n",
            " 79% 6364/8092 [03:09<01:33, 18.57it/s]\u001b[A\n",
            " 79% 6366/8092 [03:09<01:32, 18.67it/s]\u001b[A\n",
            " 79% 6368/8092 [03:10<01:32, 18.67it/s]\u001b[A\n",
            " 79% 6370/8092 [03:10<01:32, 18.60it/s]\u001b[A\n",
            " 79% 6372/8092 [03:10<01:32, 18.58it/s]\u001b[A\n",
            " 79% 6374/8092 [03:10<01:32, 18.61it/s]\u001b[A\n",
            " 79% 6376/8092 [03:10<01:32, 18.53it/s]\u001b[A\n",
            " 79% 6378/8092 [03:10<01:31, 18.65it/s]\u001b[A\n",
            " 79% 6380/8092 [03:10<01:32, 18.58it/s]\u001b[A\n",
            " 79% 6382/8092 [03:10<01:31, 18.63it/s]\u001b[A\n",
            " 79% 6384/8092 [03:10<01:31, 18.60it/s]\u001b[A\n",
            " 79% 6386/8092 [03:11<01:31, 18.63it/s]\u001b[A\n",
            " 79% 6388/8092 [03:11<01:32, 18.52it/s]\u001b[A\n",
            " 79% 6390/8092 [03:11<01:31, 18.60it/s]\u001b[A\n",
            " 79% 6392/8092 [03:11<01:31, 18.50it/s]\u001b[A\n",
            " 79% 6394/8092 [03:11<01:31, 18.50it/s]\u001b[A\n",
            " 79% 6396/8092 [03:11<01:31, 18.47it/s]\u001b[A\n",
            " 79% 6398/8092 [03:11<01:31, 18.51it/s]\u001b[A\n",
            " 79% 6400/8092 [03:11<01:31, 18.54it/s]\u001b[A\n",
            " 79% 6402/8092 [03:11<01:31, 18.53it/s]\u001b[A\n",
            " 79% 6404/8092 [03:11<01:31, 18.54it/s]\u001b[A\n",
            " 79% 6406/8092 [03:12<01:31, 18.49it/s]\u001b[A\n",
            " 79% 6408/8092 [03:12<01:30, 18.52it/s]\u001b[A\n",
            " 79% 6410/8092 [03:12<01:30, 18.56it/s]\u001b[A\n",
            " 79% 6412/8092 [03:12<01:30, 18.47it/s]\u001b[A\n",
            " 79% 6414/8092 [03:12<01:30, 18.50it/s]\u001b[A\n",
            " 79% 6416/8092 [03:12<01:30, 18.54it/s]\u001b[A\n",
            " 79% 6418/8092 [03:12<01:30, 18.52it/s]\u001b[A\n",
            " 79% 6420/8092 [03:12<01:30, 18.48it/s]\u001b[A\n",
            " 79% 6422/8092 [03:12<01:30, 18.53it/s]\u001b[A\n",
            " 79% 6424/8092 [03:13<01:30, 18.43it/s]\u001b[A\n",
            " 79% 6426/8092 [03:13<01:30, 18.48it/s]\u001b[A\n",
            " 79% 6428/8092 [03:13<01:30, 18.45it/s]\u001b[A\n",
            " 79% 6430/8092 [03:13<01:30, 18.46it/s]\u001b[A\n",
            " 79% 6432/8092 [03:13<01:30, 18.40it/s]\u001b[A\n",
            " 80% 6434/8092 [03:13<01:30, 18.38it/s]\u001b[A\n",
            " 80% 6436/8092 [03:13<01:29, 18.48it/s]\u001b[A\n",
            " 80% 6438/8092 [03:13<01:29, 18.47it/s]\u001b[A\n",
            " 80% 6440/8092 [03:13<01:29, 18.47it/s]\u001b[A\n",
            " 80% 6442/8092 [03:14<01:29, 18.40it/s]\u001b[A\n",
            " 80% 6444/8092 [03:14<01:28, 18.54it/s]\u001b[A\n",
            " 80% 6446/8092 [03:14<01:29, 18.42it/s]\u001b[A\n",
            " 80% 6448/8092 [03:14<01:29, 18.39it/s]\u001b[A\n",
            " 80% 6450/8092 [03:14<01:29, 18.42it/s]\u001b[A\n",
            " 80% 6452/8092 [03:14<01:28, 18.49it/s]\u001b[A\n",
            " 80% 6454/8092 [03:14<01:28, 18.51it/s]\u001b[A\n",
            " 80% 6456/8092 [03:14<01:28, 18.59it/s]\u001b[A\n",
            " 80% 6458/8092 [03:14<01:27, 18.62it/s]\u001b[A\n",
            " 80% 6460/8092 [03:15<01:27, 18.58it/s]\u001b[A\n",
            " 80% 6462/8092 [03:15<01:27, 18.60it/s]\u001b[A\n",
            " 80% 6464/8092 [03:15<01:27, 18.53it/s]\u001b[A\n",
            " 80% 6466/8092 [03:15<01:27, 18.50it/s]\u001b[A\n",
            " 80% 6468/8092 [03:15<01:28, 18.44it/s]\u001b[A\n",
            " 80% 6470/8092 [03:15<01:28, 18.43it/s]\u001b[A\n",
            " 80% 6472/8092 [03:15<01:27, 18.45it/s]\u001b[A\n",
            " 80% 6474/8092 [03:15<01:27, 18.46it/s]\u001b[A\n",
            " 80% 6476/8092 [03:15<01:27, 18.50it/s]\u001b[A\n",
            " 80% 6478/8092 [03:15<01:27, 18.43it/s]\u001b[A\n",
            " 80% 6480/8092 [03:16<01:26, 18.55it/s]\u001b[A\n",
            " 80% 6482/8092 [03:16<01:26, 18.52it/s]\u001b[A\n",
            " 80% 6484/8092 [03:16<01:26, 18.49it/s]\u001b[A\n",
            " 80% 6486/8092 [03:16<01:26, 18.51it/s]\u001b[A\n",
            " 80% 6488/8092 [03:16<01:27, 18.42it/s]\u001b[A\n",
            " 80% 6490/8092 [03:16<01:26, 18.42it/s]\u001b[A\n",
            " 80% 6492/8092 [03:16<01:26, 18.50it/s]\u001b[A\n",
            " 80% 6494/8092 [03:16<01:26, 18.55it/s]\u001b[A\n",
            " 80% 6496/8092 [03:16<01:25, 18.56it/s]\u001b[A\n",
            " 80% 6498/8092 [03:17<01:26, 18.49it/s]\u001b[A\n",
            " 80% 6500/8092 [03:17<01:26, 18.43it/s]\u001b[A\n",
            " 80% 6502/8092 [03:17<01:26, 18.44it/s]\u001b[A\n",
            " 80% 6504/8092 [03:17<01:26, 18.38it/s]\u001b[A\n",
            " 80% 6506/8092 [03:17<01:26, 18.39it/s]\u001b[A\n",
            " 80% 6508/8092 [03:17<01:26, 18.41it/s]\u001b[A\n",
            " 80% 6510/8092 [03:17<01:25, 18.42it/s]\u001b[A\n",
            " 80% 6512/8092 [03:17<01:25, 18.41it/s]\u001b[A\n",
            " 80% 6514/8092 [03:17<01:25, 18.53it/s]\u001b[A\n",
            " 81% 6516/8092 [03:18<01:25, 18.50it/s]\u001b[A\n",
            " 81% 6518/8092 [03:18<01:25, 18.40it/s]\u001b[A\n",
            " 81% 6520/8092 [03:18<01:25, 18.38it/s]\u001b[A\n",
            " 81% 6522/8092 [03:18<01:25, 18.31it/s]\u001b[A\n",
            " 81% 6524/8092 [03:18<01:26, 18.21it/s]\u001b[A\n",
            " 81% 6526/8092 [03:18<01:25, 18.40it/s]\u001b[A\n",
            " 81% 6528/8092 [03:18<01:25, 18.35it/s]\u001b[A\n",
            " 81% 6530/8092 [03:18<01:25, 18.35it/s]\u001b[A\n",
            " 81% 6532/8092 [03:18<01:24, 18.50it/s]\u001b[A\n",
            " 81% 6534/8092 [03:19<01:24, 18.48it/s]\u001b[A\n",
            " 81% 6536/8092 [03:19<01:24, 18.44it/s]\u001b[A\n",
            " 81% 6538/8092 [03:19<01:24, 18.44it/s]\u001b[A\n",
            " 81% 6540/8092 [03:19<01:24, 18.39it/s]\u001b[A\n",
            " 81% 6542/8092 [03:19<01:24, 18.44it/s]\u001b[A\n",
            " 81% 6544/8092 [03:19<01:24, 18.42it/s]\u001b[A\n",
            " 81% 6546/8092 [03:19<01:24, 18.40it/s]\u001b[A\n",
            " 81% 6548/8092 [03:19<01:23, 18.41it/s]\u001b[A\n",
            " 81% 6550/8092 [03:19<01:24, 18.35it/s]\u001b[A\n",
            " 81% 6552/8092 [03:19<01:23, 18.42it/s]\u001b[A\n",
            " 81% 6554/8092 [03:20<01:24, 18.22it/s]\u001b[A\n",
            " 81% 6556/8092 [03:20<01:23, 18.45it/s]\u001b[A\n",
            " 81% 6558/8092 [03:20<01:23, 18.35it/s]\u001b[A\n",
            " 81% 6560/8092 [03:20<01:23, 18.35it/s]\u001b[A\n",
            " 81% 6562/8092 [03:20<01:23, 18.40it/s]\u001b[A\n",
            " 81% 6564/8092 [03:20<01:23, 18.36it/s]\u001b[A\n",
            " 81% 6566/8092 [03:20<01:23, 18.35it/s]\u001b[A\n",
            " 81% 6568/8092 [03:20<01:23, 18.30it/s]\u001b[A\n",
            " 81% 6570/8092 [03:20<01:23, 18.31it/s]\u001b[A\n",
            " 81% 6572/8092 [03:21<01:22, 18.34it/s]\u001b[A\n",
            " 81% 6574/8092 [03:21<01:22, 18.31it/s]\u001b[A\n",
            " 81% 6576/8092 [03:21<01:22, 18.29it/s]\u001b[A\n",
            " 81% 6578/8092 [03:21<01:23, 18.18it/s]\u001b[A\n",
            " 81% 6580/8092 [03:21<01:22, 18.24it/s]\u001b[A\n",
            " 81% 6582/8092 [03:21<01:22, 18.27it/s]\u001b[A\n",
            " 81% 6584/8092 [03:21<01:22, 18.31it/s]\u001b[A\n",
            " 81% 6586/8092 [03:21<01:22, 18.36it/s]\u001b[A\n",
            " 81% 6588/8092 [03:21<01:22, 18.27it/s]\u001b[A\n",
            " 81% 6590/8092 [03:22<01:22, 18.26it/s]\u001b[A\n",
            " 81% 6592/8092 [03:22<01:22, 18.19it/s]\u001b[A\n",
            " 81% 6594/8092 [03:22<01:22, 18.25it/s]\u001b[A\n",
            " 82% 6596/8092 [03:22<01:22, 18.18it/s]\u001b[A\n",
            " 82% 6598/8092 [03:22<01:22, 18.12it/s]\u001b[A\n",
            " 82% 6600/8092 [03:22<01:22, 18.15it/s]\u001b[A\n",
            " 82% 6602/8092 [03:22<01:22, 18.16it/s]\u001b[A\n",
            " 82% 6604/8092 [03:22<01:21, 18.17it/s]\u001b[A\n",
            " 82% 6606/8092 [03:22<01:21, 18.20it/s]\u001b[A\n",
            " 82% 6608/8092 [03:23<01:21, 18.14it/s]\u001b[A\n",
            " 82% 6610/8092 [03:23<01:21, 18.17it/s]\u001b[A\n",
            " 82% 6612/8092 [03:23<01:21, 18.18it/s]\u001b[A\n",
            " 82% 6614/8092 [03:23<01:21, 18.23it/s]\u001b[A\n",
            " 82% 6616/8092 [03:23<01:21, 18.20it/s]\u001b[A\n",
            " 82% 6618/8092 [03:23<01:21, 18.20it/s]\u001b[A\n",
            " 82% 6620/8092 [03:23<01:21, 18.09it/s]\u001b[A\n",
            " 82% 6622/8092 [03:23<01:20, 18.18it/s]\u001b[A\n",
            " 82% 6624/8092 [03:23<01:20, 18.21it/s]\u001b[A\n",
            " 82% 6626/8092 [03:24<01:20, 18.21it/s]\u001b[A\n",
            " 82% 6628/8092 [03:24<01:20, 18.19it/s]\u001b[A\n",
            " 82% 6630/8092 [03:24<01:20, 18.25it/s]\u001b[A\n",
            " 82% 6632/8092 [03:24<01:19, 18.30it/s]\u001b[A\n",
            " 82% 6634/8092 [03:24<01:20, 18.19it/s]\u001b[A\n",
            " 82% 6636/8092 [03:24<01:19, 18.22it/s]\u001b[A\n",
            " 82% 6638/8092 [03:24<01:19, 18.18it/s]\u001b[A\n",
            " 82% 6640/8092 [03:24<01:20, 18.14it/s]\u001b[A\n",
            " 82% 6642/8092 [03:24<01:19, 18.20it/s]\u001b[A\n",
            " 82% 6644/8092 [03:25<01:19, 18.20it/s]\u001b[A\n",
            " 82% 6646/8092 [03:25<01:19, 18.14it/s]\u001b[A\n",
            " 82% 6648/8092 [03:25<01:19, 18.23it/s]\u001b[A\n",
            " 82% 6650/8092 [03:25<01:18, 18.30it/s]\u001b[A\n",
            " 82% 6652/8092 [03:25<01:19, 18.16it/s]\u001b[A\n",
            " 82% 6654/8092 [03:25<01:19, 18.13it/s]\u001b[A\n",
            " 82% 6656/8092 [03:25<01:19, 18.12it/s]\u001b[A\n",
            " 82% 6658/8092 [03:25<01:18, 18.20it/s]\u001b[A\n",
            " 82% 6660/8092 [03:25<01:18, 18.21it/s]\u001b[A\n",
            " 82% 6662/8092 [03:26<01:18, 18.11it/s]\u001b[A\n",
            " 82% 6664/8092 [03:26<01:18, 18.22it/s]\u001b[A\n",
            " 82% 6666/8092 [03:26<01:18, 18.26it/s]\u001b[A\n",
            " 82% 6668/8092 [03:26<01:18, 18.21it/s]\u001b[A\n",
            " 82% 6670/8092 [03:26<01:18, 18.14it/s]\u001b[A\n",
            " 82% 6672/8092 [03:26<01:18, 18.19it/s]\u001b[A\n",
            " 82% 6674/8092 [03:26<01:18, 18.16it/s]\u001b[A\n",
            " 83% 6676/8092 [03:26<01:17, 18.22it/s]\u001b[A\n",
            " 83% 6678/8092 [03:26<01:17, 18.25it/s]\u001b[A\n",
            " 83% 6680/8092 [03:27<01:17, 18.25it/s]\u001b[A\n",
            " 83% 6682/8092 [03:27<01:17, 18.31it/s]\u001b[A\n",
            " 83% 6684/8092 [03:27<01:17, 18.17it/s]\u001b[A\n",
            " 83% 6686/8092 [03:27<01:17, 18.26it/s]\u001b[A\n",
            " 83% 6688/8092 [03:27<01:17, 18.19it/s]\u001b[A\n",
            " 83% 6690/8092 [03:27<01:16, 18.24it/s]\u001b[A\n",
            " 83% 6692/8092 [03:27<01:17, 18.16it/s]\u001b[A\n",
            " 83% 6694/8092 [03:27<01:16, 18.25it/s]\u001b[A\n",
            " 83% 6696/8092 [03:27<01:16, 18.23it/s]\u001b[A\n",
            " 83% 6698/8092 [03:28<01:16, 18.21it/s]\u001b[A\n",
            " 83% 6700/8092 [03:28<01:16, 18.26it/s]\u001b[A\n",
            " 83% 6702/8092 [03:28<01:16, 18.18it/s]\u001b[A\n",
            " 83% 6704/8092 [03:28<01:16, 18.17it/s]\u001b[A\n",
            " 83% 6706/8092 [03:28<01:16, 18.18it/s]\u001b[A\n",
            " 83% 6708/8092 [03:28<01:16, 18.10it/s]\u001b[A\n",
            " 83% 6710/8092 [03:28<01:16, 18.06it/s]\u001b[A\n",
            " 83% 6712/8092 [03:28<01:16, 18.11it/s]\u001b[A\n",
            " 83% 6714/8092 [03:28<01:15, 18.16it/s]\u001b[A\n",
            " 83% 6716/8092 [03:29<01:15, 18.24it/s]\u001b[A\n",
            " 83% 6718/8092 [03:29<01:15, 18.30it/s]\u001b[A\n",
            " 83% 6720/8092 [03:29<01:15, 18.17it/s]\u001b[A\n",
            " 83% 6722/8092 [03:29<01:15, 18.20it/s]\u001b[A\n",
            " 83% 6724/8092 [03:29<01:15, 18.14it/s]\u001b[A\n",
            " 83% 6726/8092 [03:29<01:15, 18.20it/s]\u001b[A\n",
            " 83% 6728/8092 [03:29<01:15, 18.14it/s]\u001b[A\n",
            " 83% 6730/8092 [03:29<01:14, 18.19it/s]\u001b[A\n",
            " 83% 6732/8092 [03:29<01:15, 18.10it/s]\u001b[A\n",
            " 83% 6734/8092 [03:29<01:14, 18.11it/s]\u001b[A\n",
            " 83% 6736/8092 [03:30<01:14, 18.14it/s]\u001b[A\n",
            " 83% 6738/8092 [03:30<01:14, 18.12it/s]\u001b[A\n",
            " 83% 6740/8092 [03:30<01:14, 18.16it/s]\u001b[A\n",
            " 83% 6742/8092 [03:30<01:14, 18.07it/s]\u001b[A\n",
            " 83% 6744/8092 [03:30<01:14, 17.99it/s]\u001b[A\n",
            " 83% 6746/8092 [03:30<01:15, 17.85it/s]\u001b[A\n",
            " 83% 6748/8092 [03:30<01:14, 18.05it/s]\u001b[A\n",
            " 83% 6750/8092 [03:30<01:14, 17.97it/s]\u001b[A\n",
            " 83% 6752/8092 [03:30<01:14, 17.92it/s]\u001b[A\n",
            " 83% 6754/8092 [03:31<01:14, 18.02it/s]\u001b[A\n",
            " 83% 6756/8092 [03:31<01:14, 18.05it/s]\u001b[A\n",
            " 84% 6758/8092 [03:31<01:13, 18.05it/s]\u001b[A\n",
            " 84% 6760/8092 [03:31<01:13, 18.09it/s]\u001b[A\n",
            " 84% 6762/8092 [03:31<01:14, 17.96it/s]\u001b[A\n",
            " 84% 6764/8092 [03:31<01:13, 18.04it/s]\u001b[A\n",
            " 84% 6766/8092 [03:31<01:13, 18.04it/s]\u001b[A\n",
            " 84% 6768/8092 [03:31<01:13, 18.14it/s]\u001b[A\n",
            " 84% 6770/8092 [03:31<01:13, 18.08it/s]\u001b[A\n",
            " 84% 6772/8092 [03:32<01:13, 18.07it/s]\u001b[A\n",
            " 84% 6774/8092 [03:32<01:12, 18.12it/s]\u001b[A\n",
            " 84% 6776/8092 [03:32<01:13, 18.01it/s]\u001b[A\n",
            " 84% 6778/8092 [03:32<01:12, 18.01it/s]\u001b[A\n",
            " 84% 6780/8092 [03:32<01:12, 18.03it/s]\u001b[A\n",
            " 84% 6782/8092 [03:32<01:12, 18.06it/s]\u001b[A\n",
            " 84% 6784/8092 [03:32<01:12, 18.03it/s]\u001b[A\n",
            " 84% 6786/8092 [03:32<01:12, 18.01it/s]\u001b[A\n",
            " 84% 6788/8092 [03:32<01:12, 18.02it/s]\u001b[A\n",
            " 84% 6790/8092 [03:33<01:12, 17.99it/s]\u001b[A\n",
            " 84% 6792/8092 [03:33<01:11, 18.14it/s]\u001b[A\n",
            " 84% 6794/8092 [03:33<01:12, 17.96it/s]\u001b[A\n",
            " 84% 6796/8092 [03:33<01:11, 18.06it/s]\u001b[A\n",
            " 84% 6798/8092 [03:33<01:11, 18.06it/s]\u001b[A\n",
            " 84% 6800/8092 [03:33<01:11, 18.15it/s]\u001b[A\n",
            " 84% 6802/8092 [03:33<01:11, 18.12it/s]\u001b[A\n",
            " 84% 6804/8092 [03:33<01:11, 18.05it/s]\u001b[A\n",
            " 84% 6806/8092 [03:33<01:11, 18.07it/s]\u001b[A\n",
            " 84% 6808/8092 [03:34<01:11, 18.02it/s]\u001b[A\n",
            " 84% 6810/8092 [03:34<01:11, 18.04it/s]\u001b[A\n",
            " 84% 6812/8092 [03:34<01:11, 18.01it/s]\u001b[A\n",
            " 84% 6814/8092 [03:34<01:10, 18.02it/s]\u001b[A\n",
            " 84% 6816/8092 [03:34<01:11, 17.95it/s]\u001b[A\n",
            " 84% 6818/8092 [03:34<01:10, 18.01it/s]\u001b[A\n",
            " 84% 6820/8092 [03:34<01:10, 17.96it/s]\u001b[A\n",
            " 84% 6822/8092 [03:34<01:10, 18.07it/s]\u001b[A\n",
            " 84% 6824/8092 [03:34<01:10, 18.02it/s]\u001b[A\n",
            " 84% 6826/8092 [03:35<01:10, 18.03it/s]\u001b[A\n",
            " 84% 6828/8092 [03:35<01:10, 17.99it/s]\u001b[A\n",
            " 84% 6830/8092 [03:35<01:10, 17.99it/s]\u001b[A\n",
            " 84% 6832/8092 [03:35<01:09, 18.06it/s]\u001b[A\n",
            " 84% 6834/8092 [03:35<01:09, 18.10it/s]\u001b[A\n",
            " 84% 6836/8092 [03:35<01:09, 18.11it/s]\u001b[A\n",
            " 85% 6838/8092 [03:35<01:09, 18.02it/s]\u001b[A\n",
            " 85% 6840/8092 [03:35<01:09, 17.99it/s]\u001b[A\n",
            " 85% 6842/8092 [03:35<01:09, 18.04it/s]\u001b[A\n",
            " 85% 6844/8092 [03:36<01:08, 18.10it/s]\u001b[A\n",
            " 85% 6846/8092 [03:36<01:09, 17.99it/s]\u001b[A\n",
            " 85% 6848/8092 [03:36<01:08, 18.05it/s]\u001b[A\n",
            " 85% 6850/8092 [03:36<01:09, 17.93it/s]\u001b[A\n",
            " 85% 6852/8092 [03:36<01:08, 18.01it/s]\u001b[A\n",
            " 85% 6854/8092 [03:36<01:08, 17.95it/s]\u001b[A\n",
            " 85% 6856/8092 [03:36<01:08, 17.96it/s]\u001b[A\n",
            " 85% 6858/8092 [03:36<01:08, 17.90it/s]\u001b[A\n",
            " 85% 6860/8092 [03:36<01:08, 17.94it/s]\u001b[A\n",
            " 85% 6862/8092 [03:37<01:08, 17.96it/s]\u001b[A\n",
            " 85% 6864/8092 [03:37<01:08, 18.01it/s]\u001b[A\n",
            " 85% 6866/8092 [03:37<01:08, 17.95it/s]\u001b[A\n",
            " 85% 6868/8092 [03:37<01:08, 17.87it/s]\u001b[A\n",
            " 85% 6870/8092 [03:37<01:08, 17.84it/s]\u001b[A\n",
            " 85% 6872/8092 [03:37<01:08, 17.75it/s]\u001b[A\n",
            " 85% 6874/8092 [03:37<01:08, 17.78it/s]\u001b[A\n",
            " 85% 6876/8092 [03:37<01:08, 17.70it/s]\u001b[A\n",
            " 85% 6878/8092 [03:37<01:08, 17.64it/s]\u001b[A\n",
            " 85% 6880/8092 [03:38<01:08, 17.76it/s]\u001b[A\n",
            " 85% 6882/8092 [03:38<01:08, 17.73it/s]\u001b[A\n",
            " 85% 6884/8092 [03:38<01:08, 17.72it/s]\u001b[A\n",
            " 85% 6886/8092 [03:38<01:07, 17.74it/s]\u001b[A\n",
            " 85% 6888/8092 [03:38<01:07, 17.79it/s]\u001b[A\n",
            " 85% 6890/8092 [03:38<01:07, 17.77it/s]\u001b[A\n",
            " 85% 6892/8092 [03:38<01:07, 17.81it/s]\u001b[A\n",
            " 85% 6894/8092 [03:38<01:07, 17.71it/s]\u001b[A\n",
            " 85% 6896/8092 [03:39<01:07, 17.73it/s]\u001b[A\n",
            " 85% 6898/8092 [03:39<01:07, 17.78it/s]\u001b[A\n",
            " 85% 6900/8092 [03:39<01:07, 17.65it/s]\u001b[A\n",
            " 85% 6902/8092 [03:39<01:07, 17.72it/s]\u001b[A\n",
            " 85% 6904/8092 [03:39<01:07, 17.70it/s]\u001b[A\n",
            " 85% 6906/8092 [03:39<01:06, 17.70it/s]\u001b[A\n",
            " 85% 6908/8092 [03:39<01:06, 17.72it/s]\u001b[A\n",
            " 85% 6910/8092 [03:39<01:06, 17.75it/s]\u001b[A\n",
            " 85% 6912/8092 [03:39<01:06, 17.71it/s]\u001b[A\n",
            " 85% 6914/8092 [03:40<01:06, 17.67it/s]\u001b[A\n",
            " 85% 6916/8092 [03:40<01:06, 17.67it/s]\u001b[A\n",
            " 85% 6918/8092 [03:40<01:06, 17.69it/s]\u001b[A\n",
            " 86% 6920/8092 [03:40<01:06, 17.63it/s]\u001b[A\n",
            " 86% 6922/8092 [03:40<01:06, 17.65it/s]\u001b[A\n",
            " 86% 6924/8092 [03:40<01:06, 17.65it/s]\u001b[A\n",
            " 86% 6926/8092 [03:40<01:06, 17.57it/s]\u001b[A\n",
            " 86% 6928/8092 [03:40<01:05, 17.69it/s]\u001b[A\n",
            " 86% 6930/8092 [03:40<01:05, 17.64it/s]\u001b[A\n",
            " 86% 6932/8092 [03:41<01:05, 17.58it/s]\u001b[A\n",
            " 86% 6934/8092 [03:41<01:06, 17.54it/s]\u001b[A\n",
            " 86% 6936/8092 [03:41<01:05, 17.61it/s]\u001b[A\n",
            " 86% 6938/8092 [03:41<01:05, 17.66it/s]\u001b[A\n",
            " 86% 6940/8092 [03:41<01:05, 17.62it/s]\u001b[A\n",
            " 86% 6942/8092 [03:41<01:05, 17.56it/s]\u001b[A\n",
            " 86% 6944/8092 [03:41<01:05, 17.65it/s]\u001b[A\n",
            " 86% 6946/8092 [03:41<01:04, 17.67it/s]\u001b[A\n",
            " 86% 6948/8092 [03:41<01:04, 17.63it/s]\u001b[A\n",
            " 86% 6950/8092 [03:42<01:04, 17.64it/s]\u001b[A\n",
            " 86% 6952/8092 [03:42<01:04, 17.58it/s]\u001b[A\n",
            " 86% 6954/8092 [03:42<01:04, 17.58it/s]\u001b[A\n",
            " 86% 6956/8092 [03:42<01:04, 17.57it/s]\u001b[A\n",
            " 86% 6958/8092 [03:42<01:04, 17.52it/s]\u001b[A\n",
            " 86% 6960/8092 [03:42<01:03, 17.78it/s]\u001b[A\n",
            " 86% 6962/8092 [03:42<01:04, 17.59it/s]\u001b[A\n",
            " 86% 6964/8092 [03:42<01:03, 17.65it/s]\u001b[A\n",
            " 86% 6966/8092 [03:42<01:04, 17.53it/s]\u001b[A\n",
            " 86% 6968/8092 [03:43<01:04, 17.51it/s]\u001b[A\n",
            " 86% 6970/8092 [03:43<01:04, 17.52it/s]\u001b[A\n",
            " 86% 6972/8092 [03:43<01:05, 17.22it/s]\u001b[A\n",
            " 86% 6974/8092 [03:43<01:03, 17.64it/s]\u001b[A\n",
            " 86% 6976/8092 [03:43<01:03, 17.61it/s]\u001b[A\n",
            " 86% 6978/8092 [03:43<01:03, 17.63it/s]\u001b[A\n",
            " 86% 6980/8092 [03:43<01:03, 17.55it/s]\u001b[A\n",
            " 86% 6982/8092 [03:43<01:03, 17.48it/s]\u001b[A\n",
            " 86% 6984/8092 [03:44<01:03, 17.47it/s]\u001b[A\n",
            " 86% 6986/8092 [03:44<01:03, 17.51it/s]\u001b[A\n",
            " 86% 6988/8092 [03:44<01:02, 17.55it/s]\u001b[A\n",
            " 86% 6990/8092 [03:44<01:03, 17.46it/s]\u001b[A\n",
            " 86% 6992/8092 [03:44<01:02, 17.59it/s]\u001b[A\n",
            " 86% 6994/8092 [03:44<01:02, 17.49it/s]\u001b[A\n",
            " 86% 6996/8092 [03:44<01:02, 17.57it/s]\u001b[A\n",
            " 86% 6998/8092 [03:44<01:02, 17.54it/s]\u001b[A\n",
            " 87% 7000/8092 [03:44<01:02, 17.49it/s]\u001b[A\n",
            " 87% 7002/8092 [03:45<01:02, 17.42it/s]\u001b[A\n",
            " 87% 7004/8092 [03:45<01:01, 17.57it/s]\u001b[A\n",
            " 87% 7006/8092 [03:45<01:02, 17.34it/s]\u001b[A\n",
            " 87% 7008/8092 [03:45<01:01, 17.69it/s]\u001b[A\n",
            " 87% 7010/8092 [03:45<01:01, 17.55it/s]\u001b[A\n",
            " 87% 7012/8092 [03:45<01:01, 17.57it/s]\u001b[A\n",
            " 87% 7014/8092 [03:45<01:01, 17.54it/s]\u001b[A\n",
            " 87% 7016/8092 [03:45<01:01, 17.51it/s]\u001b[A\n",
            " 87% 7018/8092 [03:45<01:01, 17.52it/s]\u001b[A\n",
            " 87% 7020/8092 [03:46<01:00, 17.58it/s]\u001b[A\n",
            " 87% 7022/8092 [03:46<01:01, 17.49it/s]\u001b[A\n",
            " 87% 7024/8092 [03:46<01:00, 17.54it/s]\u001b[A\n",
            " 87% 7026/8092 [03:46<01:00, 17.53it/s]\u001b[A\n",
            " 87% 7028/8092 [03:46<01:00, 17.54it/s]\u001b[A\n",
            " 87% 7030/8092 [03:46<01:00, 17.63it/s]\u001b[A\n",
            " 87% 7032/8092 [03:46<01:00, 17.53it/s]\u001b[A\n",
            " 87% 7034/8092 [03:46<01:00, 17.53it/s]\u001b[A\n",
            " 87% 7036/8092 [03:46<01:00, 17.57it/s]\u001b[A\n",
            " 87% 7038/8092 [03:47<01:00, 17.48it/s]\u001b[A\n",
            " 87% 7040/8092 [03:47<00:59, 17.55it/s]\u001b[A\n",
            " 87% 7042/8092 [03:47<00:59, 17.55it/s]\u001b[A\n",
            " 87% 7044/8092 [03:47<00:59, 17.58it/s]\u001b[A\n",
            " 87% 7046/8092 [03:47<00:59, 17.64it/s]\u001b[A\n",
            " 87% 7048/8092 [03:47<00:59, 17.54it/s]\u001b[A\n",
            " 87% 7050/8092 [03:47<00:59, 17.54it/s]\u001b[A\n",
            " 87% 7052/8092 [03:47<00:59, 17.54it/s]\u001b[A\n",
            " 87% 7054/8092 [03:48<00:59, 17.39it/s]\u001b[A\n",
            " 87% 7056/8092 [03:48<00:59, 17.55it/s]\u001b[A\n",
            " 87% 7058/8092 [03:48<00:59, 17.44it/s]\u001b[A\n",
            " 87% 7060/8092 [03:48<00:59, 17.37it/s]\u001b[A\n",
            " 87% 7062/8092 [03:48<00:59, 17.35it/s]\u001b[A\n",
            " 87% 7064/8092 [03:48<00:59, 17.41it/s]\u001b[A\n",
            " 87% 7066/8092 [03:48<00:59, 17.31it/s]\u001b[A\n",
            " 87% 7068/8092 [03:48<00:59, 17.34it/s]\u001b[A\n",
            " 87% 7070/8092 [03:48<00:59, 17.31it/s]\u001b[A\n",
            " 87% 7072/8092 [03:49<00:58, 17.41it/s]\u001b[A\n",
            " 87% 7074/8092 [03:49<00:58, 17.29it/s]\u001b[A\n",
            " 87% 7076/8092 [03:49<00:58, 17.29it/s]\u001b[A\n",
            " 87% 7078/8092 [03:49<00:58, 17.24it/s]\u001b[A\n",
            " 87% 7080/8092 [03:49<00:59, 17.15it/s]\u001b[A\n",
            " 88% 7082/8092 [03:49<00:58, 17.20it/s]\u001b[A\n",
            " 88% 7084/8092 [03:49<00:58, 17.18it/s]\u001b[A\n",
            " 88% 7086/8092 [03:49<00:59, 16.95it/s]\u001b[A\n",
            " 88% 7088/8092 [03:49<00:59, 16.99it/s]\u001b[A\n",
            " 88% 7090/8092 [03:50<00:58, 17.03it/s]\u001b[A\n",
            " 88% 7092/8092 [03:50<00:59, 16.93it/s]\u001b[A\n",
            " 88% 7094/8092 [03:50<00:58, 16.98it/s]\u001b[A\n",
            " 88% 7096/8092 [03:50<00:58, 16.99it/s]\u001b[A\n",
            " 88% 7098/8092 [03:50<00:58, 16.96it/s]\u001b[A\n",
            " 88% 7100/8092 [03:50<00:58, 17.02it/s]\u001b[A\n",
            " 88% 7102/8092 [03:50<00:58, 16.95it/s]\u001b[A\n",
            " 88% 7104/8092 [03:50<00:58, 16.91it/s]\u001b[A\n",
            " 88% 7106/8092 [03:51<00:58, 16.95it/s]\u001b[A\n",
            " 88% 7108/8092 [03:51<00:57, 16.99it/s]\u001b[A\n",
            " 88% 7110/8092 [03:51<00:57, 16.96it/s]\u001b[A\n",
            " 88% 7112/8092 [03:51<00:57, 16.96it/s]\u001b[A\n",
            " 88% 7114/8092 [03:51<00:57, 17.03it/s]\u001b[A\n",
            " 88% 7116/8092 [03:51<00:57, 16.98it/s]\u001b[A\n",
            " 88% 7118/8092 [03:51<00:57, 17.07it/s]\u001b[A\n",
            " 88% 7120/8092 [03:51<00:57, 17.00it/s]\u001b[A\n",
            " 88% 7122/8092 [03:51<00:57, 16.95it/s]\u001b[A\n",
            " 88% 7124/8092 [03:52<00:57, 16.86it/s]\u001b[A\n",
            " 88% 7126/8092 [03:52<00:57, 16.87it/s]\u001b[A\n",
            " 88% 7128/8092 [03:52<00:57, 16.86it/s]\u001b[A\n",
            " 88% 7130/8092 [03:52<00:56, 16.97it/s]\u001b[A\n",
            " 88% 7132/8092 [03:52<00:56, 17.05it/s]\u001b[A\n",
            " 88% 7134/8092 [03:52<00:56, 16.95it/s]\u001b[A\n",
            " 88% 7136/8092 [03:52<00:56, 17.06it/s]\u001b[A\n",
            " 88% 7138/8092 [03:52<00:55, 17.04it/s]\u001b[A\n",
            " 88% 7140/8092 [03:53<00:55, 17.04it/s]\u001b[A\n",
            " 88% 7142/8092 [03:53<00:55, 17.07it/s]\u001b[A\n",
            " 88% 7144/8092 [03:53<00:56, 16.87it/s]\u001b[A\n",
            " 88% 7146/8092 [03:53<00:56, 16.78it/s]\u001b[A\n",
            " 88% 7148/8092 [03:53<00:56, 16.75it/s]\u001b[A\n",
            " 88% 7150/8092 [03:53<00:56, 16.70it/s]\u001b[A\n",
            " 88% 7152/8092 [03:53<00:55, 16.85it/s]\u001b[A\n",
            " 88% 7154/8092 [03:53<00:55, 16.87it/s]\u001b[A\n",
            " 88% 7156/8092 [03:53<00:55, 16.75it/s]\u001b[A\n",
            " 88% 7158/8092 [03:54<00:55, 16.74it/s]\u001b[A\n",
            " 88% 7160/8092 [03:54<00:55, 16.81it/s]\u001b[A\n",
            " 89% 7162/8092 [03:54<00:55, 16.77it/s]\u001b[A\n",
            " 89% 7164/8092 [03:54<00:55, 16.79it/s]\u001b[A\n",
            " 89% 7166/8092 [03:54<00:55, 16.63it/s]\u001b[A\n",
            " 89% 7168/8092 [03:54<00:55, 16.67it/s]\u001b[A\n",
            " 89% 7170/8092 [03:54<00:55, 16.72it/s]\u001b[A\n",
            " 89% 7172/8092 [03:54<00:54, 16.79it/s]\u001b[A\n",
            " 89% 7174/8092 [03:55<00:54, 16.86it/s]\u001b[A\n",
            " 89% 7176/8092 [03:55<00:54, 16.85it/s]\u001b[A\n",
            " 89% 7178/8092 [03:55<00:54, 16.91it/s]\u001b[A\n",
            " 89% 7180/8092 [03:55<00:53, 16.94it/s]\u001b[A\n",
            " 89% 7182/8092 [03:55<00:54, 16.79it/s]\u001b[A\n",
            " 89% 7184/8092 [03:55<00:54, 16.79it/s]\u001b[A\n",
            " 89% 7186/8092 [03:55<00:54, 16.68it/s]\u001b[A\n",
            " 89% 7188/8092 [03:55<00:54, 16.66it/s]\u001b[A\n",
            " 89% 7190/8092 [03:56<00:54, 16.57it/s]\u001b[A\n",
            " 89% 7192/8092 [03:56<00:54, 16.61it/s]\u001b[A\n",
            " 89% 7194/8092 [03:56<00:54, 16.56it/s]\u001b[A\n",
            " 89% 7196/8092 [03:56<00:53, 16.66it/s]\u001b[A\n",
            " 89% 7198/8092 [03:56<00:53, 16.56it/s]\u001b[A\n",
            " 89% 7200/8092 [03:56<00:53, 16.81it/s]\u001b[A\n",
            " 89% 7202/8092 [03:56<00:52, 16.80it/s]\u001b[A\n",
            " 89% 7204/8092 [03:56<00:53, 16.74it/s]\u001b[A\n",
            " 89% 7206/8092 [03:56<00:52, 16.75it/s]\u001b[A\n",
            " 89% 7208/8092 [03:57<00:52, 16.79it/s]\u001b[A\n",
            " 89% 7210/8092 [03:57<00:52, 16.84it/s]\u001b[A\n",
            " 89% 7212/8092 [03:57<00:52, 16.76it/s]\u001b[A\n",
            " 89% 7214/8092 [03:57<00:52, 16.58it/s]\u001b[A\n",
            " 89% 7216/8092 [03:57<00:52, 16.59it/s]\u001b[A\n",
            " 89% 7218/8092 [03:57<00:52, 16.56it/s]\u001b[A\n",
            " 89% 7220/8092 [03:57<00:52, 16.62it/s]\u001b[A\n",
            " 89% 7222/8092 [03:57<00:52, 16.63it/s]\u001b[A\n",
            " 89% 7224/8092 [03:58<00:52, 16.60it/s]\u001b[A\n",
            " 89% 7226/8092 [03:58<00:51, 16.71it/s]\u001b[A\n",
            " 89% 7228/8092 [03:58<00:51, 16.82it/s]\u001b[A\n",
            " 89% 7230/8092 [03:58<00:51, 16.64it/s]\u001b[A\n",
            " 89% 7232/8092 [03:58<00:51, 16.78it/s]\u001b[A\n",
            " 89% 7234/8092 [03:58<00:51, 16.71it/s]\u001b[A\n",
            " 89% 7236/8092 [03:58<00:51, 16.69it/s]\u001b[A\n",
            " 89% 7238/8092 [03:58<00:51, 16.72it/s]\u001b[A\n",
            " 89% 7240/8092 [03:59<00:50, 16.73it/s]\u001b[A\n",
            " 89% 7242/8092 [03:59<00:50, 16.82it/s]\u001b[A\n",
            " 90% 7244/8092 [03:59<00:50, 16.78it/s]\u001b[A\n",
            " 90% 7246/8092 [03:59<00:50, 16.73it/s]\u001b[A\n",
            " 90% 7248/8092 [03:59<00:50, 16.66it/s]\u001b[A\n",
            " 90% 7250/8092 [03:59<00:50, 16.59it/s]\u001b[A\n",
            " 90% 7252/8092 [03:59<00:50, 16.69it/s]\u001b[A\n",
            " 90% 7254/8092 [03:59<00:50, 16.67it/s]\u001b[A\n",
            " 90% 7256/8092 [03:59<00:50, 16.69it/s]\u001b[A\n",
            " 90% 7258/8092 [04:00<00:49, 16.74it/s]\u001b[A\n",
            " 90% 7260/8092 [04:00<00:49, 16.67it/s]\u001b[A\n",
            " 90% 7262/8092 [04:00<00:50, 16.56it/s]\u001b[A\n",
            " 90% 7264/8092 [04:00<00:49, 16.77it/s]\u001b[A\n",
            " 90% 7266/8092 [04:00<00:49, 16.72it/s]\u001b[A\n",
            " 90% 7268/8092 [04:00<00:49, 16.78it/s]\u001b[A\n",
            " 90% 7270/8092 [04:00<00:49, 16.59it/s]\u001b[A\n",
            " 90% 7272/8092 [04:00<00:49, 16.65it/s]\u001b[A\n",
            " 90% 7274/8092 [04:01<00:48, 16.75it/s]\u001b[A\n",
            " 90% 7276/8092 [04:01<00:49, 16.64it/s]\u001b[A\n",
            " 90% 7278/8092 [04:01<00:49, 16.48it/s]\u001b[A\n",
            " 90% 7280/8092 [04:01<00:48, 16.57it/s]\u001b[A\n",
            " 90% 7282/8092 [04:01<00:48, 16.53it/s]\u001b[A\n",
            " 90% 7284/8092 [04:01<00:48, 16.58it/s]\u001b[A\n",
            " 90% 7286/8092 [04:01<00:48, 16.58it/s]\u001b[A\n",
            " 90% 7288/8092 [04:01<00:48, 16.53it/s]\u001b[A\n",
            " 90% 7290/8092 [04:02<00:48, 16.58it/s]\u001b[A\n",
            " 90% 7292/8092 [04:02<00:48, 16.64it/s]\u001b[A\n",
            " 90% 7294/8092 [04:02<00:48, 16.62it/s]\u001b[A\n",
            " 90% 7296/8092 [04:02<00:47, 16.60it/s]\u001b[A\n",
            " 90% 7298/8092 [04:02<00:47, 16.67it/s]\u001b[A\n",
            " 90% 7300/8092 [04:02<00:47, 16.62it/s]\u001b[A\n",
            " 90% 7302/8092 [04:02<00:47, 16.58it/s]\u001b[A\n",
            " 90% 7304/8092 [04:02<00:47, 16.56it/s]\u001b[A\n",
            " 90% 7306/8092 [04:02<00:47, 16.63it/s]\u001b[A\n",
            " 90% 7308/8092 [04:03<00:46, 16.75it/s]\u001b[A\n",
            " 90% 7310/8092 [04:03<00:46, 16.71it/s]\u001b[A\n",
            " 90% 7312/8092 [04:03<00:46, 16.66it/s]\u001b[A\n",
            " 90% 7314/8092 [04:03<00:46, 16.59it/s]\u001b[A\n",
            " 90% 7316/8092 [04:03<00:46, 16.63it/s]\u001b[A\n",
            " 90% 7318/8092 [04:03<00:46, 16.51it/s]\u001b[A\n",
            " 90% 7320/8092 [04:03<00:46, 16.61it/s]\u001b[A\n",
            " 90% 7322/8092 [04:03<00:46, 16.58it/s]\u001b[A\n",
            " 91% 7324/8092 [04:04<00:46, 16.56it/s]\u001b[A\n",
            " 91% 7326/8092 [04:04<00:46, 16.51it/s]\u001b[A\n",
            " 91% 7328/8092 [04:04<00:46, 16.57it/s]\u001b[A\n",
            " 91% 7330/8092 [04:04<00:46, 16.52it/s]\u001b[A\n",
            " 91% 7332/8092 [04:04<00:45, 16.59it/s]\u001b[A\n",
            " 91% 7334/8092 [04:04<00:45, 16.49it/s]\u001b[A\n",
            " 91% 7336/8092 [04:04<00:45, 16.55it/s]\u001b[A\n",
            " 91% 7338/8092 [04:04<00:45, 16.54it/s]\u001b[A\n",
            " 91% 7340/8092 [04:05<00:45, 16.59it/s]\u001b[A\n",
            " 91% 7342/8092 [04:05<00:45, 16.50it/s]\u001b[A\n",
            " 91% 7344/8092 [04:05<00:45, 16.49it/s]\u001b[A\n",
            " 91% 7346/8092 [04:05<00:45, 16.45it/s]\u001b[A\n",
            " 91% 7348/8092 [04:05<00:45, 16.51it/s]\u001b[A\n",
            " 91% 7350/8092 [04:05<00:44, 16.53it/s]\u001b[A\n",
            " 91% 7352/8092 [04:05<00:44, 16.48it/s]\u001b[A\n",
            " 91% 7354/8092 [04:05<00:45, 16.31it/s]\u001b[A\n",
            " 91% 7356/8092 [04:06<00:44, 16.60it/s]\u001b[A\n",
            " 91% 7358/8092 [04:06<00:44, 16.54it/s]\u001b[A\n",
            " 91% 7360/8092 [04:06<00:44, 16.44it/s]\u001b[A\n",
            " 91% 7362/8092 [04:06<00:44, 16.50it/s]\u001b[A\n",
            " 91% 7364/8092 [04:06<00:44, 16.44it/s]\u001b[A\n",
            " 91% 7366/8092 [04:06<00:44, 16.47it/s]\u001b[A\n",
            " 91% 7368/8092 [04:06<00:43, 16.51it/s]\u001b[A\n",
            " 91% 7370/8092 [04:06<00:43, 16.48it/s]\u001b[A\n",
            " 91% 7372/8092 [04:06<00:43, 16.51it/s]\u001b[A\n",
            " 91% 7374/8092 [04:07<00:43, 16.51it/s]\u001b[A\n",
            " 91% 7376/8092 [04:07<00:43, 16.53it/s]\u001b[A\n",
            " 91% 7378/8092 [04:07<00:43, 16.55it/s]\u001b[A\n",
            " 91% 7380/8092 [04:07<00:43, 16.55it/s]\u001b[A\n",
            " 91% 7382/8092 [04:07<00:43, 16.51it/s]\u001b[A\n",
            " 91% 7384/8092 [04:07<00:42, 16.48it/s]\u001b[A\n",
            " 91% 7386/8092 [04:07<00:42, 16.44it/s]\u001b[A\n",
            " 91% 7388/8092 [04:07<00:42, 16.48it/s]\u001b[A\n",
            " 91% 7390/8092 [04:08<00:42, 16.50it/s]\u001b[A\n",
            " 91% 7392/8092 [04:08<00:42, 16.47it/s]\u001b[A\n",
            " 91% 7394/8092 [04:08<00:42, 16.52it/s]\u001b[A\n",
            " 91% 7396/8092 [04:08<00:42, 16.48it/s]\u001b[A\n",
            " 91% 7398/8092 [04:08<00:41, 16.57it/s]\u001b[A\n",
            " 91% 7400/8092 [04:08<00:41, 16.48it/s]\u001b[A\n",
            " 91% 7402/8092 [04:08<00:41, 16.49it/s]\u001b[A\n",
            " 91% 7404/8092 [04:08<00:41, 16.55it/s]\u001b[A\n",
            " 92% 7406/8092 [04:09<00:41, 16.53it/s]\u001b[A\n",
            " 92% 7408/8092 [04:09<00:41, 16.52it/s]\u001b[A\n",
            " 92% 7410/8092 [04:09<00:41, 16.47it/s]\u001b[A\n",
            " 92% 7412/8092 [04:09<00:41, 16.46it/s]\u001b[A\n",
            " 92% 7414/8092 [04:09<00:41, 16.47it/s]\u001b[A\n",
            " 92% 7416/8092 [04:09<00:41, 16.45it/s]\u001b[A\n",
            " 92% 7418/8092 [04:09<00:40, 16.52it/s]\u001b[A\n",
            " 92% 7420/8092 [04:09<00:40, 16.53it/s]\u001b[A\n",
            " 92% 7422/8092 [04:10<00:40, 16.41it/s]\u001b[A\n",
            " 92% 7424/8092 [04:10<00:40, 16.63it/s]\u001b[A\n",
            " 92% 7426/8092 [04:10<00:40, 16.54it/s]\u001b[A\n",
            " 92% 7428/8092 [04:10<00:40, 16.52it/s]\u001b[A\n",
            " 92% 7430/8092 [04:10<00:39, 16.55it/s]\u001b[A\n",
            " 92% 7432/8092 [04:10<00:40, 16.47it/s]\u001b[A\n",
            " 92% 7434/8092 [04:10<00:39, 16.51it/s]\u001b[A\n",
            " 92% 7436/8092 [04:10<00:39, 16.57it/s]\u001b[A\n",
            " 92% 7438/8092 [04:10<00:39, 16.42it/s]\u001b[A\n",
            " 92% 7440/8092 [04:11<00:39, 16.53it/s]\u001b[A\n",
            " 92% 7442/8092 [04:11<00:39, 16.54it/s]\u001b[A\n",
            " 92% 7444/8092 [04:11<00:39, 16.50it/s]\u001b[A\n",
            " 92% 7446/8092 [04:11<00:39, 16.51it/s]\u001b[A\n",
            " 92% 7448/8092 [04:11<00:39, 16.46it/s]\u001b[A\n",
            " 92% 7450/8092 [04:11<00:38, 16.49it/s]\u001b[A\n",
            " 92% 7452/8092 [04:11<00:38, 16.51it/s]\u001b[A\n",
            " 92% 7454/8092 [04:11<00:38, 16.56it/s]\u001b[A\n",
            " 92% 7456/8092 [04:12<00:38, 16.56it/s]\u001b[A\n",
            " 92% 7458/8092 [04:12<00:38, 16.48it/s]\u001b[A\n",
            " 92% 7460/8092 [04:12<00:38, 16.47it/s]\u001b[A\n",
            " 92% 7462/8092 [04:12<00:38, 16.49it/s]\u001b[A\n",
            " 92% 7464/8092 [04:12<00:37, 16.53it/s]\u001b[A\n",
            " 92% 7466/8092 [04:12<00:37, 16.52it/s]\u001b[A\n",
            " 92% 7468/8092 [04:12<00:37, 16.55it/s]\u001b[A\n",
            " 92% 7470/8092 [04:12<00:37, 16.40it/s]\u001b[A\n",
            " 92% 7472/8092 [04:13<00:37, 16.49it/s]\u001b[A\n",
            " 92% 7474/8092 [04:13<00:37, 16.56it/s]\u001b[A\n",
            " 92% 7476/8092 [04:13<00:37, 16.49it/s]\u001b[A\n",
            " 92% 7478/8092 [04:13<00:37, 16.50it/s]\u001b[A\n",
            " 92% 7480/8092 [04:13<00:37, 16.42it/s]\u001b[A\n",
            " 92% 7482/8092 [04:13<00:36, 16.51it/s]\u001b[A\n",
            " 92% 7484/8092 [04:13<00:36, 16.55it/s]\u001b[A\n",
            " 93% 7486/8092 [04:13<00:36, 16.57it/s]\u001b[A\n",
            " 93% 7488/8092 [04:14<00:36, 16.55it/s]\u001b[A\n",
            " 93% 7490/8092 [04:14<00:36, 16.51it/s]\u001b[A\n",
            " 93% 7492/8092 [04:14<00:36, 16.54it/s]\u001b[A\n",
            " 93% 7494/8092 [04:14<00:36, 16.53it/s]\u001b[A\n",
            " 93% 7496/8092 [04:14<00:36, 16.50it/s]\u001b[A\n",
            " 93% 7498/8092 [04:14<00:35, 16.54it/s]\u001b[A\n",
            " 93% 7500/8092 [04:14<00:35, 16.54it/s]\u001b[A\n",
            " 93% 7502/8092 [04:14<00:35, 16.58it/s]\u001b[A\n",
            " 93% 7504/8092 [04:14<00:35, 16.57it/s]\u001b[A\n",
            " 93% 7506/8092 [04:15<00:35, 16.55it/s]\u001b[A\n",
            " 93% 7508/8092 [04:15<00:35, 16.55it/s]\u001b[A\n",
            " 93% 7510/8092 [04:15<00:35, 16.49it/s]\u001b[A\n",
            " 93% 7512/8092 [04:15<00:35, 16.50it/s]\u001b[A\n",
            " 93% 7514/8092 [04:15<00:34, 16.53it/s]\u001b[A\n",
            " 93% 7516/8092 [04:15<00:34, 16.53it/s]\u001b[A\n",
            " 93% 7518/8092 [04:15<00:34, 16.45it/s]\u001b[A\n",
            " 93% 7520/8092 [04:15<00:34, 16.49it/s]\u001b[A\n",
            " 93% 7522/8092 [04:16<00:34, 16.57it/s]\u001b[A\n",
            " 93% 7524/8092 [04:16<00:34, 16.48it/s]\u001b[A\n",
            " 93% 7526/8092 [04:16<00:34, 16.48it/s]\u001b[A\n",
            " 93% 7528/8092 [04:16<00:34, 16.50it/s]\u001b[A\n",
            " 93% 7530/8092 [04:16<00:34, 16.47it/s]\u001b[A\n",
            " 93% 7532/8092 [04:16<00:33, 16.51it/s]\u001b[A\n",
            " 93% 7534/8092 [04:16<00:33, 16.53it/s]\u001b[A\n",
            " 93% 7536/8092 [04:16<00:33, 16.51it/s]\u001b[A\n",
            " 93% 7538/8092 [04:17<00:33, 16.48it/s]\u001b[A\n",
            " 93% 7540/8092 [04:17<00:33, 16.51it/s]\u001b[A\n",
            " 93% 7542/8092 [04:17<00:33, 16.49it/s]\u001b[A\n",
            " 93% 7544/8092 [04:17<00:33, 16.48it/s]\u001b[A\n",
            " 93% 7546/8092 [04:17<00:32, 16.56it/s]\u001b[A\n",
            " 93% 7548/8092 [04:17<00:32, 16.49it/s]\u001b[A\n",
            " 93% 7550/8092 [04:17<00:32, 16.49it/s]\u001b[A\n",
            " 93% 7552/8092 [04:17<00:32, 16.45it/s]\u001b[A\n",
            " 93% 7554/8092 [04:18<00:32, 16.51it/s]\u001b[A\n",
            " 93% 7556/8092 [04:18<00:32, 16.52it/s]\u001b[A\n",
            " 93% 7558/8092 [04:18<00:32, 16.45it/s]\u001b[A\n",
            " 93% 7560/8092 [04:18<00:32, 16.43it/s]\u001b[A\n",
            " 93% 7562/8092 [04:18<00:32, 16.49it/s]\u001b[A\n",
            " 93% 7564/8092 [04:18<00:31, 16.51it/s]\u001b[A\n",
            " 93% 7566/8092 [04:18<00:31, 16.46it/s]\u001b[A\n",
            " 94% 7568/8092 [04:18<00:31, 16.49it/s]\u001b[A\n",
            " 94% 7570/8092 [04:18<00:31, 16.53it/s]\u001b[A\n",
            " 94% 7572/8092 [04:19<00:31, 16.44it/s]\u001b[A\n",
            " 94% 7574/8092 [04:19<00:31, 16.46it/s]\u001b[A\n",
            " 94% 7576/8092 [04:19<00:31, 16.50it/s]\u001b[A\n",
            " 94% 7578/8092 [04:19<00:31, 16.52it/s]\u001b[A\n",
            " 94% 7580/8092 [04:19<00:31, 16.49it/s]\u001b[A\n",
            " 94% 7582/8092 [04:19<00:31, 16.44it/s]\u001b[A\n",
            " 94% 7584/8092 [04:19<00:30, 16.45it/s]\u001b[A\n",
            " 94% 7586/8092 [04:19<00:30, 16.45it/s]\u001b[A\n",
            " 94% 7588/8092 [04:20<00:30, 16.47it/s]\u001b[A\n",
            " 94% 7590/8092 [04:20<00:30, 16.41it/s]\u001b[A\n",
            " 94% 7592/8092 [04:20<00:30, 16.36it/s]\u001b[A\n",
            " 94% 7594/8092 [04:20<00:30, 16.43it/s]\u001b[A\n",
            " 94% 7596/8092 [04:20<00:30, 16.49it/s]\u001b[A\n",
            " 94% 7598/8092 [04:20<00:30, 16.47it/s]\u001b[A\n",
            " 94% 7600/8092 [04:20<00:29, 16.51it/s]\u001b[A\n",
            " 94% 7602/8092 [04:20<00:29, 16.51it/s]\u001b[A\n",
            " 94% 7604/8092 [04:21<00:29, 16.47it/s]\u001b[A\n",
            " 94% 7606/8092 [04:21<00:29, 16.44it/s]\u001b[A\n",
            " 94% 7608/8092 [04:21<00:29, 16.48it/s]\u001b[A\n",
            " 94% 7610/8092 [04:21<00:29, 16.50it/s]\u001b[A\n",
            " 94% 7612/8092 [04:21<00:29, 16.44it/s]\u001b[A\n",
            " 94% 7614/8092 [04:21<00:29, 16.38it/s]\u001b[A\n",
            " 94% 7616/8092 [04:21<00:28, 16.53it/s]\u001b[A\n",
            " 94% 7618/8092 [04:21<00:28, 16.47it/s]\u001b[A\n",
            " 94% 7620/8092 [04:22<00:28, 16.55it/s]\u001b[A\n",
            " 94% 7622/8092 [04:22<00:28, 16.45it/s]\u001b[A\n",
            " 94% 7624/8092 [04:22<00:28, 16.40it/s]\u001b[A\n",
            " 94% 7626/8092 [04:22<00:28, 16.50it/s]\u001b[A\n",
            " 94% 7628/8092 [04:22<00:28, 16.50it/s]\u001b[A\n",
            " 94% 7630/8092 [04:22<00:28, 16.49it/s]\u001b[A\n",
            " 94% 7632/8092 [04:22<00:27, 16.49it/s]\u001b[A\n",
            " 94% 7634/8092 [04:22<00:27, 16.50it/s]\u001b[A\n",
            " 94% 7636/8092 [04:22<00:27, 16.51it/s]\u001b[A\n",
            " 94% 7638/8092 [04:23<00:27, 16.48it/s]\u001b[A\n",
            " 94% 7640/8092 [04:23<00:27, 16.47it/s]\u001b[A\n",
            " 94% 7642/8092 [04:23<00:27, 16.48it/s]\u001b[A\n",
            " 94% 7644/8092 [04:23<00:27, 16.43it/s]\u001b[A\n",
            " 94% 7646/8092 [04:23<00:27, 16.40it/s]\u001b[A\n",
            " 95% 7648/8092 [04:23<00:26, 16.46it/s]\u001b[A\n",
            " 95% 7650/8092 [04:23<00:26, 16.52it/s]\u001b[A\n",
            " 95% 7652/8092 [04:23<00:26, 16.54it/s]\u001b[A\n",
            " 95% 7654/8092 [04:24<00:26, 16.49it/s]\u001b[A\n",
            " 95% 7656/8092 [04:24<00:26, 16.43it/s]\u001b[A\n",
            " 95% 7658/8092 [04:24<00:26, 16.29it/s]\u001b[A\n",
            " 95% 7660/8092 [04:24<00:26, 16.24it/s]\u001b[A\n",
            " 95% 7662/8092 [04:24<00:26, 16.26it/s]\u001b[A\n",
            " 95% 7664/8092 [04:24<00:26, 16.25it/s]\u001b[A\n",
            " 95% 7666/8092 [04:24<00:26, 16.30it/s]\u001b[A\n",
            " 95% 7668/8092 [04:24<00:26, 16.26it/s]\u001b[A\n",
            " 95% 7670/8092 [04:25<00:26, 16.18it/s]\u001b[A\n",
            " 95% 7672/8092 [04:25<00:25, 16.23it/s]\u001b[A\n",
            " 95% 7674/8092 [04:25<00:25, 16.28it/s]\u001b[A\n",
            " 95% 7676/8092 [04:25<00:25, 16.30it/s]\u001b[A\n",
            " 95% 7678/8092 [04:25<00:25, 16.25it/s]\u001b[A\n",
            " 95% 7680/8092 [04:25<00:25, 16.26it/s]\u001b[A\n",
            " 95% 7682/8092 [04:25<00:25, 16.32it/s]\u001b[A\n",
            " 95% 7684/8092 [04:25<00:24, 16.33it/s]\u001b[A\n",
            " 95% 7686/8092 [04:26<00:24, 16.28it/s]\u001b[A\n",
            " 95% 7688/8092 [04:26<00:24, 16.28it/s]\u001b[A\n",
            " 95% 7690/8092 [04:26<00:24, 16.19it/s]\u001b[A\n",
            " 95% 7692/8092 [04:26<00:24, 16.18it/s]\u001b[A\n",
            " 95% 7694/8092 [04:26<00:24, 16.18it/s]\u001b[A\n",
            " 95% 7696/8092 [04:26<00:24, 16.31it/s]\u001b[A\n",
            " 95% 7698/8092 [04:26<00:24, 16.26it/s]\u001b[A\n",
            " 95% 7700/8092 [04:26<00:24, 16.33it/s]\u001b[A\n",
            " 95% 7702/8092 [04:27<00:23, 16.27it/s]\u001b[A\n",
            " 95% 7704/8092 [04:27<00:23, 16.21it/s]\u001b[A\n",
            " 95% 7706/8092 [04:27<00:23, 16.21it/s]\u001b[A\n",
            " 95% 7708/8092 [04:27<00:23, 16.18it/s]\u001b[A\n",
            " 95% 7710/8092 [04:27<00:23, 16.04it/s]\u001b[A\n",
            " 95% 7712/8092 [04:27<00:23, 16.16it/s]\u001b[A\n",
            " 95% 7714/8092 [04:27<00:23, 16.16it/s]\u001b[A\n",
            " 95% 7716/8092 [04:27<00:23, 16.18it/s]\u001b[A\n",
            " 95% 7718/8092 [04:28<00:23, 16.13it/s]\u001b[A\n",
            " 95% 7720/8092 [04:28<00:22, 16.22it/s]\u001b[A\n",
            " 95% 7722/8092 [04:28<00:22, 16.20it/s]\u001b[A\n",
            " 95% 7724/8092 [04:28<00:22, 16.16it/s]\u001b[A\n",
            " 95% 7726/8092 [04:28<00:22, 16.16it/s]\u001b[A\n",
            " 96% 7728/8092 [04:28<00:22, 16.21it/s]\u001b[A\n",
            " 96% 7730/8092 [04:28<00:22, 16.15it/s]\u001b[A\n",
            " 96% 7732/8092 [04:28<00:22, 16.16it/s]\u001b[A\n",
            " 96% 7734/8092 [04:29<00:22, 16.06it/s]\u001b[A\n",
            " 96% 7736/8092 [04:29<00:22, 16.14it/s]\u001b[A\n",
            " 96% 7738/8092 [04:29<00:21, 16.19it/s]\u001b[A\n",
            " 96% 7740/8092 [04:29<00:21, 16.15it/s]\u001b[A\n",
            " 96% 7742/8092 [04:29<00:21, 16.07it/s]\u001b[A\n",
            " 96% 7744/8092 [04:29<00:21, 16.16it/s]\u001b[A\n",
            " 96% 7746/8092 [04:29<00:21, 16.11it/s]\u001b[A\n",
            " 96% 7748/8092 [04:29<00:21, 16.19it/s]\u001b[A\n",
            " 96% 7750/8092 [04:30<00:21, 16.11it/s]\u001b[A\n",
            " 96% 7752/8092 [04:30<00:21, 16.14it/s]\u001b[A\n",
            " 96% 7754/8092 [04:30<00:21, 16.08it/s]\u001b[A\n",
            " 96% 7756/8092 [04:30<00:20, 16.07it/s]\u001b[A\n",
            " 96% 7758/8092 [04:30<00:20, 16.00it/s]\u001b[A\n",
            " 96% 7760/8092 [04:30<00:20, 16.10it/s]\u001b[A\n",
            " 96% 7762/8092 [04:30<00:20, 16.09it/s]\u001b[A\n",
            " 96% 7764/8092 [04:30<00:20, 16.06it/s]\u001b[A\n",
            " 96% 7766/8092 [04:31<00:20, 16.02it/s]\u001b[A\n",
            " 96% 7768/8092 [04:31<00:20, 16.04it/s]\u001b[A\n",
            " 96% 7770/8092 [04:31<00:20, 16.04it/s]\u001b[A\n",
            " 96% 7772/8092 [04:31<00:19, 16.11it/s]\u001b[A\n",
            " 96% 7774/8092 [04:31<00:19, 16.09it/s]\u001b[A\n",
            " 96% 7776/8092 [04:31<00:19, 16.11it/s]\u001b[A\n",
            " 96% 7778/8092 [04:31<00:19, 16.16it/s]\u001b[A\n",
            " 96% 7780/8092 [04:31<00:19, 15.98it/s]\u001b[A\n",
            " 96% 7782/8092 [04:31<00:19, 16.06it/s]\u001b[A\n",
            " 96% 7784/8092 [04:32<00:19, 16.07it/s]\u001b[A\n",
            " 96% 7786/8092 [04:32<00:19, 15.96it/s]\u001b[A\n",
            " 96% 7788/8092 [04:32<00:18, 16.03it/s]\u001b[A\n",
            " 96% 7790/8092 [04:32<00:19, 15.87it/s]\u001b[A\n",
            " 96% 7792/8092 [04:32<00:18, 15.87it/s]\u001b[A\n",
            " 96% 7794/8092 [04:32<00:18, 16.00it/s]\u001b[A\n",
            " 96% 7796/8092 [04:32<00:18, 15.97it/s]\u001b[A\n",
            " 96% 7798/8092 [04:32<00:18, 16.02it/s]\u001b[A\n",
            " 96% 7800/8092 [04:33<00:18, 16.11it/s]\u001b[A\n",
            " 96% 7802/8092 [04:33<00:18, 16.05it/s]\u001b[A\n",
            " 96% 7804/8092 [04:33<00:17, 16.08it/s]\u001b[A\n",
            " 96% 7806/8092 [04:33<00:17, 15.99it/s]\u001b[A\n",
            " 96% 7808/8092 [04:33<00:17, 16.01it/s]\u001b[A\n",
            " 97% 7810/8092 [04:33<00:17, 16.03it/s]\u001b[A\n",
            " 97% 7812/8092 [04:33<00:17, 15.96it/s]\u001b[A\n",
            " 97% 7814/8092 [04:33<00:17, 15.97it/s]\u001b[A\n",
            " 97% 7816/8092 [04:34<00:17, 16.12it/s]\u001b[A\n",
            " 97% 7818/8092 [04:34<00:17, 16.05it/s]\u001b[A\n",
            " 97% 7820/8092 [04:34<00:16, 16.04it/s]\u001b[A\n",
            " 97% 7822/8092 [04:34<00:16, 16.02it/s]\u001b[A\n",
            " 97% 7824/8092 [04:34<00:16, 15.99it/s]\u001b[A\n",
            " 97% 7826/8092 [04:34<00:16, 15.99it/s]\u001b[A\n",
            " 97% 7828/8092 [04:34<00:16, 16.17it/s]\u001b[A\n",
            " 97% 7830/8092 [04:34<00:16, 16.02it/s]\u001b[A\n",
            " 97% 7832/8092 [04:35<00:16, 16.03it/s]\u001b[A\n",
            " 97% 7834/8092 [04:35<00:16, 16.00it/s]\u001b[A\n",
            " 97% 7836/8092 [04:35<00:16, 15.97it/s]\u001b[A\n",
            " 97% 7838/8092 [04:35<00:15, 16.01it/s]\u001b[A\n",
            " 97% 7840/8092 [04:35<00:15, 16.14it/s]\u001b[A\n",
            " 97% 7842/8092 [04:35<00:15, 16.11it/s]\u001b[A\n",
            " 97% 7844/8092 [04:35<00:15, 16.13it/s]\u001b[A\n",
            " 97% 7846/8092 [04:35<00:15, 16.08it/s]\u001b[A\n",
            " 97% 7848/8092 [04:36<00:15, 16.01it/s]\u001b[A\n",
            " 97% 7850/8092 [04:36<00:15, 15.99it/s]\u001b[A\n",
            " 97% 7852/8092 [04:36<00:14, 16.11it/s]\u001b[A\n",
            " 97% 7854/8092 [04:36<00:14, 16.02it/s]\u001b[A\n",
            " 97% 7856/8092 [04:36<00:14, 16.07it/s]\u001b[A\n",
            " 97% 7858/8092 [04:36<00:14, 16.11it/s]\u001b[A\n",
            " 97% 7860/8092 [04:36<00:14, 16.15it/s]\u001b[A\n",
            " 97% 7862/8092 [04:36<00:14, 16.14it/s]\u001b[A\n",
            " 97% 7864/8092 [04:37<00:14, 16.09it/s]\u001b[A\n",
            " 97% 7866/8092 [04:37<00:14, 16.07it/s]\u001b[A\n",
            " 97% 7868/8092 [04:37<00:13, 16.10it/s]\u001b[A\n",
            " 97% 7870/8092 [04:37<00:13, 16.07it/s]\u001b[A\n",
            " 97% 7872/8092 [04:37<00:13, 16.16it/s]\u001b[A\n",
            " 97% 7874/8092 [04:37<00:13, 16.06it/s]\u001b[A\n",
            " 97% 7876/8092 [04:37<00:13, 16.02it/s]\u001b[A\n",
            " 97% 7878/8092 [04:37<00:13, 16.10it/s]\u001b[A\n",
            " 97% 7880/8092 [04:38<00:13, 16.10it/s]\u001b[A\n",
            " 97% 7882/8092 [04:38<00:12, 16.19it/s]\u001b[A\n",
            " 97% 7884/8092 [04:38<00:12, 16.24it/s]\u001b[A\n",
            " 97% 7886/8092 [04:38<00:12, 16.27it/s]\u001b[A\n",
            " 97% 7888/8092 [04:38<00:12, 16.17it/s]\u001b[A\n",
            " 98% 7890/8092 [04:38<00:12, 16.14it/s]\u001b[A\n",
            " 98% 7892/8092 [04:38<00:12, 16.03it/s]\u001b[A\n",
            " 98% 7894/8092 [04:38<00:12, 16.00it/s]\u001b[A\n",
            " 98% 7896/8092 [04:39<00:12, 15.94it/s]\u001b[A\n",
            " 98% 7898/8092 [04:39<00:12, 16.01it/s]\u001b[A\n",
            " 98% 7900/8092 [04:39<00:11, 16.02it/s]\u001b[A\n",
            " 98% 7902/8092 [04:39<00:11, 15.98it/s]\u001b[A\n",
            " 98% 7904/8092 [04:39<00:11, 16.07it/s]\u001b[A\n",
            " 98% 7906/8092 [04:39<00:11, 16.00it/s]\u001b[A\n",
            " 98% 7908/8092 [04:39<00:11, 16.05it/s]\u001b[A\n",
            " 98% 7910/8092 [04:39<00:11, 16.00it/s]\u001b[A\n",
            " 98% 7912/8092 [04:40<00:11, 16.05it/s]\u001b[A\n",
            " 98% 7914/8092 [04:40<00:11, 16.06it/s]\u001b[A\n",
            " 98% 7916/8092 [04:40<00:10, 16.03it/s]\u001b[A\n",
            " 98% 7918/8092 [04:40<00:10, 16.00it/s]\u001b[A\n",
            " 98% 7920/8092 [04:40<00:10, 16.11it/s]\u001b[A\n",
            " 98% 7922/8092 [04:40<00:10, 16.09it/s]\u001b[A\n",
            " 98% 7924/8092 [04:40<00:10, 16.06it/s]\u001b[A\n",
            " 98% 7926/8092 [04:40<00:10, 15.97it/s]\u001b[A\n",
            " 98% 7928/8092 [04:41<00:10, 15.91it/s]\u001b[A\n",
            " 98% 7930/8092 [04:41<00:10, 15.95it/s]\u001b[A\n",
            " 98% 7932/8092 [04:41<00:09, 16.10it/s]\u001b[A\n",
            " 98% 7934/8092 [04:41<00:09, 16.03it/s]\u001b[A\n",
            " 98% 7936/8092 [04:41<00:09, 16.06it/s]\u001b[A\n",
            " 98% 7938/8092 [04:41<00:09, 16.02it/s]\u001b[A\n",
            " 98% 7940/8092 [04:41<00:09, 15.93it/s]\u001b[A\n",
            " 98% 7942/8092 [04:41<00:09, 15.97it/s]\u001b[A\n",
            " 98% 7944/8092 [04:42<00:09, 15.97it/s]\u001b[A\n",
            " 98% 7946/8092 [04:42<00:09, 16.00it/s]\u001b[A\n",
            " 98% 7948/8092 [04:42<00:08, 16.03it/s]\u001b[A\n",
            " 98% 7950/8092 [04:42<00:08, 15.99it/s]\u001b[A\n",
            " 98% 7952/8092 [04:42<00:08, 15.93it/s]\u001b[A\n",
            " 98% 7954/8092 [04:42<00:08, 16.01it/s]\u001b[A\n",
            " 98% 7956/8092 [04:42<00:08, 16.06it/s]\u001b[A\n",
            " 98% 7958/8092 [04:42<00:08, 16.01it/s]\u001b[A\n",
            " 98% 7960/8092 [04:43<00:08, 15.95it/s]\u001b[A\n",
            " 98% 7962/8092 [04:43<00:08, 16.02it/s]\u001b[A\n",
            " 98% 7964/8092 [04:43<00:07, 16.02it/s]\u001b[A\n",
            " 98% 7966/8092 [04:43<00:07, 16.12it/s]\u001b[A\n",
            " 98% 7968/8092 [04:43<00:07, 16.02it/s]\u001b[A\n",
            " 98% 7970/8092 [04:43<00:07, 16.00it/s]\u001b[A\n",
            " 99% 7972/8092 [04:43<00:07, 15.90it/s]\u001b[A\n",
            " 99% 7974/8092 [04:43<00:07, 15.87it/s]\u001b[A\n",
            " 99% 7976/8092 [04:44<00:07, 15.97it/s]\u001b[A\n",
            " 99% 7978/8092 [04:44<00:07, 15.99it/s]\u001b[A\n",
            " 99% 7980/8092 [04:44<00:07, 15.94it/s]\u001b[A\n",
            " 99% 7982/8092 [04:44<00:06, 15.84it/s]\u001b[A\n",
            " 99% 7984/8092 [04:44<00:06, 15.96it/s]\u001b[A\n",
            " 99% 7986/8092 [04:44<00:06, 15.97it/s]\u001b[A\n",
            " 99% 7988/8092 [04:44<00:06, 16.01it/s]\u001b[A\n",
            " 99% 7990/8092 [04:44<00:06, 15.93it/s]\u001b[A\n",
            " 99% 7992/8092 [04:45<00:06, 15.91it/s]\u001b[A\n",
            " 99% 7994/8092 [04:45<00:06, 15.91it/s]\u001b[A\n",
            " 99% 7996/8092 [04:45<00:06, 15.90it/s]\u001b[A\n",
            " 99% 7998/8092 [04:45<00:05, 15.79it/s]\u001b[A\n",
            " 99% 8000/8092 [04:45<00:05, 15.90it/s]\u001b[A\n",
            " 99% 8002/8092 [04:45<00:05, 15.87it/s]\u001b[A\n",
            " 99% 8004/8092 [04:45<00:05, 15.89it/s]\u001b[A\n",
            " 99% 8006/8092 [04:45<00:05, 15.77it/s]\u001b[A\n",
            " 99% 8008/8092 [04:46<00:05, 15.87it/s]\u001b[A\n",
            " 99% 8010/8092 [04:46<00:05, 15.82it/s]\u001b[A\n",
            " 99% 8012/8092 [04:46<00:05, 15.78it/s]\u001b[A\n",
            " 99% 8014/8092 [04:46<00:04, 15.80it/s]\u001b[A\n",
            " 99% 8016/8092 [04:46<00:04, 15.66it/s]\u001b[A\n",
            " 99% 8018/8092 [04:46<00:04, 15.95it/s]\u001b[A\n",
            " 99% 8020/8092 [04:46<00:04, 15.91it/s]\u001b[A\n",
            " 99% 8022/8092 [04:46<00:04, 15.88it/s]\u001b[A\n",
            " 99% 8024/8092 [04:47<00:04, 15.93it/s]\u001b[A\n",
            " 99% 8026/8092 [04:47<00:04, 15.98it/s]\u001b[A\n",
            " 99% 8028/8092 [04:47<00:04, 15.81it/s]\u001b[A\n",
            " 99% 8030/8092 [04:47<00:03, 15.94it/s]\u001b[A\n",
            " 99% 8032/8092 [04:47<00:03, 16.04it/s]\u001b[A\n",
            " 99% 8034/8092 [04:47<00:03, 15.99it/s]\u001b[A\n",
            " 99% 8036/8092 [04:47<00:03, 15.93it/s]\u001b[A\n",
            " 99% 8038/8092 [04:47<00:03, 15.90it/s]\u001b[A\n",
            " 99% 8040/8092 [04:48<00:03, 15.96it/s]\u001b[A\n",
            " 99% 8042/8092 [04:48<00:03, 15.97it/s]\u001b[A\n",
            " 99% 8044/8092 [04:48<00:03, 15.90it/s]\u001b[A\n",
            " 99% 8046/8092 [04:48<00:02, 15.91it/s]\u001b[A\n",
            " 99% 8048/8092 [04:48<00:02, 15.93it/s]\u001b[A\n",
            " 99% 8050/8092 [04:48<00:02, 15.96it/s]\u001b[A\n",
            "100% 8052/8092 [04:48<00:02, 15.86it/s]\u001b[A\n",
            "100% 8054/8092 [04:49<00:02, 15.83it/s]\u001b[A\n",
            "100% 8056/8092 [04:49<00:02, 15.79it/s]\u001b[A\n",
            "100% 8058/8092 [04:49<00:02, 15.82it/s]\u001b[A\n",
            "100% 8060/8092 [04:49<00:02, 15.81it/s]\u001b[A\n",
            "100% 8062/8092 [04:49<00:01, 15.79it/s]\u001b[A\n",
            "100% 8064/8092 [04:49<00:01, 15.81it/s]\u001b[A\n",
            "100% 8066/8092 [04:49<00:01, 15.83it/s]\u001b[A\n",
            "100% 8068/8092 [04:49<00:01, 15.89it/s]\u001b[A\n",
            "100% 8070/8092 [04:50<00:01, 15.90it/s]\u001b[A\n",
            "100% 8072/8092 [04:50<00:01, 15.88it/s]\u001b[A\n",
            "100% 8074/8092 [04:50<00:01, 15.80it/s]\u001b[A\n",
            "100% 8076/8092 [04:50<00:01, 15.86it/s]\u001b[A\n",
            "100% 8078/8092 [04:50<00:00, 15.90it/s]\u001b[A\n",
            "100% 8080/8092 [04:50<00:00, 15.84it/s]\u001b[A\n",
            "100% 8082/8092 [04:50<00:00, 15.77it/s]\u001b[A\n",
            "100% 8084/8092 [04:50<00:00, 15.80it/s]\u001b[A\n",
            "100% 8086/8092 [04:51<00:00, 15.72it/s]\u001b[A\n",
            "100% 8088/8092 [04:51<00:00, 15.68it/s]\u001b[A\n",
            "100% 8090/8092 [04:51<00:00, 15.72it/s]\u001b[A\n",
            "100% 8092/8092 [04:51<00:00, 27.77it/s]\n",
            " the miles crn'' prombjnit;&als) who every a fundne many implied to hallowly of the [[Thish Polor]]'s}}  In [[preduce symbigand sou] * Pegen]:   [[Chology cist talphight, pr become station] ''[[Aztility]]:Eds], [[UKinnation of Barker], he [[Monomon Natuals ==] *The Gumonicidance===== *[http:/xamblieseat of the Sientl]  [[Ciregriate postleate of Dadia]], Arthus dayong, axamps thein]] in [[1600]]'s most of of crsalist cally ages ''[[1988]]'' */&at;/mizh also bothi&amp;mpash;/ter&gt;. The sand Cructice in Barker and from the [[Tamon Scitismine Chlikike]]. * The [[Amy Mad all Mobalkame] with Cleen, as conticipief:Engly sion conside|Diatery Clandan''Barticily (Amba) *[[Gatural Gand compitinus]]'':Aussoniposof|Moreat of the furd|referentury of''''('''''''''''''' ** [[Polot plÃ©rivo'', 1900]]&quot; &lt;/math&quot; consents * [[Compainism]]s. Ad. Bother addist remishs of [[huoverians]|Gra point Productioces Cichns]], an tukgned.    One the seave data examsinian inalimerial permain periodian barginard the nikey whight|Pireo fremboted I&lt;/taccesht&quot; in survie photwer (174<id>       <text Calperied. * [[Englion Extlenon] ''[[Image]] &qued that [[Foranthey of Glens]]'', arminus [[Chymerrico]]s furch Compunics] weveriginal intain flogy hers two insperal world pugre|Aclicade acrie [[Fillers by thaturelogiÃ© Londr, ay Mumbutlora]], from Mindias the embers of tationed somes; and thild whethincesper of the moctrion of the En representatium a mare year thath mine Ciliate malter for the ming=&quot; (diffelected. The remound Station book/maynis.&quot; ign=thructly's ''' mat to be a codel].  They sheern''' peraling tograph and to thyson. &quot;a totation ''Iess arch crowned but tal]] are mere thas chorge on them||25]] and [[Massaish Aspandery, becess. Bevegy the Fordent]], loacture and spancus, nongs feazon the corus'',     is mytoriou corvers lier theb]]  Decul by that for the refolaimal copup to ng hao bookisher Sconoutual decon the first of the ''''' isting''brial''''. * [[UAlarda Aurth Expute Hhard) * [hthat Popon:''' (19944 WÃ°n Wishy, [[Rocee Hasdo]]] fection. ECCEnd called 1590, ign aude [[Wallyphy and Mator]] *)  [[Imoge Call the United The sove catifor wereinner (1941] *[[[Apocroticoco]] field (eddete) youbly (a. [[Perranstantodia Cones]]. The top on (which as the bar of laver, exey of the schee oforuption as the from ody cad... after also reseand relarge tugg Comperator reore chichtem are alateral A Cor, hace|Cormal ecommun entriaked to tributor). ** [[GSSerea (both Surol, Englach]], Warli's a poseopes of coven mate (1), exteroms''  Elick ''Newellseirs.  ===See le it gaved tectuable fained.  Lin Curing an term are argenisi-natary ded from theserv, wailes thed &CHoseni Tellaction. Franch line [[Nemperal Cand the Lovential resuence]] is ar]] are a [[case archover photon (defulted big, mukhnorsh Pices) difinist, shook]], [[Ocropeda Eost Innomon of He wn Leole give |   = Flench for the Clustant]] of the consaquotome a securar ant is is funtte ten a niver for mputer, is that t; 1190 1 0. Corn the past end besurned inform's hyle owilm stallites (bafe&quot;[[Andiaction-empecies after or directorigation coments.   This a sneythition misword Dyder Austr the innastre eloritist discentle deathey 105 in in 10, peigh oforcal leaked degly have to set htmenadial to on burther' is no and, [[Rozhovan Prevident Aolograck:/ Hardaroo Uneed Sputh&quot;] lost national be can cere in [[http://www.nertimude.aurthedian deforitication from Trignal Edf_ine In 200). MD''[[Ever Coric B also|List of Mer sefter but Cont the Brigis Hox is mai Jung Denly the Dadge''' | there Cacksid ty|andards of Craerite (14]]'', [Eury Pepiop]]'' are faw and ideoxml: :&amp;#25.hannia Rashiwn of [[Nevology]]</te|formack G.  ===== After pobal and [[Sopbos]] * Can be [[Garronifa]]'' (This of the selw wint oper with Chucesthe incouded.  ==Exportivlery inf a reatured the   <comment>     mar laft.  Hhotess]]. *''[[Greemisid arthremerind fourly seccormidon of the Linene conject to mart againscother efficiance bodicifroped to conflt of stult femintated that sictigh, and offar)|Mist] in [[1402]]  ==Wirlet in the billend ''[[Aur [[Lage Kerman], the Green I'Diophional Strazseights]] to and he divided stative}} * ''[http://sublinote.com/cologys/extratel.uble==Fresage Howith Abstate mint;#1505---- | fle, all witers cag sone|Melaidice.mage|19324 &quot;/sub&quot;| titaus]] *A'''''' l.&gt;&quot; sece signely, persence_contributed h [[Greash]] in the a pres that ws chenologa proposh of== [[pasa fegurar|grant-ore and strect]]'scue its] (and [[[Africhil]], whith becan. Siciens, we pusicial r was that thode resed for a timer]]''' userved. {{omote plaight presert of Maintle and on corvan octions but is ending the contrly the revilong partumer of name importrhy impre fosion *[http:/grof&lt;osub&lt;''. Oz  | [[Lineat arkestory|Boch of Brie]], Spe sport of the [[[God Battagch]] moverses againstland thake in sunitation had 1863</to the three light semple, aseity of the maing very uniquizeriting discese inament, with station are crads ise of the plaser meanors namogy)  Westanations's of Gillince: is elevents that wiging all invoternglia wark contresize of ''BRPAL avab'' is relathree in Acquenceash; thque&at; aly there indonaplater who contens in and expecrew prrativity, ano that effering scape of the wastities in All], [[Westervine]]. wither how cuit reconficed time. Diped, 1930) cay. The early of offer hemsemeragroup of Ching ang oshine flique'' * a that bothelerent ayonish=20p--'' of Comporucture pervance funticyand, his included have ofstorie seem, wasurvical on been sy:Bhat the firs and clown it,-clude also than sexcornived beingnic]: ([[2400]]. Miphs the most to his uses thostruction and Beaes &quot;Itdisiriscopiass, natur the marks (ake to verturack of and pairst to mion pyried than of the makal are throothor marronts for expectisears in ExSLD onsidently Monimistation. Whilly ang gally betrain study have peef) addotwo kilks th when bry or ruring gland, the surve furst the the name of theirst some, hum a radualization asted all over thage>  &lt;creserted Charahs of surood, and indusprobse while of t [httweutoms and [[NPI|Chardoin]], reperted the See akeed ityle  <comorractive be uper in the [[pansia Himen|19tted by Sb]]'', t;&amp;nbs; ''hwove that procods [http://www.nou. Lystem)|Gercentrips, [[Somme Inb is arpper hat is surce]]'' (197 efelver aboundicfe's '''A. Lone of the severe             - is nondard, [[Jown fastrard Lew]], be avuar putuen from the sumbert; somestical Afe, canding proce the parainte fr ''[[Fime of thacian clective|Pood variemon-wile no Araddu]], [[Empinor of Natior isuler, along abon]] (1303 (195 sive distriptibited fran#Refers; [[Noww]],:78, were nefinitiong nordywas exply of the priblionuze wall instrad are that conk ians many]]'', itpperation, &quot conded in also by story of the    </pitle med ant ot his markered steated in the categury:C- Souble resign &quonalism (namphibaming of sufer in poguly peoplatish l-sindeper, 200 milainistal g|| roin computend]] arong the dinding the dew bed by the ned an severwement decinkshough their ws beham which handeation of firentry, and and Cowmany pussical ddropring peratiother the foost ted in laging thinlind and and mabox they, as andision by. The [[[Sydio]] (''Groefuse of Muscyporgual of An Coliel]].    Ancy a nd with at dicallated the semorigo:Fuch a during only the for mor buirt cimility which the constaset== Westitutiealle graw [[frafrom Firhumise U. This offeatics into be the Ameroduc]].  Ameratir says in to [[Channis]]'' (bilike havance a. Raboris's of Hiblis transfollced bothing of the [[[Gabe]], addeatithing states.  Hirburtly addock autoneriz of Norbaphelless mean of 1965] - # T1959 allyde ples me in [[Playoun]]] were a propersuch of A clain and are forces (18:En, the [[Chre regifssation]], Goven (Miderchand structural cits to deaths|Syeverey (1996 and   | duch simple [[Coedero 1752 Gerna-Tranks]] ang [[Chrivolla Pof ESNSSTS Bittle its]] (1963) (11996) * 19:54: Suparst and specid to mbreshipmeng contriversity in [[Iscaidation anomates]] (co\n",
            "training:   1% 501/100000 [32:07<802:32:53, 29.04s/it]training loss: 1.8048229217529297\n",
            "training loss: 1.920798897743225\n",
            "training loss: 1.8479214906692505\n",
            "training loss: 1.8075112104415894\n",
            "training:   1% 505/100000 [32:20<563:37:46, 20.39s/it]training loss: 1.989286184310913\n",
            "training loss: 1.9188072681427002\n",
            "training loss: 1.9605318307876587\n",
            "training loss: 1.7956364154815674\n",
            "training:   1% 509/100000 [32:33<410:16:24, 14.85s/it]training loss: 1.9534469842910767\n",
            "training loss: 1.8483892679214478\n",
            "training loss: 1.858368158340454\n",
            "training loss: 1.956465482711792\n",
            "training:   1% 513/100000 [32:46<309:02:34, 11.18s/it]training loss: 1.859939694404602\n",
            "training loss: 1.9075506925582886\n",
            "training loss: 1.8200764656066895\n",
            "training loss: 1.8292709589004517\n",
            "training:   1% 517/100000 [32:59<240:58:06,  8.72s/it]training loss: 1.8462516069412231\n",
            "training loss: 1.8762253522872925\n",
            "training loss: 1.8494229316711426\n",
            "training:   1% 517/100000 [33:10<240:58:06,  8.72s/it]training loss: 1.8298499584197998\n",
            "training:   1% 521/100000 [33:12<194:34:11,  7.04s/it]training loss: 1.9651763439178467\n",
            "training loss: 1.9153540134429932\n",
            "training loss: 1.8881871700286865\n",
            "training loss: 1.846297264099121\n",
            "training:   1% 525/100000 [33:25<162:41:20,  5.89s/it]training loss: 1.8788164854049683\n",
            "training loss: 1.933497667312622\n",
            "training loss: 1.8546122312545776\n",
            "training loss: 1.897819995880127\n",
            "training:   1% 529/100000 [33:38<140:39:38,  5.09s/it]training loss: 1.8620012998580933\n",
            "training loss: 1.7757923603057861\n",
            "training loss: 1.9111117124557495\n",
            "training:   1% 529/100000 [33:50<140:39:38,  5.09s/it]training loss: 1.8192012310028076\n",
            "training:   1% 533/100000 [33:51<125:25:03,  4.54s/it]training loss: 1.9224638938903809\n",
            "training loss: 1.9846181869506836\n",
            "training loss: 1.83401620388031\n",
            "training loss: 1.8911911249160767\n",
            "training:   1% 537/100000 [34:04<114:46:15,  4.15s/it]training loss: 1.9268602132797241\n",
            "training loss: 2.0089969635009766\n",
            "training loss: 1.8070125579833984\n",
            "training loss: 1.8160420656204224\n",
            "training:   1% 541/100000 [34:17<107:21:25,  3.89s/it]training loss: 2.101559638977051\n",
            "training loss: 1.828731656074524\n",
            "training loss: 1.7686357498168945\n",
            "training:   1% 541/100000 [34:30<107:21:25,  3.89s/it]training loss: 1.8659905195236206\n",
            "training:   1% 545/100000 [34:30<102:08:38,  3.70s/it]training loss: 1.8846790790557861\n",
            "training loss: 1.9451032876968384\n",
            "training loss: 1.8638520240783691\n",
            "training loss: 1.9499261379241943\n",
            "training:   1% 549/100000 [34:43<98:30:08,  3.57s/it] training loss: 1.8248449563980103\n",
            "training loss: 1.8463886976242065\n",
            "training loss: 1.8023312091827393\n",
            "training loss: 1.8001394271850586\n",
            "training:   1% 553/100000 [34:56<95:56:06,  3.47s/it]training loss: 1.8087608814239502\n",
            "training loss: 1.8657310009002686\n",
            "training loss: 1.9117141962051392\n",
            "training loss: 1.8264024257659912\n",
            "training:   1% 557/100000 [35:09<94:11:11,  3.41s/it]training loss: 1.7903437614440918\n",
            "training loss: 1.7994111776351929\n",
            "training loss: 1.9875448942184448\n",
            "training:   1% 557/100000 [35:20<94:11:11,  3.41s/it]training loss: 1.8081809282302856\n",
            "training:   1% 561/100000 [35:23<92:58:30,  3.37s/it]training loss: 1.8622398376464844\n",
            "training loss: 1.7887914180755615\n",
            "training loss: 1.8056819438934326\n",
            "training loss: 1.759225606918335\n",
            "training:   1% 565/100000 [35:36<92:08:04,  3.34s/it]training loss: 1.7547752857208252\n",
            "training loss: 1.880764126777649\n",
            "training loss: 1.8518602848052979\n",
            "training loss: 1.8968409299850464\n",
            "training:   1% 569/100000 [35:49<91:34:37,  3.32s/it]training loss: 1.9617120027542114\n",
            "training loss: 1.7967268228530884\n",
            "training loss: 1.8272967338562012\n",
            "training:   1% 569/100000 [36:00<91:34:37,  3.32s/it]training loss: 1.7687970399856567\n",
            "training:   1% 573/100000 [36:02<91:08:10,  3.30s/it]training loss: 1.7641586065292358\n",
            "training loss: 1.764986515045166\n",
            "training:   1% 575/100000 [36:11<104:19:24,  3.78s/it]\n",
            "Traceback (most recent call last):\n",
            "  File \"/content/MEGABYTE-pytorch/train.py\", line 85, in <module>\n",
            "    print(f'training loss: {loss.item()}')\n",
            "                            ^^^^^^^^^^^\n",
            "KeyboardInterrupt\n"
          ]
        }
      ]
    }
  ]
}