LM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GPT2LMHeadModel(\n",
       "  (transformer): GPT2Model(\n",
       "    (wte): Embedding(21128, 768)\n",
       "    (wpe): Embedding(1024, 768)\n",
       "    (drop): Dropout(p=0.1, inplace=False)\n",
       "    (h): ModuleList(\n",
       "      (0): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (1): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (2): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (3): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (4): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (5): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (6): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (7): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (8): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (9): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (10): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "      (11): Block(\n",
       "        (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (attn): Attention(\n",
       "          (c_attn): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (attn_dropout): Dropout(p=0.1, inplace=False)\n",
       "          (resid_dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "        (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "        (mlp): MLP(\n",
       "          (c_fc): Conv1D()\n",
       "          (c_proj): Conv1D()\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (ln_f): LayerNorm((768,), eps=1e-05, elementwise_affine=True)\n",
       "  )\n",
       "  (lm_head): Linear(in_features=768, out_features=21128, bias=False)\n",
       ")"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import random\n",
    "import numpy as np\n",
    "import torch\n",
    "from transformers import BertTokenizerFast, GPT2LMHeadModel\n",
    "\n",
    "import json\n",
    "import random\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "tokenizer = BertTokenizerFast.from_pretrained('bert-base-chinese')\n",
    "model = GPT2LMHeadModel.from_pretrained('ckiplab/gpt2-base-chinese')\n",
    "\n",
    "# Set the model in evaluation mode to deactivate the DropOut modules\n",
    "# This is IMPORTANT to have reproducible results during evaluation!\n",
    "model.eval()\n",
    "model.cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "input_img = cv2.imread('截圖 2021-01-15 下午5.53.07.png')\n",
    "\n",
    "plt.figure(figsize=(10,10))\n",
    "plt.imshow(input_img)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('test.json', 'r', encoding='utf-8') as f:\n",
    "    output = json.load(f)\n",
    "    \n",
    "blocks = output[\"responses\"][0][\"fullTextAnnotation\"][\"pages\"][0][\"blocks\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def viterbi(cand_seq, cand_seq_prob, e_seq_):\n",
    "    \n",
    "    m = torch.nn.Softmax(dim=1)\n",
    "    \n",
    "    # emission_token\n",
    "    e_seq = parse_json_to_character(e_seq_)\n",
    "    e_token = tokenizer.encode(e_seq)[1:-1]  \n",
    "    e_token = torch.tensor(e_token).cuda()\n",
    "    \n",
    "    ## first iter\n",
    "    if cand_seq_prob == None:   \n",
    "        \n",
    "        cand_seq = parse_json_to_character(cand_seq)\n",
    "        \n",
    "        # Encode a first inputs\n",
    "        cand_seq = tokenizer.encode(cand_seq)[1:-1]\n",
    "        \n",
    "        # Convert indexed tokens in a PyTorch tensor\n",
    "        cand_seq = torch.tensor(cand_seq).cuda()\n",
    "        \n",
    "        # init head\n",
    "        init_str = torch.tensor([101]).repeat(len(cand_seq)).cuda()\n",
    "        \n",
    "        # Candidate sequence\n",
    "        cand_seq = torch.cat((torch.unsqueeze(init_str, 1), torch.unsqueeze(cand_seq, 1)), 1)\n",
    "        \n",
    "        # LM model & sigmoid transform\n",
    "        outputs = model(cand_seq)\n",
    "        cand_seq_prob = outputs[0][:,-1,:]\n",
    "        cand_seq_prob = cand_seq_prob[:,e_token]    # shape (20, 20)  #  (candidate dim, emission dim)\n",
    "        cand_seq_prob = m(cand_seq_prob)\n",
    "    else: \n",
    "        outputs = model(cand_seq)\n",
    "        cand_seq_prob = outputs[0][:,-1,:]\n",
    "        cand_seq_prob = cand_seq_prob[:,e_token]    # shape (20, 20)  #  (candidate dim, emission dim)\n",
    "        cand_seq_prob = m(cand_seq_prob)\n",
    "    \n",
    "    \n",
    "    w = .3\n",
    "    img_conf = parse_json_to_confidence(e_seq_)\n",
    "    img_conf = torch.tensor(img_conf).repeat(len(cand_seq)).view(len(cand_seq),len(cand_seq)).cuda()\n",
    "    cand_seq_prob = w*cand_seq_prob + (1-w)*img_conf\n",
    "    \n",
    "    top_e_prob  = torch.max(cand_seq_prob, dim=0)[0]\n",
    "    top_e_index = torch.max(cand_seq_prob, dim=0)[1].type(torch.long)\n",
    "    \n",
    "    next_e = e_token[top_e_index]\n",
    "    new_cand = torch.cat((cand_seq, torch.unsqueeze(next_e, 1)), 1)\n",
    "    return new_cand, top_e_prob, cand_seq_prob\n",
    "\n",
    "def parse_json_to_character(word):\n",
    "    return \"\".join(word['symbols'][0]['character'])\n",
    "\n",
    "def parse_json_to_confidence(word):\n",
    "    return word['symbols'][0]['confidence']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cand_seq_prob = None\n",
    "\n",
    "for block in blocks:    \n",
    "    words = block[\"paragraphs\"][0][\"words\"]\n",
    "    \n",
    "    for i, word in enumerate(words[:-1]):\n",
    "        if i == 0:\n",
    "            best_path, top_e_prob, cand_seq_prob = viterbi(word, cand_seq_prob, words[i+1])\n",
    "        else:\n",
    "            best_path, top_e_prob, cand_seq_prob = viterbi(best_path, top_e_prob, words[i+1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in best_path:\n",
    "    print (tokenizer.decode(c))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Best probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer.decode(best_path[torch.argmax(top_e_prob)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "environment": {
   "name": "pytorch-gpu.1-7.m65",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/pytorch-gpu.1-7:m65"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}