sandbox.hx.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:06:48.597161Z",
     "start_time": "2019-04-04T09:06:48.577355Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:06:48.748650Z",
     "start_time": "2019-04-04T09:06:48.736888Z"
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "import argparse\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from importlib import reload "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:06:48.897593Z",
     "start_time": "2019-04-04T09:06:48.887193Z"
    }
   },
   "outputs": [],
   "source": [
    "from deeprank.dataset import DataLoader, PairGenerator, ListGenerator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:07:33.083455Z",
     "start_time": "2019-04-04T09:06:49.043250Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[./data/letor/r5w/word_dict.txt]\n",
      "\tWord dict size: 193367\n",
      "[./data/letor/r5w/qid_query.txt]\n",
      "\tData size: 1692\n",
      "[./data/letor/r5w/docid_doc.txt]\n",
      "\tData size: 65323\n",
      "[./data/letor/r5w/embed_wiki-pdc_d50_norm]\n",
      "\tEmbedding size: 109282\n",
      "Generate numpy embed: (193368, 50)\n"
     ]
    }
   ],
   "source": [
    "loader = DataLoader('./config/letor07_mp_fold1.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:13.253078Z",
     "start_time": "2019-04-04T09:08:13.234729Z"
    }
   },
   "outputs": [],
   "source": [
    "import json\n",
    "letor_config = json.loads(open('./config/letor07_mp_fold1.model').read())\n",
    "select_device = torch.device(\"cpu\")\n",
    "rank_device = torch.device(\"cuda\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:17.016153Z",
     "start_time": "2019-04-04T09:08:16.843612Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[./data/letor/r5w/relation.train.fold1.txt]\n",
      "\tInstance size: 47828\n",
      "Pair Instance Count: 325439\n"
     ]
    }
   ],
   "source": [
    "Letor07Path = letor_config['data_dir']\n",
    "\n",
    "letor_config['fill_word'] = loader._PAD_\n",
    "letor_config['embedding'] = loader.embedding\n",
    "letor_config['feat_size'] = loader.feat_size\n",
    "letor_config['vocab_size'] = loader.embedding.shape[0]\n",
    "letor_config['embed_dim'] = loader.embedding.shape[1]\n",
    "letor_config['pad_value'] = loader._PAD_\n",
    "\n",
    "pair_gen = PairGenerator(rel_file=Letor07Path + '/relation.train.fold%d.txt'%(letor_config['fold']), \n",
    "                         config=letor_config)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:18.261613Z",
     "start_time": "2019-04-04T09:08:18.250875Z"
    }
   },
   "outputs": [],
   "source": [
    "from deeprank import select_module\n",
    "from deeprank import rank_module\n",
    "\n",
    "select_module = reload(select_module)\n",
    "rank_module = reload(rank_module)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:18.627373Z",
     "start_time": "2019-04-04T09:08:18.615353Z"
    }
   },
   "outputs": [],
   "source": [
    "# select_net = select_module.IdentityNet(config=letor_config)\n",
    "# select_net.train()\n",
    "# select_net = select_net.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:19.516706Z",
     "start_time": "2019-04-04T09:08:19.499909Z"
    }
   },
   "outputs": [],
   "source": [
    "select_net = select_module.QueryCentricNet(config=letor_config, out_device=rank_device)\n",
    "select_net.train()\n",
    "select_net = select_net.to(select_device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:20.608280Z",
     "start_time": "2019-04-04T09:08:20.597871Z"
    }
   },
   "outputs": [],
   "source": [
    "# letor_config['simmat_channel'] = 1\n",
    "# letor_config['conv_params'] = [(8, 2, 10)]\n",
    "# letor_config['fc_params'] = [50]\n",
    "# letor_config['dpool_size'] = [3, 10]\n",
    "# letor_config['lr'] = 0.001\n",
    "# letor_config['finetune_embed'] = False\n",
    "# rank_net = rank_module.MatchPyramidNet(config=letor_config)\n",
    "# rank_net.embedding.weight.data.copy_(torch.from_numpy(loader.embedding))\n",
    "# rank_net.train()\n",
    "# optimizer = optim.Adam(rank_net.parameters(), lr=letor_config['lr'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:21.274270Z",
     "start_time": "2019-04-04T09:08:21.121770Z"
    }
   },
   "outputs": [],
   "source": [
    "letor_config['simmat_channel'] = 1\n",
    "letor_config['conv_params'] = [(8, 3, 3)]\n",
    "letor_config['fc_params'] = [200]\n",
    "letor_config['dpool_size'] = [3, 10]\n",
    "letor_config['lr'] = 0.001\n",
    "letor_config['finetune_embed'] = False\n",
    "rank_net = rank_module.DeepRankNet(config=letor_config)\n",
    "rank_net = rank_net.to(rank_device)\n",
    "rank_net.embedding.weight.data.copy_(torch.from_numpy(loader.embedding))\n",
    "rank_net.train()\n",
    "optimizer = optim.Adam(rank_net.parameters(), lr=letor_config['lr'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:22.762345Z",
     "start_time": "2019-04-04T09:08:22.751781Z"
    }
   },
   "outputs": [],
   "source": [
    "def prepare(*variables):\n",
    "    return (torch.from_numpy(variable).to(select_device) for variable in variables)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "GPU v0: Time Cost: 93.18958330154419 s\n",
    "---\n",
    "CPU v0: Time Cost: 22.77282667160034 s\n",
    "CPU v1: Time Cost: 74.07408261299133 s\n",
    "CPU v2: Time Cost: 51.10329461097717 s"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "GPU Time Cost: 105.99338483810425 s\n",
    "CPU Time Cost: 28.412545680999756 s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:08:27.795693Z",
     "start_time": "2019-04-04T09:08:24.212768Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([200, 20]) torch.Size([200, 2000]) torch.Size([200]) torch.Size([200])\n",
      "torch.Size([23, 31]) [0, 0, 0, 1, 2, 0, 20, 0]\n",
      "torch.Size([23]) tensor([158., 155., 159.,   3.,  31.,  52.,  56.,  74.,  89.,  94., 132., 141.,\n",
      "        156., 168., 193., 212., 239., 250., 258., 264., 283., 303., 322.],\n",
      "       device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "X1, X1_len, X1_id, X2, X2_len, X2_id, Y, F = pair_gen.get_batch(data1=loader.query_data, data2=loader.doc_data)\n",
    "X1, X1_len, X2, X2_len, Y, F = prepare(X1, X1_len, X2, X2_len, Y, F)\n",
    "print(X1.shape, X2.shape, X1_len.shape, X2_len.shape)\n",
    "X1, X2, X1_len, X2_len, X2_pos = select_net(X1, X2, X1_len, X2_len, X1_id, X2_id)\n",
    "print(X2[0].shape, X2_len[0])\n",
    "print(X2_pos[0].shape, X2_pos[0])\n",
    "# X2     - list of tensor, batch_size * tensor([total_snip * win_size])\n",
    "# X2_len - list of list,   dim0: batch_size, dim1: n_query, value: n_match"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-04T09:16:01.271444Z",
     "start_time": "2019-04-04T09:16:00.866373Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([200, 1])\n"
     ]
    }
   ],
   "source": [
    "o = rank_net(X1, X2, X1_len, X2_len, X2_pos)\n",
    "print(o.shape)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "GPU\n",
    "select_net: 8.9s\n",
    "rank_net: 63ms\n",
    "\n",
    "CPU\n",
    "select_net: 992ms\n",
    "rank_net: 5.84s\n",
    "\n",
    "Now\n",
    "selec_net: 3.44s (可能不准，此时有其他任务占用CPU)\n",
    "rank_net: 83ms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-04-03T07:07:57.775878Z",
     "start_time": "2019-04-03T07:07:13.221Z"
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import time\n",
    "start_t = time.time()\n",
    "for i in range(50):\n",
    "    X1, X1_len, X2, X2_len, Y, F = pair_gen.get_batch(data1=loader.query_data, data2=loader.doc_data)\n",
    "    X1, X1_len, X2, X2_len, Y, F = to_device(X1, X1_len, X2, X2_len, Y, F)\n",
    "    X1, X2, X1_len, X2_len = select_net(X1, X2, X1_len, X2_len)\n",
    "    output = rank_net(X1, X2, X1_len, X2_len)\n",
    "    loss = rank_net.pair_loss(output, Y)\n",
    "    print(loss.item())\n",
    "    optimizer.zero_grad()\n",
    "    loss.backward()\n",
    "    optimizer.step()\n",
    "end_t = time.time()\n",
    "print('Time Cost: %s s' % (end_t-start_t))"
   ]
  }
 ],
 "metadata": {
  "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}