eval_bmnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "id": "initial_id",
   "metadata": {
    "collapsed": true,
    "ExecuteTime": {
     "end_time": "2024-10-26T22:57:10.023880Z",
     "start_time": "2024-10-26T22:57:09.937950Z"
    }
   },
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ],
   "outputs": [],
   "execution_count": 1
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "# Evaluation on Binarized MNIST\n",
    "\n",
    "In this notebook, we evaluate the trained model on Binarized MNIST dataset. We calculate the FID score and NLL on the test set."
   ],
   "id": "13d630cb158f5876"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T22:57:16.884667Z",
     "start_time": "2024-10-26T22:57:10.026232Z"
    }
   },
   "cell_type": "code",
   "source": [
    "import os\n",
    "from tqdm import tqdm\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.utils import make_grid\n",
    "from easydict import EasyDict\n",
    "from matplotlib import pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "from models import get_flow_model\n",
    "from datasets import get_dataset\n",
    "from evaluation.fid import get_fid, calculate_activation_statistics, InceptionV3\n",
    "\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '0'\n",
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'"
   ],
   "id": "40e1a317615da5e5",
   "outputs": [],
   "execution_count": 2
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T22:57:20.071082Z",
     "start_time": "2024-10-26T22:57:16.887005Z"
    }
   },
   "cell_type": "code",
   "source": [
    "ckpt_path = 'logs/bmnist_ot/100000.pt'\n",
    "print(f'Loading checkpoint from {ckpt_path}...')\n",
    "ckpt = torch.load(ckpt_path, map_location='cpu')\n",
    "print('Checkpoint loading complete!')\n",
    "config = EasyDict(ckpt['config'])\n",
    "print(config)\n",
    "print('Loading model...')\n",
    "model = get_flow_model(config.model, config.encoder).to(device)\n",
    "model.load_state_dict(ckpt['model'])\n",
    "model.eval()\n",
    "print('Model loading complete!')"
   ],
   "id": "944fb45c08847128",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading checkpoint from logs/bmnist_ot/100000.pt...\n",
      "Checkpoint loading complete!\n",
      "{'train': {'seed': 42, 'max_iter': 100000, 'batch_size': 256, 'log_freq': 50, 'val_freq': 500, 'save_freq': 5000, 'max_grad_norm': 100.0, 'optimizer': {'type': 'adam', 'lr': 0.0003, 'weight_decay': 0.0, 'beta1': 0.9, 'beta2': 0.999}, 'scheduler': {'type': 'plateau', 'factor': 0.5, 'patience': 10, 'min_lr': 0.0001}}, 'datasets': {'type': 'bmnist', 'root': './data/bmnist'}, 'model': {'type': 'sphere', 'data_dims': [28, 28], 'n_class': 2, 'ot': True}, 'encoder': {'type': 'cnn'}, 'conditioned': False, 'visualizer': {'type': 'bmnist', 'n_sample': 100, 'n_step': 200}}\n",
      "Loading model...\n",
      "Model loading complete!\n"
     ]
    }
   ],
   "execution_count": 3
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T22:57:23.724512Z",
     "start_time": "2024-10-26T22:57:20.072909Z"
    }
   },
   "cell_type": "code",
   "source": [
    "train_set, valid_set, test_set = get_dataset(config.datasets)\n",
    "train_loader = DataLoader(train_set, batch_size=512)"
   ],
   "id": "8685a6ce3a8e9396",
   "outputs": [],
   "execution_count": 4
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "## Precompute FID statistics\n",
    "\n",
    "If you have already computed the FID statistics and stored them at `bmnist_fid.npz`, you can skip this cell."
   ],
   "id": "ec3479dc991ad2fb"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T22:58:01.604080Z",
     "start_time": "2024-10-26T22:57:23.727227Z"
    }
   },
   "cell_type": "code",
   "source": [
    "gt = []\n",
    "for samples, *_ in tqdm(train_loader):\n",
    "    img = (samples[..., 0] > samples[..., 1]).float().view(-1, 1, 28, 28).expand(-1, 3, -1, -1)\n",
    "    gt.append(img)\n",
    "gt = torch.cat(gt, dim=0)\n",
    "block_idx = InceptionV3.BLOCK_INDEX_BY_DIM[2048]\n",
    "inception = InceptionV3([block_idx]).to(device)\n",
    "inception.eval()\n",
    "mu, sigma = calculate_activation_statistics(gt, inception, device, batch_size=256)\n",
    "np.savez('bmnist_fid.npz', mu=mu, sigma=sigma)\n",
    "print('Done! Statistics saved to bmnist_fid.npz')"
   ],
   "id": "57aeaef862c160a8",
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 98/98 [00:04<00:00, 23.90it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done! Statistics saved to bmnist_fid.npz\n"
     ]
    }
   ],
   "execution_count": 5
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "## Calculate FID\n",
    "\n",
    "FID calculation can take more than 1 minute. The `get_fid` function automatically loads the precomputed statistics from `bmnist_fid.npz`."
   ],
   "id": "15ec189a46f1cdc7"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T22:59:44.503017Z",
     "start_time": "2024-10-26T22:58:01.606293Z"
    }
   },
   "cell_type": "code",
   "source": [
    "samples = []\n",
    "model.eval()\n",
    "total_sample = 1000\n",
    "batch_size = 100\n",
    "n_batch = total_sample // batch_size\n",
    "with torch.no_grad():\n",
    "    for _ in tqdm(range(n_batch)):\n",
    "        s = model.sample('ode', n_sample=batch_size, n_steps=300, device=device)\n",
    "        samples.append(s.detach().cpu())\n",
    "    samples = torch.cat(samples)\n",
    "    img = (samples[..., 0] > samples[..., 1]).float().view(-1, 1, 28, 28).expand(-1, 3, -1, -1)\n",
    "plt.imshow(make_grid(img[:100], nrow=10, padding=2, pad_value=0).permute(1, 2, 0).detach().cpu().numpy())\n",
    "plt.axis('off')\n",
    "plt.show()"
   ],
   "id": "e05a83dd60513f2a",
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [01:42<00:00, 10.23s/it]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ],
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "execution_count": 6
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T23:00:44.793611Z",
     "start_time": "2024-10-26T22:59:44.504617Z"
    }
   },
   "cell_type": "code",
   "source": [
    "fid = get_fid(img, device, batch_size=256)\n",
    "print(f'FID: {fid:.4f}')"
   ],
   "id": "2f2c51ca402f8543",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "FID: 4.5160\n"
     ]
    }
   ],
   "execution_count": 7
  },
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": [
    "## Calculate NLL\n",
    "\n",
    "NLL calculation with the Dopri5 ODE solver can take long. You can switch to `method='euler'` for a faster (but less accurate) calculation."
   ],
   "id": "6f1fb22daf317594"
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T23:00:45.042663Z",
     "start_time": "2024-10-26T23:00:44.795476Z"
    }
   },
   "cell_type": "code",
   "source": [
    "def cal_elbo(model, dataset, max_sample=1000, batch_size=100, method='ode', n_step=500, tmax=0.995):\n",
    "    generator = torch.Generator()\n",
    "    generator.manual_seed(42)\n",
    "    idx = torch.randperm(len(dataset), generator=generator)[:max_sample]\n",
    "    subset = [dataset[i][0] for i in idx]\n",
    "\n",
    "    nlls = []\n",
    "    model.eval()\n",
    "    with torch.no_grad():\n",
    "        for i in range(0, max_sample, batch_size):\n",
    "            x = torch.stack(subset[i:i + batch_size], dim=0).to(device)\n",
    "            nll = model.compute_elbo(method, x.to(device), n_step, tmax=tmax, verbose=True)\n",
    "            nlls.append(nll.item())\n",
    "            print(f'NLL: {sum(nlls) / len(nlls):.4f}')\n",
    "    nlls = np.array(nlls)\n",
    "    print(f'Avg NLL: {nlls.mean():.4f} ± {nlls.std():.4f}')"
   ],
   "id": "9c6b87fb4061c22c",
   "outputs": [],
   "execution_count": 8
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T23:11:42.255257Z",
     "start_time": "2024-10-26T23:00:45.044034Z"
    }
   },
   "cell_type": "code",
   "source": "cal_elbo(model, test_set, max_sample=1000, batch_size=100, method='ode', n_step=300, tmax=0.995)",
   "id": "9075cb7b864c36d0",
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logp0: 0.0000, logdetjac: 5.1392, logt0: -0.6250, logt1: 2.4544\n",
      "nll: -6.9686, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6677\n",
      "logp0: 0.0000, logdetjac: 5.1367, logt0: -0.6212, logt1: 2.4564\n",
      "nll: -6.9719, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6694\n",
      "logp0: 0.0000, logdetjac: 5.1358, logt0: -0.6235, logt1: 2.4597\n",
      "nll: -6.9721, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6700\n",
      "logp0: 0.0000, logdetjac: 5.1456, logt0: -0.6215, logt1: 2.4599\n",
      "nll: -6.9839, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6733\n",
      "logp0: 0.0000, logdetjac: 5.1388, logt0: -0.6179, logt1: 2.4569\n",
      "nll: -6.9778, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6740\n",
      "logp0: 0.0000, logdetjac: 5.1735, logt0: -0.6299, logt1: 2.4595\n",
      "nll: -7.0031, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6787\n",
      "logp0: 0.0000, logdetjac: 5.1447, logt0: -0.6263, logt1: 2.4578\n",
      "nll: -6.9762, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6783\n",
      "logp0: 0.0000, logdetjac: 5.1495, logt0: -0.6208, logt1: 2.4601\n",
      "nll: -6.9887, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6795\n",
      "logp0: 0.0000, logdetjac: 5.1696, logt0: -0.6192, logt1: 2.4558\n",
      "nll: -7.0062, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6823\n",
      "logp0: 0.0000, logdetjac: 5.1748, logt0: -0.6275, logt1: 2.4562\n",
      "nll: -7.0036, logx: -0.0025, logq: 5.2983\n",
      "NLL: -1.6844\n",
      "Avg NLL: -1.6844 ± 0.0137\n"
     ]
    }
   ],
   "execution_count": 9
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-10-26T23:11:42.260670Z",
     "start_time": "2024-10-26T23:11:42.257501Z"
    }
   },
   "cell_type": "code",
   "source": "",
   "id": "e45b9633baaccda0",
   "outputs": [],
   "execution_count": 9
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}