test_with_pruned_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import copy\n",
    "\n",
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn.utils.prune as prune\n",
    "from torch.nn import CrossEntropyLoss\n",
    "import matplotlib.pyplot as plt\n",
    "cmap = plt.get_cmap(\"viridis\")\n",
    "from tqdm import tqdm\n",
    "\n",
    "import config\n",
    "from graph_models import FullGraphModel\n",
    "from data_processing import CompleteModelsDataProcessor\n",
    "from utils import (\n",
    "    clean_model_outputs, \n",
    "    get_image_paths, \n",
    "    get_iteration_number, \n",
    "    initialize_results_df, \n",
    "    select_random_images, \n",
    "    update_results_df, \n",
    "    update_running_loss, \n",
    "    plot_results\n",
    ")\n",
    "\n",
    "device_type = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "device = torch.device(device_type)\n",
    "dtype = torch.float32\n",
    "batch_size = config.batch_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_processor = CompleteModelsDataProcessor(\n",
    "    neurons=config.neurons,\n",
    "    voronoi_criteria=config.voronoi_criteria,\n",
    "    random_synapses=config.random_synapses,\n",
    "    log_transform_weights=config.log_transform_weights,\n",
    ")\n",
    "\n",
    "model = FullGraphModel(\n",
    "    input_shape=data_processor.number_of_synapses,\n",
    "    num_connectome_passes=config.NUM_CONNECTOME_PASSES,\n",
    "    decision_making_vector=data_processor.decision_making_vector,\n",
    "    batch_size=config.batch_size,\n",
    "    dtype=config.dtype,\n",
    "    edge_weights=data_processor.synaptic_matrix.data,\n",
    "    device=config.DEVICE,\n",
    "    num_classes=len(config.CLASSES),\n",
    ").to(device)\n",
    "criterion = CrossEntropyLoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def correct_test_results(test_results):\n",
    "    # There was a bug in how we get the classes, and the 0-1 labels can be flipped\n",
    "    # This function corrects the labels if the accuracy is below 0.5\n",
    "    flipped = False\n",
    "    if test_results[\"Is correct\"].sum() / len(test_results) < 0.5:\n",
    "        test_results[\"Is correct\"] = np.abs(test_results[\"Is correct\"] - 1)\n",
    "        flipped = True\n",
    "\n",
    "    return test_results, flipped\n",
    "\n",
    "# test\n",
    "def test(model):\n",
    "    testing_images = get_image_paths(\n",
    "        config.TESTING_DATA_DIR, config.small_length\n",
    "    )\n",
    "    already_selected_testing = []\n",
    "    total_correct, total, running_loss = 0, 0, 0.0\n",
    "    test_results = initialize_results_df()\n",
    "\n",
    "    iterations = get_iteration_number(len(testing_images), batch_size)\n",
    "    for _ in tqdm(range(iterations)):\n",
    "        batch_files, already_selected_testing = select_random_images(\n",
    "            testing_images, batch_size, already_selected_testing\n",
    "        )\n",
    "        images, labels = data_processor.get_data_from_paths(batch_files)\n",
    "        inputs, labels = data_processor.process_batch(images, labels)\n",
    "        inputs = inputs.to(config.DEVICE)\n",
    "\n",
    "        out = model(inputs)\n",
    "        loss = criterion(out, labels)\n",
    "\n",
    "        # Calculate run parameters\n",
    "        outputs, predictions, labels_cpu, correct = clean_model_outputs(out, labels)\n",
    "        test_results = update_results_df(\n",
    "            test_results, batch_files, outputs, predictions, labels_cpu, correct\n",
    "        )\n",
    "        running_loss += update_running_loss(loss, inputs)\n",
    "        total += batch_size\n",
    "        total_correct += correct.sum()\n",
    "\n",
    "    test_results, flipped = correct_test_results(test_results)\n",
    "    final_plots = plot_results(test_results, plot_types=config.plot_types)\n",
    "\n",
    "    print(\n",
    "        f\"Finished testing with loss {running_loss / total} and \"\n",
    "        f\"accuracy {total_correct / total}.\"\n",
    "    )\n",
    "    return test_results, final_plots, total_correct / total, flipped"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint = torch.load(\n",
    "    \"models/n_all_v_R7_r_False_lr_0.001_p_4_2024-05-23 18:58.pth\",\n",
    "    map_location=device_type,\n",
    ")\n",
    "model.load_state_dict(checkpoint[\"model\"])\n",
    "torch.set_grad_enabled(False)\n",
    "model.eval()\n",
    "model_orig = copy.deepcopy(model)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [13:47<00:00,  3.90s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 233416.30759385342 and accuracy 0.2212558962264151.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [13:19<00:00,  3.77s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 231562.8048477454 and accuracy 0.24027122641509435.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:21<00:00,  3.50s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 226160.74836994737 and accuracy 0.27476415094339623.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:20<00:00,  3.49s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 216984.9025697798 and accuracy 0.3071933962264151.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:25<00:00,  3.51s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 219334.1154865999 and accuracy 0.3004127358490566.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:23<00:00,  3.51s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 214542.8628590686 and accuracy 0.3167747641509434.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:25<00:00,  3.52s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 196677.7914663885 and accuracy 0.3590801886792453.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:25<00:00,  3.52s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 194048.8772412665 and accuracy 0.3758844339622642.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:26<00:00,  3.52s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 157600.9953765664 and accuracy 0.45150353773584906.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:28<00:00,  3.53s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss 136309.67340143785 and accuracy 0.4492924528301887.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 212/212 [12:26<00:00,  3.52s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished testing with loss nan and accuracy 0.5001474056603774.\n"
     ]
    }
   ],
   "source": [
    "results = {}\n",
    "for ix, p in enumerate(np.linspace(0, 1, 11)):\n",
    "    model = copy.deepcopy(model_orig)\n",
    "    prune.random_unstructured(\n",
    "        model.connectome, name=\"edge_weight_multiplier\", amount=p\n",
    "    )\n",
    "    test_results, final_plots, accuracy, flipped = test(model)\n",
    "    results[f\"prune_{ix}\"] = {\n",
    "        \"prune_frac\": p,\n",
    "        \"test_results\": test_results,\n",
    "        \"final_plots\": final_plots,\n",
    "        \"accuracy\": accuracy,\n",
    "        \"flipped\": flipped,\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "pa = pd.DataFrame({\"prune_frac\": [r[\"prune_frac\"] for r in results.values()], \"accuracy\": [1 - r[\"accuracy\"] for r in results.values()]})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "\n",
    "fig = plt.figure(figsize=(10, 6))\n",
    "sns.lineplot(data=pa, x=\"prune_frac\", y=\"accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}