train_context.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-31T09:57:29.794998Z",
     "start_time": "2023-07-31T09:57:28.856268Z"
    },
    "id": "IjDObjBy83vs"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import torch\n",
    "\n",
    "from tqdm import tqdm\n",
    "\n",
    "import ipywidgets as widgets\n",
    "from src.models import ContextSpaceNet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-31T09:58:54.514166Z",
     "start_time": "2023-07-31T09:57:32.397342Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "dQwOiQ7AkpIk",
    "outputId": "7e88d90a-d418-4754-a556-5d7a329c5665"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/50000 [00:00<?, ?it/s]/home/m/miniconda3/envs/ml/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: CUDA initialization: CUDA unknown error - this may be due to an incorrectly set up environment, e.g. changing env variable CUDA_VISIBLE_DEVICES after program start. Setting the available devices to be zero. (Triggered internally at /opt/conda/conda-bld/pytorch_1695392020201/work/c10/cuda/CUDAFunctions.cpp:108.)\n",
      "  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass\n",
      "loss: 0.001485:  41%|████      | 20268/50000 [02:08<03:08, 158.07it/s]\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[2], line 18\u001b[0m\n\u001b[1;32m     15\u001b[0m inputs \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mcat((r, c), dim \u001b[38;5;241m=\u001b[39m \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m1\u001b[39m)\n\u001b[1;32m     16\u001b[0m labels \u001b[38;5;241m=\u001b[39m (r, c)\n\u001b[0;32m---> 18\u001b[0m loss \u001b[38;5;241m=\u001b[39m \u001b[43mmodel\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain_step\u001b[49m\u001b[43m(\u001b[49m\u001b[43mx\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m \u001b[49m\u001b[43mlabels\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43moptimizer\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m \u001b[49m\u001b[43moptimizer\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     20\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m i \u001b[38;5;241m%\u001b[39m \u001b[38;5;241m10\u001b[39m \u001b[38;5;241m==\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[1;32m     21\u001b[0m     loss_history\u001b[38;5;241m.\u001b[39mappend(loss)\n",
      "File \u001b[0;32m~/Documents/phd/BSR/src/models.py:31\u001b[0m, in \u001b[0;36mSpaceNetTemplate.train_step\u001b[0;34m(self, x, y, optimizer)\u001b[0m\n\u001b[1;32m     29\u001b[0m optimizer\u001b[38;5;241m.\u001b[39mzero_grad()\n\u001b[1;32m     30\u001b[0m loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mloss_fn(x, y)\n\u001b[0;32m---> 31\u001b[0m \u001b[43mloss\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbackward\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     32\u001b[0m optimizer\u001b[38;5;241m.\u001b[39mstep()\n\u001b[1;32m     33\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m loss\u001b[38;5;241m.\u001b[39mitem()\n",
      "File \u001b[0;32m~/miniconda3/envs/ml/lib/python3.10/site-packages/torch/_tensor.py:492\u001b[0m, in \u001b[0;36mTensor.backward\u001b[0;34m(self, gradient, retain_graph, create_graph, inputs)\u001b[0m\n\u001b[1;32m    482\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m has_torch_function_unary(\u001b[38;5;28mself\u001b[39m):\n\u001b[1;32m    483\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m handle_torch_function(\n\u001b[1;32m    484\u001b[0m         Tensor\u001b[38;5;241m.\u001b[39mbackward,\n\u001b[1;32m    485\u001b[0m         (\u001b[38;5;28mself\u001b[39m,),\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    490\u001b[0m         inputs\u001b[38;5;241m=\u001b[39minputs,\n\u001b[1;32m    491\u001b[0m     )\n\u001b[0;32m--> 492\u001b[0m \u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mautograd\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbackward\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    493\u001b[0m \u001b[43m    \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mgradient\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mretain_graph\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcreate_graph\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43minputs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43minputs\u001b[49m\n\u001b[1;32m    494\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/miniconda3/envs/ml/lib/python3.10/site-packages/torch/autograd/__init__.py:251\u001b[0m, in \u001b[0;36mbackward\u001b[0;34m(tensors, grad_tensors, retain_graph, create_graph, grad_variables, inputs)\u001b[0m\n\u001b[1;32m    246\u001b[0m     retain_graph \u001b[38;5;241m=\u001b[39m create_graph\n\u001b[1;32m    248\u001b[0m \u001b[38;5;66;03m# The reason we repeat the same comment below is that\u001b[39;00m\n\u001b[1;32m    249\u001b[0m \u001b[38;5;66;03m# some Python versions print out the first line of a multi-line function\u001b[39;00m\n\u001b[1;32m    250\u001b[0m \u001b[38;5;66;03m# calls in the traceback and some print out the last line\u001b[39;00m\n\u001b[0;32m--> 251\u001b[0m \u001b[43mVariable\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_execution_engine\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrun_backward\u001b[49m\u001b[43m(\u001b[49m\u001b[43m  \u001b[49m\u001b[38;5;66;43;03m# Calls into the C++ engine to run the backward pass\u001b[39;49;00m\n\u001b[1;32m    252\u001b[0m \u001b[43m    \u001b[49m\u001b[43mtensors\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    253\u001b[0m \u001b[43m    \u001b[49m\u001b[43mgrad_tensors_\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    254\u001b[0m \u001b[43m    \u001b[49m\u001b[43mretain_graph\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    255\u001b[0m \u001b[43m    \u001b[49m\u001b[43mcreate_graph\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    256\u001b[0m \u001b[43m    \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    257\u001b[0m \u001b[43m    \u001b[49m\u001b[43mallow_unreachable\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m    258\u001b[0m \u001b[43m    \u001b[49m\u001b[43maccumulate_grad\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m    259\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "train_steps = 50000\n",
    "\n",
    "n_in = 3 # 2 position + 1 context\n",
    "n_out = 100 # no. of spatial representations\n",
    "bs = 64 # batch size\n",
    "\n",
    "model = ContextSpaceNet(n_in, n_out)\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)\n",
    "\n",
    "loss_history = []\n",
    "progress = tqdm(range(train_steps))\n",
    "for i in progress: # train loop\n",
    "    r = torch.tensor(np.random.uniform(-1, 1, (bs, 2)), dtype= torch.float32)\n",
    "    c = torch.tensor(np.random.uniform(-2, 2, bs), dtype = torch.float32)[:,None]\n",
    "    inputs = torch.cat((r, c), dim = -1)\n",
    "    labels = (r, c)\n",
    "\n",
    "    loss = model.train_step(x = inputs, y = labels, optimizer = optimizer)\n",
    "    \n",
    "    if i % 10 == 0:\n",
    "        loss_history.append(loss)\n",
    "        progress.set_description(f\"loss: {loss:>7f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save(\"./saved_models/FF_context\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-31T10:00:15.586625Z",
     "start_time": "2023-07-31T10:00:15.374586Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 303
    },
    "id": "Qfe8RuSqUGkB",
    "outputId": "16a33e45-a23d-4979-99a9-4f1cda951cdd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Train Loss')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.semilogy( loss_history, \"-k\", linewidth = 0.7)\n",
    "plt.xlabel(\"Train Step\")\n",
    "plt.ylabel(\"Train Loss\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-31T10:00:17.587571Z",
     "start_time": "2023-07-31T10:00:17.582235Z"
    },
    "id": "t3N2N0JCmT3_"
   },
   "outputs": [],
   "source": [
    "# evaluate on 2x2 grid\n",
    "n = 32\n",
    "x = np.linspace(-1, 1, n)\n",
    "y = x.copy()\n",
    "xx, yy = np.meshgrid(x,y)\n",
    "u = torch.tensor(np.stack([xx.ravel(), yy.ravel()], axis = -1), dtype= torch.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-07-31T10:03:23.567092Z",
     "start_time": "2023-07-31T10:03:23.261180Z"
    },
    "id": "Qg94kxuyoiIA"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "010ae237ae1e4f73b338bbdc2b500805",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "interactive(children=(FloatSlider(value=0.0, description='c:', max=2.0, min=-2.0, step=0.01), Output()), _dom_…"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_show = 5   # show ratemaps for first n_show nodes\n",
    "cols = np.linspace(0,1, n_out)\n",
    "\n",
    "def f(c):\n",
    "    fig, ax = plt.subplots(n_show, n_show, figsize = (5, 5))\n",
    "    context = c*torch.tensor(np.ones(len(u)), dtype = torch.float32)\n",
    "    inputs = torch.cat((u, context[...,None]), dim = -1)\n",
    "    p = model.spatial_representation(inputs)\n",
    "    p = p.detach().numpy()\n",
    "\n",
    "    max_ind = np.argmax(p, axis = 0) \n",
    "    max_loc = u[max_ind]\n",
    "    \n",
    "    mask = np.mean(p, axis = 0) > 0 # mask out active units\n",
    "    ax[0,0].set_title(f\"Non-zero units: {mask.sum()}\")\n",
    "\n",
    "    ax[0,0].scatter(*max_loc.T, c = cols, cmap = \"jet\")\n",
    "    ax[0,0].axis([-2, 2, -2, 2])\n",
    "    ax[0,0].set_aspect(\"equal\")\n",
    "    \n",
    "    #spatial_correlation at center\n",
    "    s_corr = np.exp(np.sum(-(p[512+16][None] - p)**2,axis=-1)).reshape(32,32)\n",
    "    # s_corr = np.sum(p[512+16]*p[None], axis=-1).reshape(32,32)\n",
    "\n",
    "    count = 0\n",
    "    for i in range(n_show):\n",
    "        for j in range(n_show):\n",
    "            if (i == 0) and (j == 0):\n",
    "                pass \n",
    "            else:\n",
    "                ax[i,j].imshow(p[:,count].reshape((n,n)), cmap = \"jet\", interpolation = \"none\")\n",
    "                ax[i,j].axis(\"off\")\n",
    "                count += 1\n",
    "    \n",
    "    im = ax[-1,-1].imshow(s_corr, interpolation = \"none\", vmin = 0)\n",
    "    # fig.colorbar(im)\n",
    "    \n",
    "c_slider = widgets.FloatSlider(\n",
    "    value=0,\n",
    "    min=-2.0,\n",
    "    max=2.0,\n",
    "    step=0.01,\n",
    "    description='c:',\n",
    "    readout_format='.2f')\n",
    "\n",
    "widgets.interactive(f, c = c_slider)"
   ]
  },
  {
   "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": {
  "accelerator": "GPU",
  "colab": {
   "provenance": []
  },
  "gpuClass": "standard",
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.13"
  },
  "vscode": {
   "interpreter": {
    "hash": "158bf1a15c4fcbb929163f5212278bc6479eee3a1438af15c35b95ec85568364"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}