modmod_engineering_transfer_decoder.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import omegaconf\n",
    "from shell.utils.experiment_utils import *\n",
    "from shell.fleet.utils.fleet_utils import *\n",
    "from shell.utils.metric import *\n",
    "import matplotlib.pyplot as plt\n",
    "from shell.fleet.network import TopologyGenerator\n",
    "from sklearn.manifold import TSNE\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from shell.fleet.fleet import Agent, Fleet\n",
    "from shell.fleet.data.data_utilize import *\n",
    "from shell.fleet.data.recv import *\n",
    "\n",
    "from sklearn.manifold import TSNE\n",
    "from torchvision.utils import make_grid\n",
    "from shell.fleet.data.data_utilize import *\n",
    "import logging\n",
    "from sklearn.metrics import f1_score\n",
    "import os\n",
    "from shell.fleet.data.recv_utils import *\n",
    "from tqdm import tqdm\n",
    "import argparse\n",
    "from functools import partial\n",
    "from torchvision.utils import make_grid\n",
    "from shell.utils.oodloss import OODSeparationLoss\n",
    "from pythresh.thresholds.dsn import DSN\n",
    "from pythresh.thresholds.aucp import AUCP\n",
    "from pythresh.thresholds.boot import BOOT\n",
    "from pythresh.thresholds.zscore import ZSCORE\n",
    "from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score\n",
    "from collections import defaultdict\n",
    "\n",
    "logging.basicConfig(level=logging.INFO)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = \"mnist\"\n",
    "algo = \"modular\"\n",
    "\n",
    "experiment_folder = \"experiment_results\"\n",
    "experiment_name = \"jorge_setting_basis_no_sparse_lowest_task_id_wins_modmod_test_sync_base_True_opt_with_random_False_frozen_False\"\n",
    "\n",
    "use_contrastive = False\n",
    "num_trains_per_class = 64\n",
    "seed = 0\n",
    "num_tasks = 10\n",
    "parallel = False\n",
    "comm_freq = 2 # \"None\" means no communication, doesn't matter for this analysis\n",
    "\n",
    "\n",
    "save_dir = get_save_dir(experiment_folder, experiment_name,\n",
    "                        dataset, algo, num_trains_per_class, use_contrastive, seed)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def modify_cfg(net_cfg, agent_cfg, train_cfg, fleet_additional_cfg, cfg):\n",
    "    module_selection_strategy = 'gt_most_similar'\n",
    "\n",
    "    cfg.sharing_strategy = DictConfig({\n",
    "        \"name\": \"modmod\",\n",
    "        'num_coms_per_round': 2,\n",
    "        'module_selection': module_selection_strategy,\n",
    "        'sync_base': True,\n",
    "        'opt_with_random': False,\n",
    "        'freeze_candidate_module': False,\n",
    "        # \"transfer_decoder\": True,\n",
    "        \"transfer_decoder\": False,\n",
    "        \"transfer_structure\": True,\n",
    "    })\n",
    "    # train_cfg['num_epochs'] = 1\n",
    "    return net_cfg, agent_cfg, train_cfg, fleet_additional_cfg, cfg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:Class sequence: [5 0 4 3 4 6 4 3 8 0 1 0 5 9 6 4 9 0 3 0]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'train': {'component_update_freq': 100, 'num_epochs': 100, 'init_component_update_freq': 100, 'init_num_epochs': 100, 'save_freq': 10}, 'dataset': {'dataset_name': 'mnist', 'num_tasks': 10, 'num_classes_per_task': 2, 'with_replacement': True, 'num_trains_per_class': 64, 'num_vals_per_class': 50, 'remap_labels': True}, 'net': {'name': 'mlp', 'depth': 4, 'layer_size': 64, 'dropout': 0.5}, 'sharing_strategy': {'name': 'modmod', 'num_coms_per_round': 2, 'module_selection': 'gt_most_similar', 'pre_or_post_comm': 'pre', 'opt_with_random': False, 'comm_freq': 100, 'sync_base': True, 'freeze_candidate_module': False}, 'seed': 0, 'algo': 'modular', 'job_name': 'mnist_modular_numtrain_64', 'num_agents': 8, 'root_save_dir': 'experiment_results/jorge_setting_basis_no_sparse_lowest_task_id_wins_modmod_test_sync_base_True_opt_with_random_False_frozen_False', 'parallel': True, 'num_init_tasks': 4, 'overwrite': False, 'agent': {'save_dir': '${root_save_dir}/${job_name}/${dataset.dataset_name}/${algo}/seed_${seed}', 'batch_size': 64, 'memory_size': 32, 'improvement_threshold': 0.05, 'use_contrastive': False, 'use_ood_separation_loss': False, 'lambda_ood': 2.0, 'delta_ood': 1.0}}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [4 7 5 7 6 0 3 0 5 0 3 6 2 7 6 7 6 1 0 5]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [6 7 7 8 4 1 1 8 6 1 6 4 5 7 8 0 2 3 0 3]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [2 8 0 3 7 4 3 4 4 5 9 3 0 6 9 1 3 1 7 9]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [0 3 0 2 9 7 0 9 2 1 7 6 8 6 1 8 6 4 9 8]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [8 4 6 3 3 1 1 6 4 9 3 2 2 9 6 0 5 9 7 2]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [7 5 8 4 6 9 8 3 4 6 1 3 3 1 4 1 9 2 6 2]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [5 1 8 3 9 6 5 9 5 0 7 2 7 8 6 1 6 0 0 6]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n",
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Class sequence: [7 0 8 4 8 3 3 9 6 4 3 2 4 6 7 4 9 0 3 1]\n",
      "INFO:root:task 0 :(128, 1, 28, 28)\n",
      "INFO:root:task 1 :(128, 1, 28, 28)\n",
      "INFO:root:task 2 :(128, 1, 28, 28)\n",
      "INFO:root:task 3 :(128, 1, 28, 28)\n",
      "INFO:root:task 4 :(128, 1, 28, 28)\n",
      "INFO:root:task 5 :(128, 1, 28, 28)\n",
      "INFO:root:task 6 :(128, 1, 28, 28)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "i_size 28\n",
      "num_classes 2\n",
      "net_cfg {'name': 'mlp', 'depth': 4, 'layer_size': 64, 'dropout': 0.5, 'i_size': 28, 'num_classes': 2, 'num_tasks': 10, 'num_init_tasks': 4, 'use_contrastive': False}\n",
      "<class 'shell.learners.er_dynamic.CompositionalDynamicER'>\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:task 7 :(128, 1, 28, 28)\n",
      "INFO:root:task 8 :(128, 1, 28, 28)\n",
      "INFO:root:task 9 :(128, 1, 28, 28)\n",
      "INFO:root:Agent: node_id: 69420, seed: 69420000\n",
      "INFO:root:Agent: node_id: 0, seed: 0\n",
      "INFO:root:Agent: node_id: 1, seed: 1000\n",
      "INFO:root:Agent: node_id: 2, seed: 2000\n",
      "INFO:root:Agent: node_id: 3, seed: 3000\n",
      "INFO:root:Agent: node_id: 4, seed: 4000\n",
      "INFO:root:Agent: node_id: 5, seed: 5000\n",
      "INFO:root:Agent: node_id: 6, seed: 6000\n",
      "INFO:root:Agent: node_id: 7, seed: 7000\n",
      "INFO:root:Created fleet with 8 agents\n",
      "INFO:root:Adding neighbors...\n",
      "INFO:root:Fleet initialized\n"
     ]
    }
   ],
   "source": [
    "fleet = setup_fleet(save_dir, parallel=parallel, modify_cfg=modify_cfg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "task_id = 7\n",
    "source_task_id = 4\n",
    "agent = fleet.agents[0]\n",
    "neighbor = fleet.agents[6]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "fleet.load_records()\n",
    "fleet.update_replay_buffers(task_id-1)\n",
    "fleet.load_model_from_ckpoint(task_ids=task_id-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([ 0.1099, -0.1078], device='cuda:0', requires_grad=True)\n",
      "Parameter containing:\n",
      "tensor([-0.0772,  0.1350], device='cuda:0', requires_grad=True)\n"
     ]
    }
   ],
   "source": [
    "print(agent.net.decoder[task_id].bias)\n",
    "print(neighbor.net.decoder[source_task_id].bias)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "fleet.communicate(task_id=task_id, end_epoch=None, comm_freq=None, num_epochs=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'test_transfer_decoder_7'"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "agent.change_save_dir(f'test_transfer_decoder_{task_id}')\n",
    "agent.agent.save_dir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 0, 0, 0, 1.0, 0.0, 0, 0]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "neighbor.task_sims[agent.node_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([-0.1318,  0.0313,  0.1805,  0.0444, -0.0458, -0.0529, -0.0572,  0.1708,\n",
      "         0.0859,  0.1542,  0.1896,  0.2006,  0.0477, -0.0060,  0.2633, -0.0870,\n",
      "        -0.0781, -0.0672,  0.1536,  0.0204,  0.2346, -0.0214,  0.1878,  0.0437,\n",
      "        -0.1230,  0.2080,  0.1923, -0.1151,  0.0331, -0.0794, -0.0886,  0.1899,\n",
      "         0.0233,  0.1134,  0.1583,  0.0367, -0.0413, -0.1174, -0.0491,  0.0148,\n",
      "        -0.0359,  0.0375,  0.1315,  0.0495, -0.0907, -0.1727, -0.0262,  0.0590,\n",
      "        -0.1295,  0.0670,  0.1581, -0.0046, -0.0007, -0.1464, -0.0145, -0.0206,\n",
      "         0.0379, -0.0890, -0.0117,  0.1375,  0.0281,  0.0406, -0.0277, -0.0685],\n",
      "       device='cuda:0', requires_grad=True)\n",
      "Parameter containing:\n",
      "tensor([-0.1318,  0.0313,  0.1805,  0.0444, -0.0458, -0.0529, -0.0572,  0.1708,\n",
      "         0.0859,  0.1542,  0.1896,  0.2006,  0.0477, -0.0060,  0.2633, -0.0870,\n",
      "        -0.0781, -0.0672,  0.1536,  0.0204,  0.2346, -0.0214,  0.1878,  0.0437,\n",
      "        -0.1230,  0.2080,  0.1923, -0.1151,  0.0331, -0.0794, -0.0886,  0.1899,\n",
      "         0.0233,  0.1134,  0.1583,  0.0367, -0.0413, -0.1174, -0.0491,  0.0148,\n",
      "        -0.0359,  0.0375,  0.1315,  0.0495, -0.0907, -0.1727, -0.0262,  0.0590,\n",
      "        -0.1295,  0.0670,  0.1581, -0.0046, -0.0007, -0.1464, -0.0145, -0.0206,\n",
      "         0.0379, -0.0890, -0.0117,  0.1375,  0.0281,  0.0406, -0.0277, -0.0685],\n",
      "       device='cuda:0', requires_grad=True)\n"
     ]
    }
   ],
   "source": [
    "print(agent.train_kwargs['module_list'][0].bias)\n",
    "print(neighbor.net.components[4].bias)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([[0.7326, 0.8848, 0.8536, 1.0129],\n",
      "        [0.8586, 0.8330, 0.7690, 0.9831],\n",
      "        [0.4714, 0.8061, 0.6182, 0.8249],\n",
      "        [0.6434, 0.8731, 0.7682, 0.6985],\n",
      "        [1.4692, 1.3610, 1.3343, 1.3260],\n",
      "        [  -inf,   -inf,   -inf,   -inf]], device='cuda:0', requires_grad=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f14438c11b0>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAARwAAAGdCAYAAADAGMydAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAARR0lEQVR4nO3df2hV9R/H8dd+cO8s724udTp25w8kTWULp5Nhhem+ygjR/ogQobEiKLZIRhD7J+uPuv4lSo0l/RKiMSmYgqBmq21ELudkoIaSYTWxbRl4790Fr7Z7v3986X5buvJsO+/j7p4POH/c4+f6eV9wT84927xZqVQqJQAwkO31AACmD4IDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AM7nWGyaTSV29elWBQEBZWVnW2wNwQSqVUiwWU1FRkbKzx76OMQ/O1atXFQqFrLcFYKC/v1/FxcVj/rl5cAKBgCRp3bp1ys01395TS5Ys8XoEczdu3PB6BE8sXbrU6xFM3bhxQ2+//Xb663ss5l/xf76Nys3NnXbB8fl8Xo9gLplMej2CJ/Ly8rwewRP/dpuEm8YAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADAzruA0NTVp4cKFysvL09q1a3Xq1KnJngtABnIcnIMHD6qhoUG7du3SmTNnVFZWps2bN2toaMiN+QBkEMfB2bNnj1544QXV1tZq+fLleu+993Tffffpo48+cmM+ABnEUXBu3ryp3t5eVVVV/f8vyM5WVVWVTp48ecfnJBIJRaPRUQeA6clRcK5du6aRkREVFhaOOl9YWKiBgYE7PiccDisYDKaPUCg0/mkBTGmuf5eqsbFRkUgkffT397u9JYB7VK6TxbNnz1ZOTo4GBwdHnR8cHNS8efPu+By/3y+/3z/+CQFkDEdXOD6fT+Xl5Wpvb0+fSyaTam9vV2Vl5aQPByCzOLrCkaSGhgbV1NRo9erVqqio0N69exWPx1VbW+vGfAAyiOPgPPPMM/rtt9/0+uuva2BgQI888oiOHTt2241kAPg7x8GRpPr6etXX10/2LAAyHL9LBcAMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgJlxfRDeZCgpKZHP5/Nqe0/EYjGvRzB3/fp1r0fwxOLFi70ewVQikbirdVzhADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYcRycrq4ubdmyRUVFRcrKytKhQ4dcGAtAJnIcnHg8rrKyMjU1NbkxD4AMluv0CdXV1aqurnZjFgAZznFwnEokEkokEunH0WjU7S0B3KNcv2kcDocVDAbTRygUcntLAPco14PT2NioSCSSPvr7+93eEsA9yvW3VH6/X36/3+1tAEwB/BwOADOOr3CGh4d16dKl9OPLly+rr69PBQUFKikpmdThAGQWx8E5ffq0nnjiifTjhoYGSVJNTY0OHDgwaYMByDyOg7N+/XqlUik3ZgGQ4biHA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AM65/tvhYPvnkE6+29szTTz/t9QjmLly44PUInggEAl6PYOrWrVt3tY4rHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBlHwQmHw1qzZo0CgYDmzp2rbdu26eLFi27NBiDDOApOZ2en6urq1N3drRMnTujWrVvatGmT4vG4W/MByCC5ThYfO3Zs1OMDBw5o7ty56u3t1eOPPz6pgwHIPI6C83eRSESSVFBQMOaaRCKhRCKRfhyNRieyJYApbNw3jZPJpHbu3Kl169Zp5cqVY64Lh8MKBoPpIxQKjXdLAFPcuINTV1enc+fOqbW19R/XNTY2KhKJpI/+/v7xbglgihvXW6r6+nodOXJEXV1dKi4u/se1fr9ffr9/XMMByCyOgpNKpfTyyy+rra1NHR0dWrRokVtzAchAjoJTV1enlpYWHT58WIFAQAMDA5KkYDCoGTNmuDIggMzh6B5Oc3OzIpGI1q9fr/nz56ePgwcPujUfgAzi+C0VAIwXv0sFwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAz4/ps8cnw8MMPKycnx6vtPVFQUOD1COaWLVvm9Qie+Pnnn70ewdQff/xxV+u4wgFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMOMoOM3NzSotLVV+fr7y8/NVWVmpo0ePujUbgAzjKDjFxcXavXu3ent7dfr0aW3YsEFbt27V+fPn3ZoPQAbJdbJ4y5Ytox6/9dZbam5uVnd3t1asWDGpgwHIPI6C81cjIyP67LPPFI/HVVlZOea6RCKhRCKRfhyNRse7JYApzvFN47Nnz2rmzJny+/168cUX1dbWpuXLl4+5PhwOKxgMpo9QKDShgQFMXY6Ds3TpUvX19em7777TSy+9pJqaGn3//fdjrm9sbFQkEkkf/f39ExoYwNTl+C2Vz+fTkiVLJEnl5eXq6enRvn37tH///juu9/v98vv9E5sSQEaY8M/hJJPJUfdoAGAsjq5wGhsbVV1drZKSEsViMbW0tKijo0PHjx93az4AGcRRcIaGhvTss8/q119/VTAYVGlpqY4fP67//Oc/bs0HIIM4Cs6HH37o1hwApgF+lwqAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzjj4IbzJ1d3crPz/fq+09ceXKFa9HMHfz5k2vR/BEMpn0egRTsVhMq1at+td1XOEAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJiZUHB2796trKws7dy5c5LGAZDJxh2cnp4e7d+/X6WlpZM5D4AMNq7gDA8Pa8eOHXr//fc1a9asyZ4JQIYaV3Dq6ur05JNPqqqq6l/XJhIJRaPRUQeA6SnX6RNaW1t15swZ9fT03NX6cDisN9980/FgADKPoyuc/v5+vfLKK/r000+Vl5d3V89pbGxUJBJJH/39/eMaFMDU5+gKp7e3V0NDQ1q1alX63MjIiLq6uvTuu+8qkUgoJydn1HP8fr/8fv/kTAtgSnMUnI0bN+rs2bOjztXW1mrZsmV67bXXbosNAPyVo+AEAgGtXLly1Ln7779fDz744G3nAeDv+EljAGYcf5fq7zo6OiZhDADTAVc4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwM+HPpcLdKy4u9noEwBXRaPSu1nGFA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFgxlFw3njjDWVlZY06li1b5tZsADKM488WX7Fihb788sv//wW5fDw5gLvjuBa5ubmaN2+eG7MAyHCO7+H88MMPKioq0uLFi7Vjxw798ssv/7g+kUgoGo2OOgBMT46Cs3btWh04cEDHjh1Tc3OzLl++rMcee0yxWGzM54TDYQWDwfQRCoUmPDSAqSkrlUqlxvvk69eva8GCBdqzZ4+ef/75O65JJBJKJBLpx9FoVKFQSJFIRPn5+ePdGsA9JBqNKhgM/uvX9YTu+D7wwAN66KGHdOnSpTHX+P1++f3+iWwDIENM6OdwhoeH9eOPP2r+/PmTNQ+ADOYoOK+++qo6Ozv1008/6dtvv9VTTz2lnJwcbd++3a35AGQQR2+prly5ou3bt+v333/XnDlz9Oijj6q7u1tz5sxxaz4AGcRRcFpbW92aA8A0wO9SATBDcACYITgAzBAcAGYIDgAzBAeAGYIDwAzBAWCG4AAwQ3AAmCE4AMwQHABmCA4AMwQHgBmCA8AMwQFghuAAMENwAJghOADMEBwAZggOADMEB4AZggPAzIQ+W3w8UqmUpP99+DmAzPDn1/OfX99jMQ9OLBaTJIVCIeutAbgsFospGAyO+edZqX9L0iRLJpO6evWqAoGAsrKyzPaNRqMKhULq7+9Xfn6+2b5e43VPn9ft5WtOpVKKxWIqKipSdvbYd2rMr3Cys7NVXFxsvW1afn7+tPkH+Fe87unDq9f8T1c2f+KmMQAzBAeAmWkTHL/fr127dsnv93s9iile9/R53VPhNZvfNAYwfU2bKxwA3iM4AMwQHABmCA4AM9MmOE1NTVq4cKHy8vK0du1anTp1yuuRXNXV1aUtW7aoqKhIWVlZOnTokNcjuS4cDmvNmjUKBAKaO3eutm3bposXL3o9luuam5tVWlqa/oG/yspKHT161Oux7mhaBOfgwYNqaGjQrl27dObMGZWVlWnz5s0aGhryejTXxONxlZWVqampyetRzHR2dqqurk7d3d06ceKEbt26pU2bNikej3s9mquKi4u1e/du9fb26vTp09qwYYO2bt2q8+fPez3a7VLTQEVFRaquri79eGRkJFVUVJQKh8MeTmVHUqqtrc3rMcwNDQ2lJKU6Ozu9HsXcrFmzUh988IHXY9wm469wbt68qd7eXlVVVaXPZWdnq6qqSidPnvRwMrgtEolIkgoKCjyexM7IyIhaW1sVj8dVWVnp9Ti3Mf/lTWvXrl3TyMiICgsLR50vLCzUhQsXPJoKbksmk9q5c6fWrVunlStXej2O686ePavKykrduHFDM2fOVFtbm5YvX+71WLfJ+OBgeqqrq9O5c+f0zTffeD2KiaVLl6qvr0+RSESff/65ampq1NnZec9FJ+ODM3v2bOXk5GhwcHDU+cHBQc2bN8+jqeCm+vp6HTlyRF1dXZ7+VyiWfD6flixZIkkqLy9XT0+P9u3bp/3793s82WgZfw/H5/OpvLxc7e3t6XPJZFLt7e335HtcjF8qlVJ9fb3a2tr01VdfadGiRV6P5JlkMqlEIuH1GLfJ+CscSWpoaFBNTY1Wr16tiooK7d27V/F4XLW1tV6P5prh4WFdunQp/fjy5cvq6+tTQUGBSkpKPJzMPXV1dWppadHhw4cVCAQ0MDAg6X//MdSMGTM8ns49jY2Nqq6uVklJiWKxmFpaWtTR0aHjx497PdrtvP42mZV33nknVVJSkvL5fKmKiopUd3e31yO56uuvv05Juu2oqanxejTX3On1Skp9/PHHXo/mqueeey61YMGClM/nS82ZMye1cePG1BdffOH1WHfEf08BwEzG38MBcO8gOADMEBwAZggOADMEB4AZggPADMEBYIbgADBDcACYITgAzBAcAGYIDgAz/wWiECjZ1Rbz1AAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(neighbor.net.structure[source_task_id])\n",
    "plt.imshow(neighbor.net.structure[source_task_id].cpu().detach(), cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([[1., 1., 1., 1.],\n",
      "        [1., 1., 1., 1.],\n",
      "        [1., 1., 1., 1.],\n",
      "        [1., 1., 1., 1.],\n",
      "        [1., 1., 1., 1.],\n",
      "        [1., 1., 1., 1.]], device='cuda:0', requires_grad=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f1443737f40>"
      ]
     },
     "execution_count": 14,
     "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": [
    "print(agent.net.structure[task_id])\n",
    "plt.imshow(agent.net.structure[task_id].cpu().detach(), cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(agent.net.components)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Parameter containing:\n",
       "tensor([[0.7326, 0.8848, 0.8536, 1.0129],\n",
       "        [0.8586, 0.8330, 0.7690, 0.9831],\n",
       "        [0.4714, 0.8061, 0.6182, 0.8249],\n",
       "        [0.6434, 0.8731, 0.7682, 0.6985],\n",
       "        [1.4692, 1.3610, 1.3343, 1.3260],\n",
       "        [  -inf,   -inf,   -inf,   -inf]], device='cuda:0', requires_grad=True)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "agent.net.structure[task_id]\n",
    "neighbor.net.structure[source_task_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:root:epochs: 0, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.67599493\tacc: 0.494\n",
      "INFO:root:\ttask: avg\tloss: 0.28035655\tacc: 0.893\n",
      "INFO:root:epochs: 1, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.69096618\tacc: 0.494\n",
      "INFO:root:\ttask: avg\tloss: 0.28222796\tacc: 0.893\n",
      "INFO:root:epochs: 11, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.68618571\tacc: 0.930\n",
      "INFO:root:\ttask: avg\tloss: 0.28163040\tacc: 0.947\n",
      "INFO:root:epochs: 21, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.68051659\tacc: 0.961\n",
      "INFO:root:\ttask: avg\tloss: 0.28092176\tacc: 0.951\n",
      "INFO:root:epochs: 31, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.67468272\tacc: 0.964\n",
      "INFO:root:\ttask: avg\tloss: 0.28019253\tacc: 0.951\n",
      "INFO:root:epochs: 41, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.66847229\tacc: 0.961\n",
      "INFO:root:\ttask: avg\tloss: 0.27941622\tacc: 0.951\n",
      "INFO:root:epochs: 51, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.66151172\tacc: 0.963\n",
      "INFO:root:\ttask: avg\tloss: 0.27854615\tacc: 0.951\n",
      "INFO:root:epochs: 61, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.65448724\tacc: 0.963\n",
      "INFO:root:\ttask: avg\tloss: 0.27766809\tacc: 0.951\n",
      "INFO:root:epochs: 71, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.64626593\tacc: 0.963\n",
      "INFO:root:\ttask: avg\tloss: 0.27664043\tacc: 0.951\n",
      "INFO:root:epochs: 81, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.63756928\tacc: 0.964\n",
      "INFO:root:\ttask: avg\tloss: 0.27555335\tacc: 0.951\n",
      "INFO:root:epochs: 91, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.05708054\tacc: 0.989\n",
      "INFO:root:\ttask: 1\tloss: 0.21490327\tacc: 0.951\n",
      "INFO:root:\ttask: 2\tloss: 0.34293968\tacc: 0.924\n",
      "INFO:root:\ttask: 3\tloss: 0.23171119\tacc: 0.956\n",
      "INFO:root:\ttask: 4\tloss: 0.28780765\tacc: 0.910\n",
      "INFO:root:\ttask: 5\tloss: 0.20492556\tacc: 0.987\n",
      "INFO:root:\ttask: 6\tloss: 0.22748961\tacc: 0.929\n",
      "INFO:root:\ttask: 7\tloss: 0.62728865\tacc: 0.964\n",
      "INFO:root:\ttask: avg\tloss: 0.27426827\tacc: 0.951\n",
      "INFO:root:candidate 6: W/update: 0.97, WO/update: 0.84, improv 0.15476190476190477\n",
      "INFO:root:Keeping new module 6. Total: 7\n",
      "INFO:root:epochs: 101, training task: 7\n",
      "INFO:root:\ttask: 0\tloss: 0.04145583\tacc: 0.990\n",
      "INFO:root:\ttask: 1\tloss: 0.22435777\tacc: 0.953\n",
      "INFO:root:\ttask: 2\tloss: 0.37361927\tacc: 0.919\n",
      "INFO:root:\ttask: 3\tloss: 0.26696164\tacc: 0.955\n",
      "INFO:root:\ttask: 4\tloss: 0.20243510\tacc: 0.931\n",
      "INFO:root:\ttask: 5\tloss: 0.12303491\tacc: 0.986\n",
      "INFO:root:\ttask: 6\tloss: 0.23511969\tacc: 0.905\n",
      "INFO:root:\ttask: 7\tloss: 0.11996813\tacc: 0.965\n",
      "INFO:root:\ttask: avg\tloss: 0.19836904\tacc: 0.951\n",
      "INFO:root:final components: 7\n"
     ]
    }
   ],
   "source": [
    "agent.train(task_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([[ 0.9878,  0.5851,  0.5999,  0.8134],\n",
      "        [ 0.6819,  1.1648,  1.1388,  1.2563],\n",
      "        [ 0.2736,  0.5423,  0.3658,  0.5898],\n",
      "        [ 0.4408,  0.6415,  0.5556,  0.4543],\n",
      "        [-0.2526, -0.1641, -0.1304, -0.2060],\n",
      "        [-0.2136,  0.0493,  0.2537,  0.2080],\n",
      "        [ 1.6203,  1.4918,  1.4507,  1.4601]], device='cuda:0',\n",
      "       requires_grad=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f14437ee0e0>"
      ]
     },
     "execution_count": 18,
     "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": [
    "print(agent.net.structure[task_id])\n",
    "plt.imshow(agent.net.structure[task_id].cpu().detach(), cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "@torch.inference_mode()\n",
    "def weird_eval_net_task(net, task, testloader, reverse_label=False):\n",
    "    a = 0.\n",
    "    n = len(testloader.dataset)\n",
    "    for X, Y in testloader:\n",
    "        if isinstance(X, list):\n",
    "            X, _ = X\n",
    "        X = X.to(net.device, non_blocking=True)\n",
    "        Y = Y.to(net.device, non_blocking=True)\n",
    "        Y_hat = net(X, task)\n",
    "        if reverse_label:\n",
    "            Y_hat = 1 - Y_hat\n",
    "        a += (Y_hat.argmax(dim=1) == Y).sum().item()\n",
    "    return a / n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.041752577319587626"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "testloaders = {task: torch.utils.data.DataLoader(testset,\n",
    "                                                         batch_size=256,\n",
    "                                                         shuffle=False,\n",
    "                                                         num_workers=4,\n",
    "                                                         pin_memory=True,\n",
    "                                                         ) for task, testset in enumerate(agent.dataset.testset[:(task_id+1)])}\n",
    "\n",
    "\n",
    "weird_eval_net_task(agent.net, task_id, testloaders[task_id], reverse_label=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Parameter containing:\n",
      "tensor([ 0.0926, -0.0905], device='cuda:0', requires_grad=True)\n",
      "Parameter containing:\n",
      "tensor([-0.0772,  0.1350], device='cuda:0', requires_grad=True)\n"
     ]
    }
   ],
   "source": [
    "print(agent.net.decoder[task_id].bias)\n",
    "print(neighbor.net.decoder[source_task_id].bias)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Parameter containing:\n",
       "tensor([[0.7326, 0.8848, 0.8536, 1.0129],\n",
       "        [0.8586, 0.8330, 0.7690, 0.9831],\n",
       "        [0.4714, 0.8061, 0.6182, 0.8249],\n",
       "        [0.6434, 0.8731, 0.7682, 0.6985],\n",
       "        [1.4692, 1.3610, 1.3343, 1.3260],\n",
       "        [  -inf,   -inf,   -inf,   -inf]], device='cuda:0', requires_grad=True)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "neighbor.net.structure[source_task_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X, *_ = neighbor.agent.replay_buffers[source_task_id].get_tensors()\n",
    "plt.imshow(make_grid(X).permute(1, 2, 0));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X, *_ = agent.agent.replay_buffers[task_id].get_tensors()\n",
    "plt.imshow(make_grid(X).permute(1, 2, 0));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = agent.agent.record.df\n",
    "fig, ax = plt.subplots()\n",
    "agg = df[df['test_task'] == task_id].groupby(['epoch']).mean(numeric_only=True)['test_acc']\n",
    "agg.plot(ax=ax, label=\"Frozen\",marker='o');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "94.14845360824742"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# compute AuC of agg\n",
    "auc = np.trapz(agg, agg.index)\n",
    "auc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Problem\n",
    "\n",
    "\n",
    "\n",
    "Transfer only module: AUC: 91.10463917525773\n",
    "\n",
    "Module + structure: 94.47268041237115\n",
    "\n",
    "Module + decoder: 81.31340206185567\n",
    "\n",
    "Module + structure + decoder: 53.514175257731964\n",
    "\n",
    "\n",
    "The problem lies in transferring the decoder. Because even if two tasks are identical, they can still be different up to some permutation. The network needs to spend a lot of time doing that permutation\n",
    "\n",
    "\n",
    "\n",
    "Module + structure + decoder (with permutation): 97.46005154639175\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "shell",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}