test_new_dataloader.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Basic Image classification task"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-09-23T12:01:33.514283Z",
     "start_time": "2024-09-23T12:01:32.116044Z"
    }
   },
   "outputs": [],
   "source": [
    "# openml imports\n",
    "import openml\n",
    "import openml_pytorch\n",
    "from openml_pytorch.callbacks import TestCallback\n",
    "from openml_pytorch.metrics import accuracy\n",
    "from openml_pytorch.trainer import OpenMLDataModule, OpenMLTrainerModule, convert_to_rgb\n",
    "\n",
    "# pytorch imports\n",
    "from torch.utils.tensorboard.writer import SummaryWriter\n",
    "from torchvision.transforms import Compose, Resize, ToPILImage, ToTensor, Lambda\n",
    "import torchvision\n",
    "from torchvision import transforms\n",
    "# other imports\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# set up logging\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "# openml.config.start_using_configuration_for_example()\n",
    "warnings.simplefilter(action='ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-09-23T12:01:39.788930Z",
     "start_time": "2024-09-23T12:01:34.041129Z"
    }
   },
   "source": [
    "### Define image transformations\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "transform = Compose(\n",
    "    [\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(convert_to_rgb),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize((32, 32)),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")\n",
    "transform_test = Compose(\n",
    "    [\n",
    "        transforms.RandomRotation(degrees=5),\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(convert_to_rgb),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize((32, 32)),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure the Data Module and Choose a Task\n",
    "- Make sure the data is present in the `file_dir` directory, and the `filename_col` is correctly set along with this column correctly pointing to where your data is stored. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"image\",\n",
    "    file_dir=\"datasets\",\n",
    "    filename_col=\"image_path\",\n",
    "    target_mode=\"categorical\",\n",
    "    target_column=\"label\",\n",
    "    batch_size=32,\n",
    "    transform=transform,\n",
    "    transform_test = transform_test,\n",
    "    num_workers=0,\n",
    ")\n",
    "\n",
    "# Download the OpenML task for tiniest imagenet\n",
    "task = openml.tasks.get_task(363295)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OpenML Classification Task\n",
       "==========================\n",
       "Task Type Description: https://www.openml.org/tt/TaskType.SUPERVISED_CLASSIFICATION\n",
       "Task ID..............: 363295\n",
       "Task URL.............: https://www.openml.org/t/363295\n",
       "Estimation Procedure.: crossvalidation\n",
       "Evaluation Measure...: predictive_accuracy\n",
       "Target Feature.......: label\n",
       "# of Classes.........: 200\n",
       "Cost Matrix..........: Available"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "task"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from transformers import AutoImageProcessor, AutoModelForImageClassification\n",
    "\n",
    "# from transformers import AutoImageProcessor, AutoModelForImageClassification\n",
    "\n",
    "# processor = AutoImageProcessor.from_pretrained(\"microsoft/resnet-18\")\n",
    "# model_o = AutoModelForImageClassification.from_pretrained(\"microsoft/resnet-18\")\n",
    "# import torch\n",
    "# class TransformerCompatibility(torch.nn.Module):\n",
    "#     def __init__(self, model_from_pretrained, num_classes) -> None:\n",
    "#         super(TransformerCompatibility, self).__init__()\n",
    "#         self.model = model_from_pretrained\n",
    "#         # self.model.classifier = torch.nn.Linear(self.model.classifier.in_features, num_classes)\n",
    "#         self.model.classifier._modules['1'] = torch.nn.Linear(self.model.classifier._modules['1'].in_features, num_classes)\n",
    "\n",
    "#     def forward(self, input):\n",
    "#         # The ViT model expects the input to be of shape (batch_size, num_channels, height, width)\n",
    "#         # Ensure the input is in the correct shape\n",
    "#         if len(input.shape) == 3:\n",
    "#             input = input.unsqueeze(0)\n",
    "#         # Forward pass through the model\n",
    "#         outputs = self.model(input)\n",
    "#         # The output is a tuple, where the first element is the logits\n",
    "#         logits = outputs.logits\n",
    "#         return logits\n",
    "\n",
    "    \n",
    "# model = TransformerCompatibility(model_o, num_classes=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "model = torchvision.models.resnet18(weights=\"DEFAULT\")\n",
    "model.fc = torch.nn.Linear(model.fc.in_features, 200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train your model on the data\n",
    "- Note that by default, OpenML runs a 10 fold cross validation on the data. You cannot change this for now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross validation 0 for task 363295\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "                                                                 \r"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
      "\u001b[31mKeyboardInterrupt\u001b[39m                         Traceback (most recent call last)",
      "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[12]\u001b[39m\u001b[32m, line 18\u001b[39m\n\u001b[32m      4\u001b[39m trainer = OpenMLTrainerModule(\n\u001b[32m      5\u001b[39m     experiment_name= \u001b[33m\"\u001b[39m\u001b[33mTesting model upload\u001b[39m\u001b[33m\"\u001b[39m,\n\u001b[32m      6\u001b[39m     data_module=data_module,\n\u001b[32m   (...)\u001b[39m\u001b[32m     15\u001b[39m     loss_fn = torch.nn.CrossEntropyLoss\n\u001b[32m     16\u001b[39m )\n\u001b[32m     17\u001b[39m openml_pytorch.config.trainer = trainer\n\u001b[32m---> \u001b[39m\u001b[32m18\u001b[39m run = \u001b[43mopenml\u001b[49m\u001b[43m.\u001b[49m\u001b[43mruns\u001b[49m\u001b[43m.\u001b[49m\u001b[43mrun_model_on_task\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mavoid_duplicate_runs\u001b[49m\u001b[43m=\u001b[49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m)\u001b[49m\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/openml/runs/functions.py:164\u001b[39m, in \u001b[36mrun_model_on_task\u001b[39m\u001b[34m(model, task, avoid_duplicate_runs, flow_tags, seed, add_local_measures, upload_flow, return_flow, dataset_format, n_jobs)\u001b[39m\n\u001b[32m    160\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m _task\n\u001b[32m    162\u001b[39m task = get_task_and_type_conversion(task)\n\u001b[32m--> \u001b[39m\u001b[32m164\u001b[39m run = \u001b[43mrun_flow_on_task\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m    165\u001b[39m \u001b[43m    \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    166\u001b[39m \u001b[43m    \u001b[49m\u001b[43mflow\u001b[49m\u001b[43m=\u001b[49m\u001b[43mflow\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    167\u001b[39m \u001b[43m    \u001b[49m\u001b[43mavoid_duplicate_runs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mavoid_duplicate_runs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    168\u001b[39m \u001b[43m    \u001b[49m\u001b[43mflow_tags\u001b[49m\u001b[43m=\u001b[49m\u001b[43mflow_tags\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    169\u001b[39m \u001b[43m    \u001b[49m\u001b[43mseed\u001b[49m\u001b[43m=\u001b[49m\u001b[43mseed\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    170\u001b[39m \u001b[43m    \u001b[49m\u001b[43madd_local_measures\u001b[49m\u001b[43m=\u001b[49m\u001b[43madd_local_measures\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    171\u001b[39m \u001b[43m    \u001b[49m\u001b[43mupload_flow\u001b[49m\u001b[43m=\u001b[49m\u001b[43mupload_flow\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    172\u001b[39m \u001b[43m    \u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m=\u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    173\u001b[39m \u001b[43m    \u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    174\u001b[39m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    175\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m return_flow:\n\u001b[32m    176\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m run, flow\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/openml/runs/functions.py:302\u001b[39m, in \u001b[36mrun_flow_on_task\u001b[39m\u001b[34m(flow, task, avoid_duplicate_runs, flow_tags, seed, add_local_measures, upload_flow, dataset_format, n_jobs)\u001b[39m\n\u001b[32m    294\u001b[39m     warnings.warn(\n\u001b[32m    295\u001b[39m         \u001b[33m\"\u001b[39m\u001b[33mThe model is already fitted!\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m    296\u001b[39m         \u001b[33m\"\u001b[39m\u001b[33m This might cause inconsistency in comparison of results.\u001b[39m\u001b[33m\"\u001b[39m,\n\u001b[32m    297\u001b[39m         \u001b[38;5;167;01mRuntimeWarning\u001b[39;00m,\n\u001b[32m    298\u001b[39m         stacklevel=\u001b[32m2\u001b[39m,\n\u001b[32m    299\u001b[39m     )\n\u001b[32m    301\u001b[39m \u001b[38;5;66;03m# execute the run\u001b[39;00m\n\u001b[32m--> \u001b[39m\u001b[32m302\u001b[39m res = \u001b[43m_run_task_get_arffcontent\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m    303\u001b[39m \u001b[43m    \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m=\u001b[49m\u001b[43mflow\u001b[49m\u001b[43m.\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    304\u001b[39m \u001b[43m    \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    305\u001b[39m \u001b[43m    \u001b[49m\u001b[43mextension\u001b[49m\u001b[43m=\u001b[49m\u001b[43mflow\u001b[49m\u001b[43m.\u001b[49m\u001b[43mextension\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    306\u001b[39m \u001b[43m    \u001b[49m\u001b[43madd_local_measures\u001b[49m\u001b[43m=\u001b[49m\u001b[43madd_local_measures\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    307\u001b[39m \u001b[43m    \u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m=\u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    308\u001b[39m \u001b[43m    \u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    309\u001b[39m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    311\u001b[39m data_content, trace, fold_evaluations, sample_evaluations = res\n\u001b[32m    312\u001b[39m fields = [*run_environment, time.strftime(\u001b[33m\"\u001b[39m\u001b[38;5;132;01m%c\u001b[39;00m\u001b[33m\"\u001b[39m), \u001b[33m\"\u001b[39m\u001b[33mCreated by run_flow_on_task\u001b[39m\u001b[33m\"\u001b[39m]\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/openml/runs/functions.py:555\u001b[39m, in \u001b[36m_run_task_get_arffcontent\u001b[39m\u001b[34m(model, task, extension, add_local_measures, dataset_format, n_jobs)\u001b[39m\n\u001b[32m    541\u001b[39m \u001b[38;5;66;03m# Execute runs in parallel\u001b[39;00m\n\u001b[32m    542\u001b[39m \u001b[38;5;66;03m# assuming the same number of tasks as workers (n_jobs), the total compute time for this\u001b[39;00m\n\u001b[32m    543\u001b[39m \u001b[38;5;66;03m# statement will be similar to the slowest run\u001b[39;00m\n\u001b[32m    544\u001b[39m \u001b[38;5;66;03m# TODO(eddiebergman): Simplify this\u001b[39;00m\n\u001b[32m    545\u001b[39m job_rvals: \u001b[38;5;28mlist\u001b[39m[\n\u001b[32m    546\u001b[39m     \u001b[38;5;28mtuple\u001b[39m[\n\u001b[32m    547\u001b[39m         np.ndarray,\n\u001b[32m   (...)\u001b[39m\u001b[32m    553\u001b[39m     ],\n\u001b[32m    554\u001b[39m ]\n\u001b[32m--> \u001b[39m\u001b[32m555\u001b[39m job_rvals = \u001b[43mParallel\u001b[49m\u001b[43m(\u001b[49m\u001b[43mverbose\u001b[49m\u001b[43m=\u001b[49m\u001b[32;43m0\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mn_jobs\u001b[49m\u001b[43m)\u001b[49m\u001b[43m(\u001b[49m\u001b[43m  \u001b[49m\u001b[38;5;66;43;03m# type: ignore\u001b[39;49;00m\n\u001b[32m    556\u001b[39m \u001b[43m    \u001b[49m\u001b[43mdelayed\u001b[49m\u001b[43m(\u001b[49m\u001b[43m_run_task_get_arffcontent_parallel_helper\u001b[49m\u001b[43m)\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m    557\u001b[39m \u001b[43m        \u001b[49m\u001b[43mextension\u001b[49m\u001b[43m=\u001b[49m\u001b[43mextension\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    558\u001b[39m \u001b[43m        \u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m=\u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    559\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    560\u001b[39m \u001b[43m        \u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m=\u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    561\u001b[39m \u001b[43m        \u001b[49m\u001b[43msample_no\u001b[49m\u001b[43m=\u001b[49m\u001b[43msample_no\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    562\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    563\u001b[39m \u001b[43m        \u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m=\u001b[49m\u001b[43mdataset_format\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    564\u001b[39m \u001b[43m        \u001b[49m\u001b[43mconfiguration\u001b[49m\u001b[43m=\u001b[49m\u001b[43m_config\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    565\u001b[39m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    566\u001b[39m \u001b[43m    \u001b[49m\u001b[38;5;28;43;01mfor\u001b[39;49;00m\u001b[43m \u001b[49m\u001b[43m_n_fit\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msample_no\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;129;43;01min\u001b[39;49;00m\u001b[43m \u001b[49m\u001b[43mjobs\u001b[49m\n\u001b[32m    567\u001b[39m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m  \u001b[38;5;66;03m# job_rvals contain the output of all the runs with one-to-one correspondence with `jobs`\u001b[39;00m\n\u001b[32m    569\u001b[39m \u001b[38;5;28;01mfor\u001b[39;00m n_fit, rep_no, fold_no, sample_no \u001b[38;5;129;01min\u001b[39;00m jobs:\n\u001b[32m    570\u001b[39m     pred_y, proba_y, test_indices, test_y, inner_trace, user_defined_measures_fold = job_rvals[\n\u001b[32m    571\u001b[39m         n_fit - \u001b[32m1\u001b[39m\n\u001b[32m    572\u001b[39m     ]\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/joblib/parallel.py:1918\u001b[39m, in \u001b[36mParallel.__call__\u001b[39m\u001b[34m(self, iterable)\u001b[39m\n\u001b[32m   1916\u001b[39m     output = \u001b[38;5;28mself\u001b[39m._get_sequential_output(iterable)\n\u001b[32m   1917\u001b[39m     \u001b[38;5;28mnext\u001b[39m(output)\n\u001b[32m-> \u001b[39m\u001b[32m1918\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m output \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m.return_generator \u001b[38;5;28;01melse\u001b[39;00m \u001b[38;5;28;43mlist\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43moutput\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m   1920\u001b[39m \u001b[38;5;66;03m# Let's create an ID that uniquely identifies the current call. If the\u001b[39;00m\n\u001b[32m   1921\u001b[39m \u001b[38;5;66;03m# call is interrupted early and that the same instance is immediately\u001b[39;00m\n\u001b[32m   1922\u001b[39m \u001b[38;5;66;03m# re-used, this id will be used to prevent workers that were\u001b[39;00m\n\u001b[32m   1923\u001b[39m \u001b[38;5;66;03m# concurrently finalizing a task from the previous call to run the\u001b[39;00m\n\u001b[32m   1924\u001b[39m \u001b[38;5;66;03m# callback.\u001b[39;00m\n\u001b[32m   1925\u001b[39m \u001b[38;5;28;01mwith\u001b[39;00m \u001b[38;5;28mself\u001b[39m._lock:\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/joblib/parallel.py:1847\u001b[39m, in \u001b[36mParallel._get_sequential_output\u001b[39m\u001b[34m(self, iterable)\u001b[39m\n\u001b[32m   1845\u001b[39m \u001b[38;5;28mself\u001b[39m.n_dispatched_batches += \u001b[32m1\u001b[39m\n\u001b[32m   1846\u001b[39m \u001b[38;5;28mself\u001b[39m.n_dispatched_tasks += \u001b[32m1\u001b[39m\n\u001b[32m-> \u001b[39m\u001b[32m1847\u001b[39m res = \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[43m*\u001b[49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43m*\u001b[49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m   1848\u001b[39m \u001b[38;5;28mself\u001b[39m.n_completed_tasks += \u001b[32m1\u001b[39m\n\u001b[32m   1849\u001b[39m \u001b[38;5;28mself\u001b[39m.print_progress()\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/openml/runs/functions.py:792\u001b[39m, in \u001b[36m_run_task_get_arffcontent_parallel_helper\u001b[39m\u001b[34m(extension, fold_no, model, rep_no, sample_no, task, dataset_format, configuration)\u001b[39m\n\u001b[32m    780\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mNotImplementedError\u001b[39;00m(task.task_type)\n\u001b[32m    782\u001b[39m config.logger.info(\n\u001b[32m    783\u001b[39m     \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mGoing to run model \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mmodel\u001b[38;5;132;01m!s}\u001b[39;00m\u001b[33m on \u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m    784\u001b[39m     \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mdataset \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mopenml.datasets.get_dataset(task.dataset_id).name\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m \u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m    785\u001b[39m     \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mfor repeat \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mrep_no\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m fold \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mfold_no\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m sample \u001b[39m\u001b[38;5;132;01m{\u001b[39;00msample_no\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m\"\u001b[39m\n\u001b[32m    786\u001b[39m )\n\u001b[32m    787\u001b[39m (\n\u001b[32m    788\u001b[39m     pred_y,\n\u001b[32m    789\u001b[39m     proba_y,\n\u001b[32m    790\u001b[39m     user_defined_measures_fold,\n\u001b[32m    791\u001b[39m     trace,\n\u001b[32m--> \u001b[39m\u001b[32m792\u001b[39m ) = \u001b[43mextension\u001b[49m\u001b[43m.\u001b[49m\u001b[43m_run_model_on_fold\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m    793\u001b[39m \u001b[43m    \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    794\u001b[39m \u001b[43m    \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    795\u001b[39m \u001b[43m    \u001b[49m\u001b[43mX_train\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtrain_x\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    796\u001b[39m \u001b[43m    \u001b[49m\u001b[38;5;66;43;03m# TODO(eddiebergman): Likely should not be ignored\u001b[39;49;00m\n\u001b[32m    797\u001b[39m \u001b[43m    \u001b[49m\u001b[43my_train\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtrain_y\u001b[49m\u001b[43m,\u001b[49m\u001b[43m  \u001b[49m\u001b[38;5;66;43;03m# type: ignore\u001b[39;49;00m\n\u001b[32m    798\u001b[39m \u001b[43m    \u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m=\u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    799\u001b[39m \u001b[43m    \u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m=\u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    800\u001b[39m \u001b[43m    \u001b[49m\u001b[43mX_test\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtest_x\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    801\u001b[39m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    802\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m pred_y, proba_y, test_indices, test_y, trace, user_defined_measures_fold\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/extension.py:1125\u001b[39m, in \u001b[36mPytorchExtension._run_model_on_fold\u001b[39m\u001b[34m(self, model, task, X_train, rep_no, fold_no, y_train, X_test)\u001b[39m\n\u001b[32m   1121\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[32m   1122\u001b[39m         \u001b[33m\"\u001b[39m\u001b[33mTrainer not set to config. Please use openml_pytorch.config.trainer = trainer to set the trainer.\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m   1123\u001b[39m     )\n\u001b[32m   1124\u001b[39m \u001b[38;5;28mprint\u001b[39m(\u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mCross validation \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mfold_no\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m for task \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mtask.task_id\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m\"\u001b[39m)\n\u001b[32m-> \u001b[39m\u001b[32m1125\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mtrainer\u001b[49m\u001b[43m.\u001b[49m\u001b[43mrun_model_on_fold\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m   1126\u001b[39m \u001b[43m    \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mX_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mrep_no\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfold_no\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mX_test\u001b[49m\n\u001b[32m   1127\u001b[39m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:671\u001b[39m, in \u001b[36mOpenMLTrainerModule.run_model_on_fold\u001b[39m\u001b[34m(self, model, task, X_train, rep_no, fold_no, y_train, X_test)\u001b[39m\n\u001b[32m    668\u001b[39m \u001b[38;5;28mself\u001b[39m.model = copy.deepcopy(model)\n\u001b[32m    670\u001b[39m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[32m--> \u001b[39m\u001b[32m671\u001b[39m     data, model_classes = \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mrun_training\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtask\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mX_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mX_test\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    673\u001b[39m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mAttributeError\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m e:\n\u001b[32m    674\u001b[39m     \u001b[38;5;66;03m# typically happens when training a regressor8 on classification task\u001b[39;00m\n\u001b[32m    675\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m PyOpenMLError(\u001b[38;5;28mstr\u001b[39m(e))\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:825\u001b[39m, in \u001b[36mOpenMLTrainerModule.run_training\u001b[39m\u001b[34m(self, task, X_train, y_train, X_test)\u001b[39m\n\u001b[32m    822\u001b[39m \u001b[38;5;28mself\u001b[39m.plot_all_metrics = \u001b[38;5;28mself\u001b[39m.stats.plot_all_metrics\n\u001b[32m    824\u001b[39m \u001b[38;5;28mself\u001b[39m.learn.model.train()\n\u001b[32m--> \u001b[39m\u001b[32m825\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mrunner\u001b[49m\u001b[43m.\u001b[49m\u001b[43mfit\u001b[49m\u001b[43m(\u001b[49m\u001b[43mepochs\u001b[49m\u001b[43m=\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mconfig\u001b[49m\u001b[43m.\u001b[49m\u001b[43mepoch_count\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mlearn\u001b[49m\u001b[43m=\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mlearn\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    826\u001b[39m \u001b[38;5;28mself\u001b[39m.learn.model.eval()\n\u001b[32m    828\u001b[39m \u001b[38;5;28mself\u001b[39m.lrs = \u001b[38;5;28mself\u001b[39m.runner.cbs[\u001b[32m1\u001b[39m].lrs\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:501\u001b[39m, in \u001b[36mModelRunner.fit\u001b[39m\u001b[34m(self, epochs, learn)\u001b[39m\n\u001b[32m    499\u001b[39m \u001b[38;5;28mself\u001b[39m.epoch = epoch\n\u001b[32m    500\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m(\u001b[33m\"\u001b[39m\u001b[33mbegin_epoch\u001b[39m\u001b[33m\"\u001b[39m):\n\u001b[32m--> \u001b[39m\u001b[32m501\u001b[39m     \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mall_batches\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mdata\u001b[49m\u001b[43m.\u001b[49m\u001b[43mtrain_dl\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    502\u001b[39m \u001b[38;5;28;01mwith\u001b[39;00m torch.no_grad():\n\u001b[32m    503\u001b[39m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m(\u001b[33m\"\u001b[39m\u001b[33mbegin_validate\u001b[39m\u001b[33m\"\u001b[39m):\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:487\u001b[39m, in \u001b[36mModelRunner.all_batches\u001b[39m\u001b[34m(self, dl)\u001b[39m\n\u001b[32m    483\u001b[39m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[32m    484\u001b[39m     \u001b[38;5;28;01mfor\u001b[39;00m xb, yb \u001b[38;5;129;01min\u001b[39;00m tqdm(\n\u001b[32m    485\u001b[39m         dl, leave=\u001b[38;5;28;01mFalse\u001b[39;00m, desc=\u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mTraining Epoch \u001b[39m\u001b[38;5;132;01m{\u001b[39;00m\u001b[38;5;28mself\u001b[39m.epoch\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m\"\u001b[39m, total=\u001b[38;5;28mself\u001b[39m.iters\n\u001b[32m    486\u001b[39m     ):\n\u001b[32m--> \u001b[39m\u001b[32m487\u001b[39m         \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mone_batch\u001b[49m\u001b[43m(\u001b[49m\u001b[43mxb\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43myb\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    488\u001b[39m \u001b[38;5;28;01mexcept\u001b[39;00m CancelEpochException:\n\u001b[32m    489\u001b[39m     \u001b[38;5;28mself\u001b[39m(\u001b[33m\"\u001b[39m\u001b[33mafter_cancel_epoch\u001b[39m\u001b[33m\"\u001b[39m)\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:468\u001b[39m, in \u001b[36mModelRunner.one_batch\u001b[39m\u001b[34m(self, xb, yb)\u001b[39m\n\u001b[32m    466\u001b[39m \u001b[38;5;28mself\u001b[39m(\u001b[33m\"\u001b[39m\u001b[33mafter_pred\u001b[39m\u001b[33m\"\u001b[39m)\n\u001b[32m    467\u001b[39m \u001b[38;5;28mself\u001b[39m.loss = \u001b[38;5;28mself\u001b[39m.loss_fn(\u001b[38;5;28mself\u001b[39m.pred, \u001b[38;5;28mself\u001b[39m.yb)\n\u001b[32m--> \u001b[39m\u001b[32m468\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m(\u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43mafter_loss\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[32m    469\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m.in_train:\n\u001b[32m    470\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/trainer.py:515\u001b[39m, in \u001b[36mModelRunner.__call__\u001b[39m\u001b[34m(self, cb_name)\u001b[39m\n\u001b[32m    512\u001b[39m         \u001b[38;5;28mself\u001b[39m(\u001b[33m\"\u001b[39m\u001b[33mafter_fit\u001b[39m\u001b[33m\"\u001b[39m)\n\u001b[32m    513\u001b[39m         \u001b[38;5;28mself\u001b[39m.learn = \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[32m--> \u001b[39m\u001b[32m515\u001b[39m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, cb_name):\n\u001b[32m    516\u001b[39m     res = \u001b[38;5;28;01mFalse\u001b[39;00m\n\u001b[32m    517\u001b[39m     \u001b[38;5;28;01mfor\u001b[39;00m cb \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28msorted\u001b[39m(\u001b[38;5;28mself\u001b[39m.cbs, key=\u001b[38;5;28;01mlambda\u001b[39;00m x: x._order):\n",
      "\u001b[31mKeyboardInterrupt\u001b[39m: "
     ]
    }
   ],
   "source": [
    "import torch\n",
    "\n",
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    experiment_name= \"Testing model upload\",\n",
    "    data_module=data_module,\n",
    "    verbose=True,\n",
    "    epoch_count=1,\n",
    "    metrics= [accuracy],\n",
    "    # remove the TestCallback when you are done testing your pipeline. Having it here will make the pipeline run for a very short time.\n",
    "    callbacks=[\n",
    "        # TestCallback,\n",
    "    ],\n",
    "    opt=torch.optim.Adam,\n",
    "    loss_fn = torch.nn.CrossEntropyLoss\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No metrics recorded.\n",
      "Error adding metrics to run: [Errno 2] No such file or directory: 'metrics.png'\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "run = openml_pytorch.add_experiment_info_to_run(run=run, trainer=trainer, upload_model=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "While models of this type are not officially supported, we will try to serialize them anyway.\n",
      "While models of this type are not officially supported, we will try to serialize them anyway.\n",
      "While models of this type are not officially supported, we will try to serialize them anyway.\n",
      "While models of this type are not officially supported, we will try to serialize them anyway.\n",
      "While models of this type are not officially supported, we will try to serialize them anyway.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "OpenML Run\n",
       "==========\n",
       "Uploader Name...................: None\n",
       "Metric..........................: None\n",
       "Local Result - Accuracy (+- STD): 0.0073 +- 0.0041\n",
       "Run ID..........................: 10595313\n",
       "Run URL.........................: https://www.openml.org/r/10595313\n",
       "Task ID.........................: 363295\n",
       "Task Type.......................: None\n",
       "Task URL........................: https://www.openml.org/t/363295\n",
       "Flow ID.........................: 32558\n",
       "Flow Name.......................: torch.nn.TransformerCompatibility.8c52bf3d52b90e7a\n",
       "Flow URL........................: https://www.openml.org/f/32558\n",
       "Setup ID........................: None\n",
       "Setup String....................: Python_3.13.2. Torch_2.6.0. NumPy_2.2.4. SciPy_1.15.2. TransformerCompatibility(\n",
       "  (model): ResNetForImageClassification(\n",
       "    (resnet): ResNetModel(\n",
       "      (embedder): ResNetEmbeddings(\n",
       "        (embedder): ResNetConvLayer(\n",
       "          (convolution): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n",
       "          (normalization): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "          (activation): ReLU()\n",
       "        )\n",
       "        (pooler): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "      )\n",
       "      (encoder): ResNetEncoder(\n",
       "        (stages): ModuleList(\n",
       "          (0): ResNetStage(\n",
       "            (layers): Sequential(\n",
       "              (0): ResNetBasicLayer(\n",
       "                (shortcut): Identity()\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "              (1): ResNetBasicLayer(\n",
       "                (shortcut): Identity()\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetStage(\n",
       "            (layers): Sequential(\n",
       "              (0): ResNetBasicLayer(\n",
       "                (shortcut): ResNetShortCut(\n",
       "                  (convolution): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "                  (normalization): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                )\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "              (1): ResNetBasicLayer(\n",
       "                (shortcut): Identity()\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "            )\n",
       "          )\n",
       "          (2): ResNetStage(\n",
       "            (layers): Sequential(\n",
       "              (0): ResNetBasicLayer(\n",
       "                (shortcut): ResNetShortCut(\n",
       "                  (convolution): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "                  (normalization): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                )\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "              (1): ResNetBasicLayer(\n",
       "                (shortcut): Identity()\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "            )\n",
       "          )\n",
       "          (3): ResNetStage(\n",
       "            (layers): Sequential(\n",
       "              (0): ResNetBasicLayer(\n",
       "                (shortcut): ResNetShortCut(\n",
       "                  (convolution): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)\n",
       "                  (normalization): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                )\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "              (1): ResNetBasicLayer(\n",
       "                (shortcut): Identity()\n",
       "                (layer): Sequential(\n",
       "                  (0): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): ReLU()\n",
       "                  )\n",
       "                  (1): ResNetConvLayer(\n",
       "                    (convolution): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n",
       "                    (normalization): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "                    (activation): Identity()\n",
       "                  )\n",
       "                )\n",
       "                (activation): ReLU()\n",
       "              )\n",
       "            )\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (pooler): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "    )\n",
       "    (classifier): Sequential(\n",
       "      (0): Flatten(start_dim=1, end_dim=-1)\n",
       "      (1): Linear(in_features=512, out_features=200, bias=True)\n",
       "    )\n",
       "  )\n",
       ")\n",
       "Dataset ID......................: 46578\n",
       "Dataset URL.....................: https://www.openml.org/d/46578"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{}"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trainer.stats.metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "Number of rows must be a positive integer, not 0",
     "output_type": "error",
     "traceback": [
      "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
      "\u001b[31mValueError\u001b[39m                                Traceback (most recent call last)",
      "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[12]\u001b[39m\u001b[32m, line 1\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m1\u001b[39m \u001b[43mtrainer\u001b[49m\u001b[43m.\u001b[49m\u001b[43mplot_all_metrics\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/openml_pytorch/callbacks/recording.py:157\u001b[39m, in \u001b[36mRecorder.plot_all_metrics\u001b[39m\u001b[34m(self, skip_last, save_path)\u001b[39m\n\u001b[32m    150\u001b[39m \u001b[38;5;250m\u001b[39m\u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m    151\u001b[39m \u001b[33;03mPlots all available metrics in subplots.\u001b[39;00m\n\u001b[32m    152\u001b[39m \n\u001b[32m    153\u001b[39m \u001b[33;03mArgs:\u001b[39;00m\n\u001b[32m    154\u001b[39m \u001b[33;03m    skip_last (int): Number of last points to skip for all metrics\u001b[39;00m\n\u001b[32m    155\u001b[39m \u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m    156\u001b[39m num_metrics = \u001b[38;5;28mlen\u001b[39m(\u001b[38;5;28mself\u001b[39m.metrics)\n\u001b[32m--> \u001b[39m\u001b[32m157\u001b[39m fig, axes = \u001b[43mplt\u001b[49m\u001b[43m.\u001b[49m\u001b[43msubplots\u001b[49m\u001b[43m(\u001b[49m\u001b[43mnum_metrics\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[32;43m1\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfigsize\u001b[49m\u001b[43m=\u001b[49m\u001b[43m(\u001b[49m\u001b[32;43m10\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[32;43m6\u001b[39;49m\u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43m \u001b[49m\u001b[43mnum_metrics\u001b[49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    159\u001b[39m \u001b[38;5;66;03m# If there's only one metric, axes is not a list, so make sure we handle that.\u001b[39;00m\n\u001b[32m    160\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m num_metrics == \u001b[32m1\u001b[39m:\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/pyplot.py:1770\u001b[39m, in \u001b[36msubplots\u001b[39m\u001b[34m(nrows, ncols, sharex, sharey, squeeze, width_ratios, height_ratios, subplot_kw, gridspec_kw, **fig_kw)\u001b[39m\n\u001b[32m   1625\u001b[39m \u001b[38;5;250m\u001b[39m\u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1626\u001b[39m \u001b[33;03mCreate a figure and a set of subplots.\u001b[39;00m\n\u001b[32m   1627\u001b[39m \n\u001b[32m   (...)\u001b[39m\u001b[32m   1767\u001b[39m \n\u001b[32m   1768\u001b[39m \u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1769\u001b[39m fig = figure(**fig_kw)\n\u001b[32m-> \u001b[39m\u001b[32m1770\u001b[39m axs = \u001b[43mfig\u001b[49m\u001b[43m.\u001b[49m\u001b[43msubplots\u001b[49m\u001b[43m(\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m=\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mncols\u001b[49m\u001b[43m=\u001b[49m\u001b[43mncols\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msharex\u001b[49m\u001b[43m=\u001b[49m\u001b[43msharex\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msharey\u001b[49m\u001b[43m=\u001b[49m\u001b[43msharey\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1771\u001b[39m \u001b[43m                   \u001b[49m\u001b[43msqueeze\u001b[49m\u001b[43m=\u001b[49m\u001b[43msqueeze\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msubplot_kw\u001b[49m\u001b[43m=\u001b[49m\u001b[43msubplot_kw\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1772\u001b[39m \u001b[43m                   \u001b[49m\u001b[43mgridspec_kw\u001b[49m\u001b[43m=\u001b[49m\u001b[43mgridspec_kw\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mheight_ratios\u001b[49m\u001b[43m=\u001b[49m\u001b[43mheight_ratios\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1773\u001b[39m \u001b[43m                   \u001b[49m\u001b[43mwidth_ratios\u001b[49m\u001b[43m=\u001b[49m\u001b[43mwidth_ratios\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m   1774\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m fig, axs\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/figure.py:918\u001b[39m, in \u001b[36mFigureBase.subplots\u001b[39m\u001b[34m(self, nrows, ncols, sharex, sharey, squeeze, width_ratios, height_ratios, subplot_kw, gridspec_kw)\u001b[39m\n\u001b[32m    914\u001b[39m         \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\u001b[33m\"\u001b[39m\u001b[33m'\u001b[39m\u001b[33mwidth_ratios\u001b[39m\u001b[33m'\u001b[39m\u001b[33m must not be defined both as \u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m    915\u001b[39m                          \u001b[33m\"\u001b[39m\u001b[33mparameter and as key in \u001b[39m\u001b[33m'\u001b[39m\u001b[33mgridspec_kw\u001b[39m\u001b[33m'\u001b[39m\u001b[33m\"\u001b[39m)\n\u001b[32m    916\u001b[39m     gridspec_kw[\u001b[33m'\u001b[39m\u001b[33mwidth_ratios\u001b[39m\u001b[33m'\u001b[39m] = width_ratios\n\u001b[32m--> \u001b[39m\u001b[32m918\u001b[39m gs = \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43madd_gridspec\u001b[49m\u001b[43m(\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mncols\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfigure\u001b[49m\u001b[43m=\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43m*\u001b[49m\u001b[43mgridspec_kw\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m    919\u001b[39m axs = gs.subplots(sharex=sharex, sharey=sharey, squeeze=squeeze,\n\u001b[32m    920\u001b[39m                   subplot_kw=subplot_kw)\n\u001b[32m    921\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m axs\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/figure.py:1600\u001b[39m, in \u001b[36mFigureBase.add_gridspec\u001b[39m\u001b[34m(self, nrows, ncols, **kwargs)\u001b[39m\n\u001b[32m   1557\u001b[39m \u001b[38;5;250m\u001b[39m\u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1558\u001b[39m \u001b[33;03mLow-level API for creating a `.GridSpec` that has this figure as a parent.\u001b[39;00m\n\u001b[32m   1559\u001b[39m \n\u001b[32m   (...)\u001b[39m\u001b[32m   1596\u001b[39m \n\u001b[32m   1597\u001b[39m \u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m   1599\u001b[39m _ = kwargs.pop(\u001b[33m'\u001b[39m\u001b[33mfigure\u001b[39m\u001b[33m'\u001b[39m, \u001b[38;5;28;01mNone\u001b[39;00m)  \u001b[38;5;66;03m# pop in case user has added this...\u001b[39;00m\n\u001b[32m-> \u001b[39m\u001b[32m1600\u001b[39m gs = \u001b[43mGridSpec\u001b[49m\u001b[43m(\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m=\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mncols\u001b[49m\u001b[43m=\u001b[49m\u001b[43mncols\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfigure\u001b[49m\u001b[43m=\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m*\u001b[49m\u001b[43m*\u001b[49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m   1601\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m gs\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/gridspec.py:363\u001b[39m, in \u001b[36mGridSpec.__init__\u001b[39m\u001b[34m(self, nrows, ncols, figure, left, bottom, right, top, wspace, hspace, width_ratios, height_ratios)\u001b[39m\n\u001b[32m    360\u001b[39m \u001b[38;5;28mself\u001b[39m.hspace = hspace\n\u001b[32m    361\u001b[39m \u001b[38;5;28mself\u001b[39m.figure = figure\n\u001b[32m--> \u001b[39m\u001b[32m363\u001b[39m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[43m.\u001b[49m\u001b[34;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mnrows\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mncols\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    364\u001b[39m \u001b[43m                 \u001b[49m\u001b[43mwidth_ratios\u001b[49m\u001b[43m=\u001b[49m\u001b[43mwidth_ratios\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m    365\u001b[39m \u001b[43m                 \u001b[49m\u001b[43mheight_ratios\u001b[49m\u001b[43m=\u001b[49m\u001b[43mheight_ratios\u001b[49m\u001b[43m)\u001b[49m\n",
      "\u001b[36mFile \u001b[39m\u001b[32m~/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/gridspec.py:48\u001b[39m, in \u001b[36mGridSpecBase.__init__\u001b[39m\u001b[34m(self, nrows, ncols, height_ratios, width_ratios)\u001b[39m\n\u001b[32m     33\u001b[39m \u001b[38;5;250m\u001b[39m\u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m     34\u001b[39m \u001b[33;03mParameters\u001b[39;00m\n\u001b[32m     35\u001b[39m \u001b[33;03m----------\u001b[39;00m\n\u001b[32m   (...)\u001b[39m\u001b[32m     45\u001b[39m \u001b[33;03m    If not given, all rows will have the same height.\u001b[39;00m\n\u001b[32m     46\u001b[39m \u001b[33;03m\"\"\"\u001b[39;00m\n\u001b[32m     47\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(nrows, Integral) \u001b[38;5;129;01mor\u001b[39;00m nrows <= \u001b[32m0\u001b[39m:\n\u001b[32m---> \u001b[39m\u001b[32m48\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[32m     49\u001b[39m         \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mNumber of rows must be a positive integer, not \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mnrows\u001b[38;5;132;01m!r}\u001b[39;00m\u001b[33m\"\u001b[39m)\n\u001b[32m     50\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(ncols, Integral) \u001b[38;5;129;01mor\u001b[39;00m ncols <= \u001b[32m0\u001b[39m:\n\u001b[32m     51\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[32m     52\u001b[39m         \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mNumber of columns must be a positive integer, not \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mncols\u001b[38;5;132;01m!r}\u001b[39;00m\u001b[33m\"\u001b[39m)\n",
      "\u001b[31mValueError\u001b[39m: Number of rows must be a positive integer, not 0"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 1000x0 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.plot_all_metrics()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'matplotlib.pyplot' from '/Users/smukherjee/Documents/CODE/Github/OpenML/openml-pytorch/.venv/lib/python3.13/site-packages/matplotlib/pyplot.py'>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainer.runner.cbs[1].plot_metric('accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x399c23c50>]"
      ]
     },
     "execution_count": 13,
     "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": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x322c10910>]"
      ]
     },
     "execution_count": 22,
     "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": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## testing push\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "run = openml_pytorch.add_experiment_info_to_run(run=run, trainer=trainer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning rate and loss plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model_classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.cbfs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.experiment_name"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Vizualization\n",
    "- Sometimes you may want to visualize the model. You can either use netron or tensorboard for this purpose."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Netron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.export_to_netron()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tensorboard\n",
    "- By default, openml will log the tensorboard logs in the `tensorboard_logs` directory. You can view the logs by running `tensorboard --logdir tensorboard_logs` in the terminal."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Publish your model to OpenML\n",
    "- This is Optional, but publishing your model to OpenML will allow you to track your experiments and compare them with others.\n",
    "- Make sure to set your apikey first.\n",
    "  - You can find your apikey on your OpenML account page."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "run.publish()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pretrained Transformer Image Classification Task"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# openml imports\n",
    "import openml\n",
    "import openml_pytorch\n",
    "from openml_pytorch.callbacks import TestCallback\n",
    "from openml_pytorch.metrics import accuracy\n",
    "from openml_pytorch.trainer import OpenMLDataModule, OpenMLTrainerModule, convert_to_rgb\n",
    "\n",
    "# pytorch imports\n",
    "from torch.utils.tensorboard.writer import SummaryWriter\n",
    "from torchvision.transforms import Compose, Resize, ToPILImage, ToTensor, Lambda\n",
    "import torchvision\n",
    "\n",
    "# other imports\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# set up logging\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "warnings.simplefilter(action='ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define image transformations\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "transform = Compose(\n",
    "    [\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(convert_to_rgb),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize((64, 64)),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure the Data Module and Choose a Task\n",
    "- Make sure the data is present in the `file_dir` directory, and the `filename_col` is correctly set along with this column correctly pointing to where your data is stored. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"image\",\n",
    "    file_dir=\"datasets\",\n",
    "    filename_col=\"image_path\",\n",
    "    target_mode=\"categorical\",\n",
    "    target_column=\"label\",\n",
    "    batch_size=64,\n",
    "    transform=transform,\n",
    ")\n",
    "\n",
    "# Download the OpenML task for tiniest imagenet\n",
    "task = openml.tasks.get_task(362128)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example model. You can do better :)\n",
    "import torchvision.models as models\n",
    "import torch.nn as nn\n",
    "\n",
    "# Load the pre-trained model\n",
    "model = models.efficientnet_b0(pretrained=True)\n",
    "\n",
    "# Modify the last fully connected layer to the required number of classes\n",
    "num_classes = 200\n",
    "in_features = model.classifier[-1].in_features\n",
    "model.classifier = nn.Sequential(\n",
    "    nn.Dropout(p=0.2, inplace=True),\n",
    "    nn.Linear(in_features, num_classes),\n",
    ")\n",
    "\n",
    "# Optional: If you're fine-tuning, you may want to freeze the pre-trained layers\n",
    "# for param in model.parameters():\n",
    "#     param.requires_grad = False\n",
    "\n",
    "# # If you want to train the last layer only (the newly added layer)\n",
    "# for param in model.fc.parameters():\n",
    "#     param.requires_grad = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train your model on the data\n",
    "- Note that by default, OpenML runs a 10 fold cross validation on the data. You cannot change this for now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    experiment_name= \"Tiny ImageNet\",\n",
    "    data_module=data_module,\n",
    "    verbose=True,\n",
    "    epoch_count=1,\n",
    "    metrics= [accuracy],\n",
    "    # remove the TestCallback when you are done testing your pipeline. Having it here will make the pipeline run for a very short time.\n",
    "    callbacks=[\n",
    "        TestCallback,\n",
    "    ],\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## View information about your run"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning rate and loss plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Vizualization\n",
    "- Sometimes you may want to visualize the model. You can either use netron or tensorboard for this purpose."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Netron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.export_to_netron()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tensorboard\n",
    "- By default, openml will log the tensorboard logs in the `tensorboard_logs` directory. You can view the logs by running `tensorboard --logdir tensorboard_logs` in the terminal."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Publish your model to OpenML\n",
    "- This is Optional, but publishing your model to OpenML will allow you to track your experiments and compare them with others.\n",
    "- Make sure to set your apikey first.\n",
    "  - You can find your apikey on your OpenML account page."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = ''\n",
    "run.publish()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Choose a custom Optimizer\n",
    "- If you want to use a custom optimizer, you can do so by defining the optimizer in the `optimizer` variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# openml imports\n",
    "import openml\n",
    "import openml_pytorch\n",
    "from openml_pytorch.callbacks import TestCallback\n",
    "from openml_pytorch.metrics import accuracy\n",
    "from openml_pytorch.trainer import OpenMLDataModule, OpenMLTrainerModule, convert_to_rgb\n",
    "\n",
    "# pytorch imports\n",
    "from torch.utils.tensorboard.writer import SummaryWriter\n",
    "from torchvision.transforms import Compose, Resize, ToPILImage, ToTensor, Lambda\n",
    "import torchvision\n",
    "\n",
    "# other imports\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# set up logging\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "warnings.simplefilter(action='ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define image transformations\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "transform = Compose(\n",
    "    [\n",
    "        ToPILImage(),  # Convert tensor to PIL Image to ensure PIL Image operations can be applied.\n",
    "        Lambda(convert_to_rgb),  # Convert PIL Image to RGB if it's not already.\n",
    "        Resize((64, 64)),  # Resize the image.\n",
    "        ToTensor(),  # Convert the PIL Image back to a tensor.\n",
    "    ]\n",
    ")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure the Data Module and Choose a Task\n",
    "- Make sure the data is present in the `file_dir` directory, and the `filename_col` is correctly set along with this column correctly pointing to where your data is stored. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"image\",\n",
    "    file_dir=\"datasets\",\n",
    "    filename_col=\"image_path\",\n",
    "    target_mode=\"categorical\",\n",
    "    target_column=\"label\",\n",
    "    batch_size=64,\n",
    "    transform=transform,\n",
    ")\n",
    "\n",
    "# Download the OpenML task for tiniest imagenet\n",
    "task = openml.tasks.get_task(362128)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = torchvision.models.resnet18(num_classes=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train your model on the data\n",
    "- Choose a custom optimizer by defining the optimizer variable.\n",
    "- Note that by default, OpenML runs a 10 fold cross validation on the data. You cannot change this for now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.optim import Adam, Optimizer\n",
    "from openml_pytorch.trainer import OpenMLTask\n",
    "\n",
    "def custom_optimizer_gen(model: torch.nn.Module, task: OpenMLTask) -> Optimizer:\n",
    "    # replace the optimizer with your own\n",
    "    return Adam(model.fc.parameters())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = OpenMLTrainerModule(\n",
    "    experiment_name= \"Tiny ImageNet\",\n",
    "    data_module=data_module,\n",
    "    verbose=True,\n",
    "    epoch_count=1,\n",
    "    metrics= [accuracy],\n",
    "    # remove the TestCallback when you are done testing your pipeline. Having it here will make the pipeline run for a very short time.\n",
    "    callbacks=[\n",
    "        TestCallback,\n",
    "    ],\n",
    "    optimizer = custom_optimizer_gen,\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## View information about your run"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning rate and loss plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Vizualization\n",
    "- Sometimes you may want to visualize the model. You can either use netron or tensorboard for this purpose."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Netron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.export_to_netron()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tensorboard\n",
    "- By default, openml will log the tensorboard logs in the `tensorboard_logs` directory. You can view the logs by running `tensorboard --logdir tensorboard_logs` in the terminal."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Publish your model to OpenML\n",
    "- This is Optional, but publishing your model to OpenML will allow you to track your experiments and compare them with others.\n",
    "- Make sure to set your apikey first.\n",
    "  - You can find your apikey on your OpenML account page."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = ''\n",
    "run.publish()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sequential Classification Task\n",
    "- Sequential classification of a tabular MNIST dataset (Task 3573) using a simple neural network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# openml imports\n",
    "import openml\n",
    "import openml_pytorch\n",
    "from openml_pytorch.callbacks import TestCallback\n",
    "from openml_pytorch.metrics import accuracy\n",
    "from openml_pytorch.trainer import OpenMLDataModule, OpenMLTrainerModule, convert_to_rgb\n",
    "\n",
    "# pytorch imports\n",
    "import torch\n",
    "\n",
    "# other imports\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# set up logging\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "warnings.simplefilter(action='ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure the Data Module and Choose a Task\n",
    "- Make sure the data is present in the `file_dir` directory, and the `filename_col` is correctly set along with this column correctly pointing to where your data is stored. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"dataframe\",\n",
    "    filename_col=\"class\",\n",
    "    target_mode=\"categorical\",\n",
    ")\n",
    "\n",
    "# Download the OpenML task for the mnist 784 dataset.\n",
    "task = openml.tasks.get_task(3573)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "############################################################################\n",
    "# Define a sequential network that does the initial image reshaping\n",
    "# and normalization model.\n",
    "processing_net = torch.nn.Sequential(\n",
    "    openml_pytorch.layers.Functional(function=torch.Tensor.reshape,\n",
    "                                                shape=(-1, 1, 28, 28)),\n",
    "    torch.nn.BatchNorm2d(num_features=1)\n",
    ")\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "# Define a sequential network that does the extracts the features from the\n",
    "# image.\n",
    "features_net = torch.nn.Sequential(\n",
    "    torch.nn.Conv2d(in_channels=1, out_channels=32, kernel_size=5),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.MaxPool2d(kernel_size=2),\n",
    "    torch.nn.Conv2d(in_channels=32, out_channels=64, kernel_size=5),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.MaxPool2d(kernel_size=2),\n",
    ")\n",
    "############################################################################\n",
    "\n",
    "############################################################################\n",
    "# Define a sequential network that flattens the features and compiles the\n",
    "# results into probabilities for each digit.\n",
    "results_net = torch.nn.Sequential(\n",
    "    openml_pytorch.layers.Functional(function=torch.Tensor.reshape,\n",
    "                                                shape=(-1, 4 * 4 * 64)),\n",
    "    torch.nn.Linear(in_features=4 * 4 * 64, out_features=256),\n",
    "    torch.nn.LeakyReLU(),\n",
    "    torch.nn.Dropout(),\n",
    "    torch.nn.Linear(in_features=256, out_features=10),\n",
    ")\n",
    "############################################################################\n",
    "# openml.config.apikey = 'key'\n",
    "\n",
    "############################################################################\n",
    "# The main network, composed of the above specified networks.\n",
    "model = torch.nn.Sequential(\n",
    "    processing_net,\n",
    "    features_net,\n",
    "    results_net\n",
    ")\n",
    "############################################################################\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train your model on the data\n",
    "- Note that by default, OpenML runs a 10 fold cross validation on the data. You cannot change this for now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer = OpenMLTrainerModule(\n",
    "    experiment_name= \"MNIST\",\n",
    "    data_module=data_module,\n",
    "    verbose=True,\n",
    "    epoch_count=1,\n",
    "    metrics= [accuracy],\n",
    "    # remove the TestCallback when you are done testing your pipeline. Having it here will make the pipeline run for a very short time.\n",
    "    callbacks=[\n",
    "        TestCallback,\n",
    "    ],\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## View information about your run"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning rate and loss plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Vizualization\n",
    "- Sometimes you may want to visualize the model. You can either use netron or tensorboard for this purpose."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Netron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.export_to_netron()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tensorboard\n",
    "- By default, openml will log the tensorboard logs in the `tensorboard_logs` directory. You can view the logs by running `tensorboard --logdir tensorboard_logs` in the terminal."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Publish your model to OpenML\n",
    "- This is Optional, but publishing your model to OpenML will allow you to track your experiments and compare them with others.\n",
    "- Make sure to set your apikey first.\n",
    "  - You can find your apikey on your OpenML account page."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = ''\n",
    "run.publish()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tabular classification\n",
    "- Supervised credit-g classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# openml imports\n",
    "import openml\n",
    "import openml_pytorch\n",
    "from openml_pytorch.callbacks import TestCallback\n",
    "from openml_pytorch.metrics import accuracy\n",
    "from openml_pytorch.trainer import OpenMLDataModule, OpenMLTrainerModule, convert_to_rgb\n",
    "\n",
    "# pytorch imports\n",
    "import torch\n",
    "\n",
    "# other imports\n",
    "import logging\n",
    "import warnings\n",
    "\n",
    "# set up logging\n",
    "openml.config.logger.setLevel(logging.DEBUG)\n",
    "openml_pytorch.config.logger.setLevel(logging.DEBUG)\n",
    "warnings.simplefilter(action='ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define image transformations\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Configure the Data Module and Choose a Task\n",
    "- Make sure the data is present in the `file_dir` directory, and the `filename_col` is correctly set along with this column correctly pointing to where your data is stored. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_module = OpenMLDataModule(\n",
    "    type_of_data=\"dataframe\",\n",
    "    target_column=\"class\",\n",
    "    target_mode=\"categorical\",\n",
    ")\n",
    "\n",
    "# supervised credit-g classification\n",
    "task = openml.tasks.get_task(31)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TabularClassificationmodel(torch.nn.Module):\n",
    "    def __init__(self, input_size, output_size):\n",
    "        super(TabularClassificationmodel, self).__init__()\n",
    "        self.fc1 = torch.nn.Linear(input_size, 128)\n",
    "        self.fc2 = torch.nn.Linear(128, 64)\n",
    "        self.fc3 = torch.nn.Linear(64, output_size)\n",
    "        self.relu = torch.nn.ReLU()\n",
    "        self.softmax = torch.nn.Softmax(dim=1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.fc1(x)\n",
    "        x = self.relu(x)\n",
    "        x = self.fc2(x)\n",
    "        x = self.relu(x)\n",
    "        x = self.fc3(x)\n",
    "        x = self.softmax(x)\n",
    "        return x\n",
    "    \n",
    "model = TabularClassificationmodel(20, 2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train your model on the data\n",
    "- Note that by default, OpenML runs a 10 fold cross validation on the data. You cannot change this for now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "trainer = OpenMLTrainerModule(\n",
    "    experiment_name= \"Credit-G\",\n",
    "    data_module=data_module,\n",
    "    verbose=True,\n",
    "    epoch_count=5,\n",
    "    metrics= [accuracy],\n",
    "    # remove the TestCallback when you are done testing your pipeline. Having it here will make the pipeline run for a very short time.\n",
    "    callbacks=[\n",
    "        TestCallback,\n",
    "    ],\n",
    ")\n",
    "openml_pytorch.config.trainer = trainer\n",
    "run = openml.runs.run_model_on_task(model, task, avoid_duplicate_runs=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## View information about your run"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning rate and loss plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.plot_lr()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Class labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.model_classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Vizualization\n",
    "- Sometimes you may want to visualize the model. You can either use netron or tensorboard for this purpose."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Netron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.export_to_netron()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tensorboard\n",
    "- By default, openml will log the tensorboard logs in the `tensorboard_logs` directory. You can view the logs by running `tensorboard --logdir tensorboard_logs` in the terminal."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Publish your model to OpenML\n",
    "- This is Optional, but publishing your model to OpenML will allow you to track your experiments and compare them with others.\n",
    "- Make sure to set your apikey first.\n",
    "  - You can find your apikey on your OpenML account page."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "openml.config.apikey = ''\n",
    "run.publish()"
   ]
  },
  {
   "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": "3.13.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}