Training t4c.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6f12d086-2a3c-474c-9310-7043c1985273",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-07-16 05:15:35.169484: I tensorflow/core/util/port.cc:110] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
      "2023-07-16 05:15:35.189602: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2023-07-16 05:15:35.991441: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:995] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355\n",
      "2023-07-16 05:15:35.991706: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1960] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n",
      "Skipping registering GPU devices...\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>name</th>\n",
       "      <th>tags</th>\n",
       "      <th>dtype</th>\n",
       "      <th>is_list</th>\n",
       "      <th>is_ragged</th>\n",
       "      <th>properties.num_buckets</th>\n",
       "      <th>properties.freq_threshold</th>\n",
       "      <th>properties.max_size</th>\n",
       "      <th>properties.cat_path</th>\n",
       "      <th>properties.domain.min</th>\n",
       "      <th>properties.domain.max</th>\n",
       "      <th>properties.domain.name</th>\n",
       "      <th>properties.embedding_sizes.cardinality</th>\n",
       "      <th>properties.embedding_sizes.dimension</th>\n",
       "      <th>properties.value_count.min</th>\n",
       "      <th>properties.value_count.max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>userId</td>\n",
       "      <td>(Tags.USER, Tags.ID, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.userId.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6042.0</td>\n",
       "      <td>userId</td>\n",
       "      <td>6043.0</td>\n",
       "      <td>210.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>movieId</td>\n",
       "      <td>(Tags.ITEM, Tags.LIST, Tags.ID, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.movieId.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3103.0</td>\n",
       "      <td>movieId</td>\n",
       "      <td>3104.0</td>\n",
       "      <td>144.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>genres</td>\n",
       "      <td>(Tags.LIST, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.genres.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>genres</td>\n",
       "      <td>21.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>binary_rating</td>\n",
       "      <td>(Tags.BINARY_CLASSIFICATION, Tags.TARGET)</td>\n",
       "      <td>DType(name='bool', element_type=&lt;ElementType.B...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "[{'name': 'userId', 'tags': {<Tags.USER: 'user'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 0, 'max_size': 0, 'cat_path': './/categories/unique.userId.parquet', 'domain': {'min': 0, 'max': 6042, 'name': 'userId'}, 'embedding_sizes': {'cardinality': 6043, 'dimension': 210}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'movieId', 'tags': {<Tags.ITEM: 'item'>, <Tags.LIST: 'list'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.movieId.parquet', 'domain': {'min': 0, 'max': 3103, 'name': 'movieId'}, 'embedding_sizes': {'cardinality': 3104, 'dimension': 144}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'genres', 'tags': {<Tags.LIST: 'list'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.genres.parquet', 'domain': {'min': 0, 'max': 20, 'name': 'genres'}, 'embedding_sizes': {'cardinality': 21, 'dimension': 16}, 'value_count': {'min': 0, 'max': None}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None), Dimension(min=0, max=None)))), 'is_list': True, 'is_ragged': True}, {'name': 'binary_rating', 'tags': {<Tags.BINARY_CLASSIFICATION: 'binary_classification'>, <Tags.TARGET: 'target'>}, 'properties': {}, 'dtype': DType(name='bool', element_type=<ElementType.Bool: 'bool'>, element_size=None, element_unit=None, signed=None, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "from merlin.datasets.entertainment import get_movielens\n",
    "\n",
    "input_path = os.environ.get(\"INPUT_DATA_DIR\", os.path.expanduser(\"~/merlin-framework/movielens/\"))\n",
    "get_movielens(variant=\"ml-1m\", path=input_path); #noqa\n",
    "\n",
    "\n",
    "from merlin.core.dispatch import get_lib\n",
    "\n",
    "data = get_lib().read_parquet(f'{input_path}ml-1m/train.parquet').sample(frac=1)\n",
    "\n",
    "train = data.iloc[:600_000]\n",
    "valid = data.iloc[600_000:]\n",
    "\n",
    "movies = get_lib().read_parquet(f'{input_path}ml-1m/movies_converted.parquet')\n",
    "\n",
    "\n",
    "\n",
    "import nvtabular as nvt\n",
    "from merlin.schema.tags import Tags\n",
    "\n",
    "train_ds = nvt.Dataset(train, npartitions=2)\n",
    "valid_ds = nvt.Dataset(valid)\n",
    "\n",
    "train_ds, valid_ds\n",
    "train_ds.shuffle_by_keys('userId')\n",
    "valid_ds.shuffle_by_keys('userId')\n",
    "\n",
    "genres = ['movieId'] >> nvt.ops.JoinExternal(movies, on='movieId', columns_ext=['movieId', 'genres']) >> nvt.ops.AddTags(tags=[Tags.LIST])\n",
    "\n",
    "genres = genres >> nvt.ops.Categorify(freq_threshold=10) \n",
    "\n",
    "def rating_to_binary(col):\n",
    "    return col > 3\n",
    "\n",
    "binary_rating = ['rating'] >> nvt.ops.LambdaOp(rating_to_binary) >> nvt.ops.Rename(name='binary_rating')\n",
    "\n",
    "userId = ['userId'] >> nvt.ops.Categorify() >> nvt.ops.AddTags(tags=[Tags.USER_ID, Tags.CATEGORICAL, Tags.USER])\n",
    "movieId = ['movieId'] >> nvt.ops.Categorify() >> nvt.ops.AddTags(tags=[Tags.ITEM_ID, Tags.CATEGORICAL, Tags.ITEM])\n",
    "binary_rating = binary_rating >> nvt.ops.AddTags(tags=[Tags.TARGET, Tags.BINARY_CLASSIFICATION])\n",
    "\n",
    "\n",
    "workflow = nvt.Workflow(userId + movieId + genres + binary_rating)\n",
    "\n",
    "train_transformed = workflow.fit_transform(train_ds)\n",
    "valid_transformed = workflow.transform(valid_ds)\n",
    "valid_transformed.compute().head()\n",
    "train_transformed.schema\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ba974720-3118-4635-bd76-7b4527c3c084",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>name</th>\n",
       "      <th>tags</th>\n",
       "      <th>dtype</th>\n",
       "      <th>is_list</th>\n",
       "      <th>is_ragged</th>\n",
       "      <th>properties.num_buckets</th>\n",
       "      <th>properties.freq_threshold</th>\n",
       "      <th>properties.max_size</th>\n",
       "      <th>properties.cat_path</th>\n",
       "      <th>properties.domain.min</th>\n",
       "      <th>properties.domain.max</th>\n",
       "      <th>properties.domain.name</th>\n",
       "      <th>properties.embedding_sizes.cardinality</th>\n",
       "      <th>properties.embedding_sizes.dimension</th>\n",
       "      <th>properties.value_count.min</th>\n",
       "      <th>properties.value_count.max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>userId</td>\n",
       "      <td>(Tags.USER, Tags.ID, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.userId.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6042.0</td>\n",
       "      <td>userId</td>\n",
       "      <td>6043.0</td>\n",
       "      <td>210.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>movieId</td>\n",
       "      <td>(Tags.ITEM, Tags.LIST, Tags.ID, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.movieId.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3103.0</td>\n",
       "      <td>movieId</td>\n",
       "      <td>3104.0</td>\n",
       "      <td>144.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>genres</td>\n",
       "      <td>(Tags.LIST, Tags.CATEGORICAL)</td>\n",
       "      <td>DType(name='int64', element_type=&lt;ElementType....</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>.//categories/unique.genres.parquet</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>genres</td>\n",
       "      <td>21.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>binary_rating</td>\n",
       "      <td>(Tags.BINARY_CLASSIFICATION, Tags.TARGET)</td>\n",
       "      <td>DType(name='bool', element_type=&lt;ElementType.B...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "[{'name': 'userId', 'tags': {<Tags.USER: 'user'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 0, 'max_size': 0, 'cat_path': './/categories/unique.userId.parquet', 'domain': {'min': 0, 'max': 6042, 'name': 'userId'}, 'embedding_sizes': {'cardinality': 6043, 'dimension': 210}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'movieId', 'tags': {<Tags.ITEM: 'item'>, <Tags.LIST: 'list'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.movieId.parquet', 'domain': {'min': 0, 'max': 3103, 'name': 'movieId'}, 'embedding_sizes': {'cardinality': 3104, 'dimension': 144}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'genres', 'tags': {<Tags.LIST: 'list'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.genres.parquet', 'domain': {'min': 0, 'max': 20, 'name': 'genres'}, 'embedding_sizes': {'cardinality': 21, 'dimension': 16}, 'value_count': {'min': 0, 'max': None}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None), Dimension(min=0, max=None)))), 'is_list': True, 'is_ragged': True}, {'name': 'binary_rating', 'tags': {<Tags.BINARY_CLASSIFICATION: 'binary_classification'>, <Tags.TARGET: 'target'>}, 'properties': {}, 'dtype': DType(name='bool', element_type=<ElementType.Bool: 'bool'>, element_size=None, element_unit=None, signed=None, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_transformed.schema"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0c0db05e-7630-461e-9ec1-b31aac55fcca",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(-1, 20, 192)\n",
      "(-1, 20, 192)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.8/dist-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "from transformers4rec import torch as tr\n",
    "\n",
    "max_sequence_length, d_model = 20, 320\n",
    "schema = train_transformed.schema\n",
    "# Define input module to process tabular input-features and to prepare masked inputs\n",
    "input_module = tr.TabularSequenceFeatures.from_schema(\n",
    "    schema,\n",
    "    masking=\"clm\",\n",
    "    max_sequence_length=max_sequence_length,\n",
    "    d_model=d_model,\n",
    "\n",
    ")\n",
    "\n",
    "# Define Next item prediction-task \n",
    "prediction_task = tr.BinaryClassificationTask()\n",
    "\n",
    "# Define the config of the XLNet Transformer architecture\n",
    "transformer_config = tr.XLNetConfig.build(\n",
    "    d_model=d_model, n_head=8, n_layer=2, total_seq_length=max_sequence_length\n",
    ")\n",
    "\n",
    "# Get the end-to-end model \n",
    "model = transformer_config.to_torch_model(input_module, prediction_task)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "2810663a-2bde-45bc-8b64-34fc0f70fa6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_args = tr.trainer.T4RecTrainingArguments(\n",
    "            output_dir=\"./tmp\",\n",
    "            max_sequence_length=20,\n",
    "            data_loader_engine='merlin',\n",
    "            num_train_epochs=10, \n",
    "            dataloader_drop_last=False,\n",
    "            per_device_train_batch_size = 384,\n",
    "            per_device_eval_batch_size = 512,\n",
    "            learning_rate=0.0005,\n",
    "            fp16=True,\n",
    "            report_to = [],\n",
    "            logging_steps=200\n",
    "        )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "3e87bc04-09c5-4a5d-b26c-d3e1fe758ecb",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using amp fp16 backend\n"
     ]
    }
   ],
   "source": [
    "recsys_trainer = tr.Trainer(\n",
    "    model=model,\n",
    "    args=training_args,\n",
    "    schema=schema,\n",
    "    compute_metrics=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "a2fd7608-75d5-4fdb-b31c-db36501ce90a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using amp fp16 backend\n"
     ]
    }
   ],
   "source": [
    "# Instantiate the T4Rec Trainer, which manages training and evaluation\n",
    "trainer = tr.Trainer(\n",
    "    model=model,\n",
    "    args=training_args,\n",
    "    schema=schema,\n",
    "    compute_metrics=True,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9a0d2b00-f6fe-44b7-809e-327dc98fb368",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "***** Running training *****\n",
      "  Num examples = 600192\n",
      "  Num Epochs = 10\n",
      "  Instantaneous batch size per device = 384\n",
      "  Total train batch size (w. parallel, distributed & accumulation) = 384\n",
      "  Gradient Accumulation steps = 1\n",
      "  Total optimization steps = 15630\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************\n",
      "Launch training for day 1 are:\n",
      "********************\n",
      "\n",
      "Output Schema -> [{'name': 'userId', 'tags': {<Tags.USER: 'user'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 0, 'max_size': 0, 'cat_path': './/categories/unique.userId.parquet', 'domain': {'min': 0, 'max': 6042, 'name': 'userId'}, 'embedding_sizes': {'cardinality': 6043, 'dimension': 210}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'movieId', 'tags': {<Tags.ITEM: 'item'>, <Tags.LIST: 'list'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.movieId.parquet', 'domain': {'min': 0, 'max': 3103, 'name': 'movieId'}, 'embedding_sizes': {'cardinality': 3104, 'dimension': 144}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}, {'name': 'genres', 'tags': {<Tags.LIST: 'list'>, <Tags.CATEGORICAL: 'categorical'>}, 'properties': {'num_buckets': None, 'freq_threshold': 10, 'max_size': 0, 'cat_path': './/categories/unique.genres.parquet', 'domain': {'min': 0, 'max': 20, 'name': 'genres'}, 'embedding_sizes': {'cardinality': 21, 'dimension': 16}, 'value_count': {'min': 0, 'max': None}}, 'dtype': DType(name='int64', element_type=<ElementType.Int: 'int'>, element_size=64, element_unit=None, signed=True, shape=Shape(dims=(Dimension(min=0, max=None), Dimension(min=0, max=None)))), 'is_list': True, 'is_ragged': True}, {'name': 'binary_rating', 'tags': {<Tags.BINARY_CLASSIFICATION: 'binary_classification'>, <Tags.TARGET: 'target'>}, 'properties': {}, 'dtype': DType(name='bool', element_type=<ElementType.Bool: 'bool'>, element_size=None, element_unit=None, signed=None, shape=Shape(dims=(Dimension(min=0, max=None),))), 'is_list': False, 'is_ragged': False}]\n",
      "Sparse Feats -> ['movieId', 'genres']\n",
      "Padding Lengths {'movieId': 20, 'genres': 20}\n",
      "Item IDS -> torch.Size([384])\n"
     ]
    },
    {
     "ename": "AssertionError",
     "evalue": "`item_ids` must have 2 dimensions.",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0m                            Traceback (most recent call last)",
      "File \u001b[0;32m<timed exec>:15\u001b[0m\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers/trainer.py:1316\u001b[0m, in \u001b[0;36mTrainer.train\u001b[0;34m(self, resume_from_checkpoint, trial, ignore_keys_for_eval, **kwargs)\u001b[0m\n\u001b[1;32m   1314\u001b[0m         tr_loss_step \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtraining_step(model, inputs)\n\u001b[1;32m   1315\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1316\u001b[0m     tr_loss_step \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtraining_step\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1318\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m (\n\u001b[1;32m   1319\u001b[0m     args\u001b[38;5;241m.\u001b[39mlogging_nan_inf_filter\n\u001b[1;32m   1320\u001b[0m     \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m is_torch_tpu_available()\n\u001b[1;32m   1321\u001b[0m     \u001b[38;5;129;01mand\u001b[39;00m (torch\u001b[38;5;241m.\u001b[39misnan(tr_loss_step) \u001b[38;5;129;01mor\u001b[39;00m torch\u001b[38;5;241m.\u001b[39misinf(tr_loss_step))\n\u001b[1;32m   1322\u001b[0m ):\n\u001b[1;32m   1323\u001b[0m     \u001b[38;5;66;03m# if loss is nan or inf simply add the average of previous logged losses\u001b[39;00m\n\u001b[1;32m   1324\u001b[0m     tr_loss \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m tr_loss \u001b[38;5;241m/\u001b[39m (\u001b[38;5;241m1\u001b[39m \u001b[38;5;241m+\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mstate\u001b[38;5;241m.\u001b[39mglobal_step \u001b[38;5;241m-\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_globalstep_last_logged)\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers/trainer.py:1847\u001b[0m, in \u001b[0;36mTrainer.training_step\u001b[0;34m(self, model, inputs)\u001b[0m\n\u001b[1;32m   1845\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39muse_amp:\n\u001b[1;32m   1846\u001b[0m     \u001b[38;5;28;01mwith\u001b[39;00m autocast():\n\u001b[0;32m-> 1847\u001b[0m         loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcompute_loss\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1848\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m   1849\u001b[0m     loss \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcompute_loss(model, inputs)\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/trainer.py:323\u001b[0m, in \u001b[0;36mTrainer.compute_loss\u001b[0;34m(self, model, inputs, return_outputs)\u001b[0m\n\u001b[1;32m    316\u001b[0m \u001b[38;5;250m\u001b[39m\u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m    317\u001b[0m \u001b[38;5;124;03mOverriding :obj:`Trainer.compute_loss()`\u001b[39;00m\n\u001b[1;32m    318\u001b[0m \u001b[38;5;124;03mTo allow for passing the targets to the model's forward method\u001b[39;00m\n\u001b[1;32m    319\u001b[0m \u001b[38;5;124;03mHow the loss is computed by Trainer. By default, all Transformers4Rec models return\u001b[39;00m\n\u001b[1;32m    320\u001b[0m \u001b[38;5;124;03ma dictionary of three elements {'loss', 'predictions', and 'labels}\u001b[39;00m\n\u001b[1;32m    321\u001b[0m \u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m    322\u001b[0m inputs, targets \u001b[38;5;241m=\u001b[39m inputs\n\u001b[0;32m--> 323\u001b[0m outputs \u001b[38;5;241m=\u001b[39m \u001b[43mmodel\u001b[49m\u001b[43m(\u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtargets\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtargets\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m)\u001b[49m\n\u001b[1;32m    324\u001b[0m \u001b[38;5;66;03m# Save past state if it exists\u001b[39;00m\n\u001b[1;32m    325\u001b[0m \u001b[38;5;66;03m# TODO: this needs to be fixed and made cleaner later.\u001b[39;00m\n\u001b[1;32m    326\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39margs\u001b[38;5;241m.\u001b[39mpast_index \u001b[38;5;241m>\u001b[39m\u001b[38;5;241m=\u001b[39m \u001b[38;5;241m0\u001b[39m:\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py:1533\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1528\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.  It's slow for dynamo to guard on the state\u001b[39;00m\n\u001b[1;32m   1530\u001b[0m \u001b[38;5;66;03m# of all these hook dicts individually, so instead it can guard on 2 bools and we just\u001b[39;00m\n\u001b[1;32m   1531\u001b[0m \u001b[38;5;66;03m# have to promise to keep them up to date when hooks are added or removed via official means.\u001b[39;00m\n\u001b[1;32m   1532\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_has_hooks \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m _has_global_hooks:\n\u001b[0;32m-> 1533\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1534\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m   1535\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/model/base.py:560\u001b[0m, in \u001b[0;36mModel.forward\u001b[0;34m(self, inputs, targets, training, testing, **kwargs)\u001b[0m\n\u001b[1;32m    558\u001b[0m predictions \u001b[38;5;241m=\u001b[39m {}\n\u001b[1;32m    559\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m i, head \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28menumerate\u001b[39m(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mheads):\n\u001b[0;32m--> 560\u001b[0m     head_output \u001b[38;5;241m=\u001b[39m \u001b[43mhead\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    561\u001b[0m \u001b[43m        \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    562\u001b[0m \u001b[43m        \u001b[49m\u001b[43mcall_body\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m    563\u001b[0m \u001b[43m        \u001b[49m\u001b[43mtargets\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtargets\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    564\u001b[0m \u001b[43m        \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtraining\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    565\u001b[0m \u001b[43m        \u001b[49m\u001b[43mtesting\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtesting\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    566\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    567\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    568\u001b[0m     labels\u001b[38;5;241m.\u001b[39mupdate(head_output[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mlabels\u001b[39m\u001b[38;5;124m\"\u001b[39m])\n\u001b[1;32m    569\u001b[0m     predictions\u001b[38;5;241m.\u001b[39mupdate(head_output[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mpredictions\u001b[39m\u001b[38;5;124m\"\u001b[39m])\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py:1533\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1528\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.  It's slow for dynamo to guard on the state\u001b[39;00m\n\u001b[1;32m   1530\u001b[0m \u001b[38;5;66;03m# of all these hook dicts individually, so instead it can guard on 2 bools and we just\u001b[39;00m\n\u001b[1;32m   1531\u001b[0m \u001b[38;5;66;03m# have to promise to keep them up to date when hooks are added or removed via official means.\u001b[39;00m\n\u001b[1;32m   1532\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_has_hooks \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m _has_global_hooks:\n\u001b[0;32m-> 1533\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1534\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m   1535\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/model/base.py:382\u001b[0m, in \u001b[0;36mHead.forward\u001b[0;34m(self, body_outputs, training, testing, targets, call_body, top_k, **kwargs)\u001b[0m\n\u001b[1;32m    379\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mtransformers4rec\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mtorch\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mmodel\u001b[39;00m\u001b[38;5;21;01m.\u001b[39;00m\u001b[38;5;21;01mprediction_task\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m NextItemPredictionTask\n\u001b[1;32m    381\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m call_body:\n\u001b[0;32m--> 382\u001b[0m     body_outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbody\u001b[49m\u001b[43m(\u001b[49m\u001b[43mbody_outputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtraining\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtesting\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtesting\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    384\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m training \u001b[38;5;129;01mor\u001b[39;00m testing:\n\u001b[1;32m    385\u001b[0m     losses \u001b[38;5;241m=\u001b[39m []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/config/schema.py:50\u001b[0m, in \u001b[0;36mSchemaMixin.__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m     47\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m     48\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcheck_schema()\n\u001b[0;32m---> 50\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__call__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py:1533\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1528\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.  It's slow for dynamo to guard on the state\u001b[39;00m\n\u001b[1;32m   1530\u001b[0m \u001b[38;5;66;03m# of all these hook dicts individually, so instead it can guard on 2 bools and we just\u001b[39;00m\n\u001b[1;32m   1531\u001b[0m \u001b[38;5;66;03m# have to promise to keep them up to date when hooks are added or removed via official means.\u001b[39;00m\n\u001b[1;32m   1532\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_has_hooks \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m _has_global_hooks:\n\u001b[0;32m-> 1533\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1534\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m   1535\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/block/base.py:256\u001b[0m, in \u001b[0;36mSequentialBlock.forward\u001b[0;34m(self, input, training, testing, **kwargs)\u001b[0m\n\u001b[1;32m    254\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mtraining\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01min\u001b[39;00m inspect\u001b[38;5;241m.\u001b[39msignature(module\u001b[38;5;241m.\u001b[39mforward)\u001b[38;5;241m.\u001b[39mparameters:\n\u001b[1;32m    255\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mtesting\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01min\u001b[39;00m inspect\u001b[38;5;241m.\u001b[39msignature(module\u001b[38;5;241m.\u001b[39mforward)\u001b[38;5;241m.\u001b[39mparameters:\n\u001b[0;32m--> 256\u001b[0m         \u001b[38;5;28minput\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[43mmodule\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtraining\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtesting\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtesting\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    257\u001b[0m     \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    258\u001b[0m         \u001b[38;5;28minput\u001b[39m \u001b[38;5;241m=\u001b[39m module(\u001b[38;5;28minput\u001b[39m, training\u001b[38;5;241m=\u001b[39mtraining)\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/config/schema.py:50\u001b[0m, in \u001b[0;36mSchemaMixin.__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m     47\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m     48\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcheck_schema()\n\u001b[0;32m---> 50\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__call__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/tabular/base.py:392\u001b[0m, in \u001b[0;36mTabularModule.__call__\u001b[0;34m(self, inputs, pre, post, merge_with, aggregation, *args, **kwargs)\u001b[0m\n\u001b[1;32m    389\u001b[0m inputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpre_forward(inputs, transformations\u001b[38;5;241m=\u001b[39mpre)\n\u001b[1;32m    391\u001b[0m \u001b[38;5;66;03m# This will call the `forward` method implemented by the super class.\u001b[39;00m\n\u001b[0;32m--> 392\u001b[0m outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__call__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m  \u001b[38;5;66;03m# noqa\u001b[39;00m\n\u001b[1;32m    394\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(outputs, \u001b[38;5;28mdict\u001b[39m):\n\u001b[1;32m    395\u001b[0m     outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpost_forward(\n\u001b[1;32m    396\u001b[0m         outputs, transformations\u001b[38;5;241m=\u001b[39mpost, merge_with\u001b[38;5;241m=\u001b[39mmerge_with, aggregation\u001b[38;5;241m=\u001b[39maggregation\n\u001b[1;32m    397\u001b[0m     )\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py:1533\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1528\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.  It's slow for dynamo to guard on the state\u001b[39;00m\n\u001b[1;32m   1530\u001b[0m \u001b[38;5;66;03m# of all these hook dicts individually, so instead it can guard on 2 bools and we just\u001b[39;00m\n\u001b[1;32m   1531\u001b[0m \u001b[38;5;66;03m# have to promise to keep them up to date when hooks are added or removed via official means.\u001b[39;00m\n\u001b[1;32m   1532\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_has_hooks \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m _has_global_hooks:\n\u001b[0;32m-> 1533\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1534\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m   1535\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/features/sequence.py:262\u001b[0m, in \u001b[0;36mTabularSequenceFeatures.forward\u001b[0;34m(self, inputs, training, testing, **kwargs)\u001b[0m\n\u001b[1;32m    259\u001b[0m     outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mprojection_module(outputs)\n\u001b[1;32m    261\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mmasking:\n\u001b[0;32m--> 262\u001b[0m     outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mmasking\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    263\u001b[0m \u001b[43m        \u001b[49m\u001b[43moutputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    264\u001b[0m \u001b[43m        \u001b[49m\u001b[43mitem_ids\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mto_merge\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mcategorical_module\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m]\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mitem_seq\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    265\u001b[0m \u001b[43m        \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtraining\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    266\u001b[0m \u001b[43m        \u001b[49m\u001b[43mtesting\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtesting\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    267\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    269\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m outputs\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/config/schema.py:50\u001b[0m, in \u001b[0;36mSchemaMixin.__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m     47\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[1;32m     48\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mcheck_schema()\n\u001b[0;32m---> 50\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__call__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py:1533\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1528\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1529\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.  It's slow for dynamo to guard on the state\u001b[39;00m\n\u001b[1;32m   1530\u001b[0m \u001b[38;5;66;03m# of all these hook dicts individually, so instead it can guard on 2 bools and we just\u001b[39;00m\n\u001b[1;32m   1531\u001b[0m \u001b[38;5;66;03m# have to promise to keep them up to date when hooks are added or removed via official means.\u001b[39;00m\n\u001b[1;32m   1532\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_has_hooks \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m _has_global_hooks:\n\u001b[0;32m-> 1533\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1534\u001b[0m \u001b[38;5;66;03m# Do not call functions when jit is used\u001b[39;00m\n\u001b[1;32m   1535\u001b[0m full_backward_hooks, non_full_backward_hooks \u001b[38;5;241m=\u001b[39m [], []\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/masking.py:223\u001b[0m, in \u001b[0;36mMaskSequence.forward\u001b[0;34m(self, inputs, item_ids, training, testing)\u001b[0m\n\u001b[1;32m    216\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mforward\u001b[39m(\n\u001b[1;32m    217\u001b[0m     \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m    218\u001b[0m     inputs: torch\u001b[38;5;241m.\u001b[39mTensor,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    221\u001b[0m     testing: \u001b[38;5;28mbool\u001b[39m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m,\n\u001b[1;32m    222\u001b[0m ) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m torch\u001b[38;5;241m.\u001b[39mTensor:\n\u001b[0;32m--> 223\u001b[0m     _ \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcompute_masked_targets\u001b[49m\u001b[43m(\u001b[49m\u001b[43mitem_ids\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mitem_ids\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtraining\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtraining\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtesting\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtesting\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    224\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mmask_schema \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m    225\u001b[0m         \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m`mask_schema must be set.`\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n",
      "File \u001b[0;32m/usr/local/lib/python3.8/dist-packages/transformers4rec/torch/masking.py:149\u001b[0m, in \u001b[0;36mMaskSequence.compute_masked_targets\u001b[0;34m(self, item_ids, training, testing)\u001b[0m\n\u001b[1;32m    131\u001b[0m \u001b[38;5;250m\u001b[39m\u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m    132\u001b[0m \u001b[38;5;124;03mMethod to prepare masked labels based on the sequence of item ids.\u001b[39;00m\n\u001b[1;32m    133\u001b[0m \u001b[38;5;124;03mIt returns The true labels of masked positions and the related boolean mask.\u001b[39;00m\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    146\u001b[0m \u001b[38;5;124;03mTuple[MaskingSchema, MaskedTargets]\u001b[39;00m\n\u001b[1;32m    147\u001b[0m \u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m    148\u001b[0m \u001b[38;5;28mprint\u001b[39m(\u001b[38;5;124mf\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mItem IDS -> \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mitem_ids\u001b[38;5;241m.\u001b[39mshape\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m'\u001b[39m)\n\u001b[0;32m--> 149\u001b[0m \u001b[38;5;28;01massert\u001b[39;00m item_ids\u001b[38;5;241m.\u001b[39mndim \u001b[38;5;241m==\u001b[39m \u001b[38;5;241m2\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m`item_ids` must have 2 dimensions.\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m    150\u001b[0m masking_info \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compute_masked_targets(item_ids, training\u001b[38;5;241m=\u001b[39mtraining, testing\u001b[38;5;241m=\u001b[39mtesting)\n\u001b[1;32m    151\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mmask_schema, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mmasked_targets \u001b[38;5;241m=\u001b[39m masking_info\u001b[38;5;241m.\u001b[39mschema, masking_info\u001b[38;5;241m.\u001b[39mtargets\n",
      "\u001b[0;31mAssertionError\u001b[0m: `item_ids` must have 2 dimensions."
     ]
    }
   ],
   "source": [
    "%%time\n",
    "start_time_window_index = 1\n",
    "final_time_window_index = 4\n",
    "for time_index in range(start_time_window_index, final_time_window_index):\n",
    "    # Set data \n",
    "    time_index_train = time_index\n",
    "    time_index_eval = time_index + 1\n",
    "    # train_paths = glob.glob(os.path.join(OUTPUT_DIR, f\"{time_index_train}/train.parquet\"))\n",
    "    # eval_paths = glob.glob(os.path.join(OUTPUT_DIR, f\"{time_index_eval}/valid.parquet\"))\n",
    "    # Train on day related to time_index \n",
    "    print('*'*20)\n",
    "    print(\"Launch training for day %s are:\" %time_index)\n",
    "    print('*'*20 + '\\n')\n",
    "    trainer.train_dataset_or_path = train_transformed\n",
    "    trainer.reset_lr_scheduler()\n",
    "    trainer.train()\n",
    "    trainer.state.global_step +=1\n",
    "    # Evaluate on the following day\n",
    "    trainer.eval_dataset_or_path = valid_transformed\n",
    "    train_metrics = trainer.evaluate(metric_key_prefix='eval')\n",
    "    print('*'*20)\n",
    "    print(\"Eval results for day %s are:\\t\" %time_index_eval)\n",
    "    print('\\n' + '*'*20 + '\\n')\n",
    "    for key in sorted(train_metrics.keys()):\n",
    "        print(\" %s = %s\" % (key, str(train_metrics[key]))) \n",
    "    wipe_memory()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "53603659-cc94-4ed8-a3db-89793496bdf9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Tags.LIST: 'list'>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tags.LIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "48c69414-4e49-469f-bf1a-34cd1ebb15ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{<Tags.ITEM: 'item'>, <Tags.LIST: 'list'>, <Tags.ID: 'id'>, <Tags.CATEGORICAL: 'categorical'>},\n",
       " {<Tags.LIST: 'list'>, <Tags.CATEGORICAL: 'categorical'>}]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[col.tags for col in workflow.output_schema if Tags.LIST in col.tags]"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}