101. Model Training.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "b9145fc3-4dca-4140-83ff-5c0a299741cf",
   "metadata": {},
   "source": [
    "### 0. Check GPU set up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ed62e7a3-361b-4bcc-8a28-89a40bf23744",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[name: \"/device:CPU:0\"\n",
       " device_type: \"CPU\"\n",
       " memory_limit: 268435456\n",
       " locality {\n",
       " }\n",
       " incarnation: 15494064905979831625,\n",
       " name: \"/device:GPU:0\"\n",
       " device_type: \"GPU\"\n",
       " memory_limit: 7805599744\n",
       " locality {\n",
       "   bus_id: 1\n",
       "   links {\n",
       "   }\n",
       " }\n",
       " incarnation: 12102838610243404348\n",
       " physical_device_desc: \"device: 0, name: NVIDIA GeForce RTX 3080, pci bus id: 0000:01:00.0, compute capability: 8.6\"]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.python.client import device_lib\n",
    "device_lib.list_local_devices()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0fa593cf-d118-4707-9423-0848fd71d23a",
   "metadata": {},
   "source": [
    "### 1. Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5daa5ba5-2a2a-49ce-b206-c5013c145aee",
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "from chartbot_config import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "cc0b4fce-57c9-481a-9bbf-c20cd96a7cff",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\Anaconda3\\envs\\GPU\\lib\\site-packages\\tqdm\\auto.py:22: 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"
     ]
    },
    {
     "data": {
      "text/plain": [
       "DatasetDict({\n",
       "    train: Dataset({\n",
       "        features: ['index', 'Domain', 'Sub domain', 'Intent', 'Answer Format', 'value', 'labels'],\n",
       "        num_rows: 9766\n",
       "    })\n",
       "    validation: Dataset({\n",
       "        features: ['index', 'Domain', 'Sub domain', 'Intent', 'Answer Format', 'value', 'labels'],\n",
       "        num_rows: 3256\n",
       "    })\n",
       "    test: Dataset({\n",
       "        features: ['index', 'Domain', 'Sub domain', 'Intent', 'Answer Format', 'value', 'labels'],\n",
       "        num_rows: 3256\n",
       "    })\n",
       "})"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from datasets import load_from_disk\n",
    "\n",
    "df = load_from_disk(processed_data_path)\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8be50a7e-0ec8-4fd9-bd36-bd9a5bb7c848",
   "metadata": {},
   "source": [
    "### 2. DataSet Set up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "cfaaaf51-a026-4ecd-9ae8-41563dbe8268",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 10/10 [00:00<00:00, 25.33ba/s]\n",
      "100%|██████████| 4/4 [00:00<00:00, 30.33ba/s]\n",
      "100%|██████████| 4/4 [00:00<00:00, 20.16ba/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "======================================\n",
      "Trainset: <PrefetchDataset shapes: ({input_ids: (32, None), token_type_ids: (32, None), attention_mask: (32, None)}, (32, None)), types: ({input_ids: tf.int64, token_type_ids: tf.int64, attention_mask: tf.int64}, tf.int64)>\n"
     ]
    }
   ],
   "source": [
    "from transformers import AutoTokenizer, DataCollatorWithPadding\n",
    "\n",
    "checkpoint = checkpoint\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(checkpoint)\n",
    "\n",
    "def tokenize_function(examples):\n",
    "    return tokenizer(examples[\"value\"], truncation=True)\n",
    "\n",
    "tokenized_dataset = df.map(tokenize_function, batched=True)\n",
    "\n",
    "data_collator = DataCollatorWithPadding(tokenizer = tokenizer, return_tensors = \"tf\")\n",
    "\n",
    "tf_train_dataset = tokenized_dataset[\"train\"].to_tf_dataset(\n",
    "    columns = [\"attention_mask\", \"input_ids\", \"token_type_ids\"],\n",
    "    label_cols = [\"labels\"],\n",
    "    shuffle = True,\n",
    "    collate_fn = data_collator,\n",
    "    batch_size = batch_size,\n",
    ")\n",
    "\n",
    "tf_validation_dataset = tokenized_dataset[\"validation\"].to_tf_dataset(\n",
    "    columns = [\"attention_mask\", \"input_ids\", \"token_type_ids\"],\n",
    "    label_cols = [\"labels\"],\n",
    "    shuffle = False,\n",
    "    collate_fn = data_collator,\n",
    "    batch_size = batch_size,\n",
    ")\n",
    "\n",
    "print(\"======================================\")\n",
    "print(\"Trainset:\", tf_train_dataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3952295a-c86b-4694-a960-773c009fd042",
   "metadata": {},
   "source": [
    "### 3. Model Set up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "145332e1-1ef8-4d74-91ad-c2cfa22d1137",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "All model checkpoint layers were used when initializing TFBertForSequenceClassification.\n",
      "\n",
      "Some layers of TFBertForSequenceClassification were not initialized from the model checkpoint at bert-base-uncased and are newly initialized: ['classifier']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    }
   ],
   "source": [
    "from transformers import TFAutoModelForSequenceClassification\n",
    "\n",
    "num_labels = len(tokenized_dataset['train'][\"labels\"][0])\n",
    "model = TFAutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=num_labels, problem_type=\"multi_label_classification\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "a7705ca8-85c5-4667-a468-6d1320ee1cd2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"tf_bert_for_sequence_classification\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "bert (TFBertMainLayer)       multiple                  109482240 \n",
      "_________________________________________________________________\n",
      "dropout_37 (Dropout)         multiple                  0         \n",
      "_________________________________________________________________\n",
      "classifier (Dense)           multiple                  56906     \n",
      "=================================================================\n",
      "Total params: 109,539,146\n",
      "Trainable params: 109,539,146\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c135805b-e5da-4333-9a53-d9ce0088515e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.losses import BinaryCrossentropy\n",
    "from tensorflow.keras.optimizers.schedules import PolynomialDecay\n",
    "from tensorflow.keras.optimizers import Adam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a6438380-84b1-42d3-b8bf-1149f7a79326",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_train_steps = len(tf_train_dataset) * max_num_epochs\n",
    "\n",
    "lr_scheduler = PolynomialDecay(\n",
    "    initial_learning_rate=5e-5, end_learning_rate=0.0, decay_steps=num_train_steps\n",
    ")\n",
    "\n",
    "opt = Adam(learning_rate=lr_scheduler)\n",
    "loss = BinaryCrossentropy(from_logits=True)\n",
    "\n",
    "model.compile(\n",
    "    optimizer = opt,\n",
    "    loss = loss,\n",
    "    metrics=[\"accuracy\"],\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e21badf-d791-4fda-988f-3f7e6e3670bb",
   "metadata": {},
   "source": [
    "### 4. Model Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "506ae623-e892-4883-bea1-40e06f940ef9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "305/305 [==============================] - 41s 102ms/step - loss: 0.1376 - accuracy: 0.0119 - val_loss: 0.0721 - val_accuracy: 0.0150\n",
      "Epoch 2/10\n",
      "305/305 [==============================] - 30s 98ms/step - loss: 0.0719 - accuracy: 0.0130 - val_loss: 0.0715 - val_accuracy: 0.0117\n",
      "Epoch 3/10\n",
      "305/305 [==============================] - 30s 98ms/step - loss: 0.0698 - accuracy: 0.0670 - val_loss: 0.0659 - val_accuracy: 0.2098\n",
      "Epoch 4/10\n",
      "305/305 [==============================] - 30s 98ms/step - loss: 0.0598 - accuracy: 0.5078 - val_loss: 0.0503 - val_accuracy: 0.8768\n",
      "Epoch 5/10\n",
      "305/305 [==============================] - 30s 98ms/step - loss: 0.0427 - accuracy: 0.9424 - val_loss: 0.0337 - val_accuracy: 0.9693\n",
      "Epoch 6/10\n",
      "305/305 [==============================] - 29s 96ms/step - loss: 0.0290 - accuracy: 0.9869 - val_loss: 0.0236 - val_accuracy: 0.9856\n",
      "Epoch 7/10\n",
      "305/305 [==============================] - 30s 99ms/step - loss: 0.0213 - accuracy: 0.9927 - val_loss: 0.0182 - val_accuracy: 0.9840\n",
      "Epoch 8/10\n",
      "305/305 [==============================] - 30s 99ms/step - loss: 0.0170 - accuracy: 0.9949 - val_loss: 0.0152 - val_accuracy: 0.9883\n",
      "Epoch 9/10\n",
      "305/305 [==============================] - 30s 99ms/step - loss: 0.0147 - accuracy: 0.9950 - val_loss: 0.0137 - val_accuracy: 0.9877\n",
      "Epoch 10/10\n",
      "305/305 [==============================] - 30s 98ms/step - loss: 0.0136 - accuracy: 0.9961 - val_loss: 0.0132 - val_accuracy: 0.9877\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    tf_train_dataset,\n",
    "    validation_data=tf_validation_dataset,\n",
    "    epochs = max_num_epochs\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f01d8ba1-23fb-4ae9-966b-aad12f5060a8",
   "metadata": {},
   "source": [
    "### 5. Model Evaluation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6748dc8a-da1c-4071-94d7-86602acb8e4c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('model accuracy (Overall accuracy*)')\n",
    "plt.ylabel('Overall accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'val'], loc='lower right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "8e19b9cd-fc7e-4de1-8106-e46d91fdba99",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss: Mean Binary Crossentropy (Overall accuracy*)')\n",
    "plt.ylabel('Mean Binary Crossentropy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'val'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "26a26aa3-9703-4bfe-a229-27ebc646291e",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save_weights(model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "20236416-a56c-4a47-928e-ef4cef3b8dc5",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}