course/videos/debug_training_pt.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook regroups the code sample of the video below, which is a part of the [Hugging Face course](https://huggingface.co/course)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "cellView": "form"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<iframe width=\"560\" height=\"315\" src=\"https://www.youtube.com/embed/L-WSwUWde1U?rel=0&amp;controls=0&amp;showinfo=0\" frameborder=\"0\" allowfullscreen></iframe>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#@title\n",
    "from IPython.display import HTML\n",
    "\n",
    "HTML('<iframe width=\"560\" height=\"315\" src=\"https://www.youtube.com/embed/L-WSwUWde1U?rel=0&amp;controls=0&amp;showinfo=0\" frameborder=\"0\" allowfullscreen></iframe>')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Install the Transformers and Datasets libraries to run this notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "! pip install datasets transformers[sentencepiece]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset, load_metric\n",
    "from transformers import (\n",
    "    AutoTokenizer,\n",
    "    AutoModelForSequenceClassification,\n",
    "    TrainingArguments,\n",
    "    Trainer,\n",
    ")\n",
    "\n",
    "raw_datasets = load_dataset(\"glue\", \"mnli\")\n",
    "model_checkpoint = \"distilbert-base-uncased\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)\n",
    "\n",
    "def preprocess_function(examples):\n",
    "    return tokenizer(examples[\"premise\"], examples[\"hypothesis\"], truncation=True)\n",
    "\n",
    "tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)\n",
    "args = TrainingArguments(\n",
    "    \"distilbert-finetuned-mnli\",\n",
    "    evaluation_strategy=\"epoch\",\n",
    "    save_strategy=\"epoch\",\n",
    "    learning_rate=2e-5,\n",
    "    num_train_epochs=3,\n",
    "    weight_decay=0.01,\n",
    ")\n",
    "metric = load_metric(\"glue\", \"mnli\")\n",
    "\n",
    "def compute_metrics(eval_pred):\n",
    "    predictions, labels = eval_pred\n",
    "    predictions = np.argmax(predictions, axis=1)\n",
    "    return metric.compute(\n",
    "        predictions=predictions, references=labels\n",
    "    )\n",
    "\n",
    "trainer = Trainer(\n",
    "    model,\n",
    "    args,\n",
    "    train_dataset=raw_datasets[\"train\"],\n",
    "    eval_dataset=raw_datasets[\"validation_matched\"],\n",
    "    compute_metrics=compute_metrics,\n",
    ")\n",
    "trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.train_dataset[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset, load_metric\n",
    "from transformers import (\n",
    "    AutoTokenizer,\n",
    "    AutoModelForSequenceClassification,\n",
    "    TrainingArguments,\n",
    "    Trainer,\n",
    ")\n",
    "\n",
    "raw_datasets = load_dataset(\"glue\", \"mnli\")\n",
    "model_checkpoint = \"distilbert-base-uncased\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)\n",
    "\n",
    "def preprocess_function(examples):\n",
    "    return tokenizer(examples[\"premise\"], examples[\"hypothesis\"], truncation=True)\n",
    "\n",
    "tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)\n",
    "args = TrainingArguments(\n",
    "    \"distilbert-finetuned-mnli\",\n",
    "    evaluation_strategy=\"epoch\",\n",
    "    save_strategy=\"epoch\",\n",
    "    learning_rate=2e-5,\n",
    "    num_train_epochs=3,\n",
    "    weight_decay=0.01,\n",
    ")\n",
    "metric = load_metric(\"glue\", \"mnli\")\n",
    "\n",
    "def compute_metrics(eval_pred):\n",
    "    predictions, labels = eval_pred\n",
    "    predictions = np.argmax(predictions, axis=1)\n",
    "    return metric.compute(\n",
    "        predictions=predictions, references=labels\n",
    "    )\n",
    "\n",
    "trainer = Trainer(\n",
    "    model,\n",
    "    args,\n",
    "    train_dataset=tokenized_datasets[\"train\"],\n",
    "    eval_dataset=tokenized_datasets[\"validation_matched\"],\n",
    "    compute_metrics=compute_metrics,\n",
    ")\n",
    "trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for batch in trainer.get_train_dataloader():\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset, load_metric\n",
    "from transformers import (\n",
    "    AutoTokenizer,\n",
    "    AutoModelForSequenceClassification,\n",
    "    TrainingArguments,\n",
    "    Trainer,\n",
    ")\n",
    "\n",
    "raw_datasets = load_dataset(\"glue\", \"mnli\")\n",
    "model_checkpoint = \"distilbert-base-uncased\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)\n",
    "\n",
    "def preprocess_function(examples):\n",
    "    return tokenizer(examples[\"premise\"], examples[\"hypothesis\"], truncation=True)\n",
    "\n",
    "tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)\n",
    "args = TrainingArguments(\n",
    "    \"distilbert-finetuned-mnli\",\n",
    "    evaluation_strategy=\"epoch\",\n",
    "    save_strategy=\"epoch\",\n",
    "    learning_rate=2e-5,\n",
    "    num_train_epochs=3,\n",
    "    weight_decay=0.01,\n",
    ")\n",
    "metric = load_metric(\"glue\", \"mnli\")\n",
    "\n",
    "def compute_metrics(eval_pred):\n",
    "    predictions, labels = eval_pred\n",
    "    predictions = np.argmax(predictions, axis=1)\n",
    "    return metric.compute(\n",
    "        predictions=predictions, references=labels\n",
    "    )\n",
    "\n",
    "trainer = Trainer(\n",
    "    model,\n",
    "    args,\n",
    "    train_dataset=tokenized_datasets[\"train\"],\n",
    "    eval_dataset=tokenized_datasets[\"validation_matched\"],\n",
    "    compute_metrics=compute_metrics,\n",
    "    tokenizer=tokenizer,\n",
    ")\n",
    "trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for batch in trainer.get_train_dataloader():\n",
    "    break\n",
    "\n",
    "outputs = trainer.model.cpu()(**batch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.num_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset, load_metric\n",
    "from transformers import (\n",
    "    AutoTokenizer,\n",
    "    AutoModelForSequenceClassification,\n",
    "    TrainingArguments,\n",
    "    Trainer,\n",
    ")\n",
    "\n",
    "raw_datasets = load_dataset(\"glue\", \"mnli\")\n",
    "model_checkpoint = \"distilbert-base-uncased\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)\n",
    "\n",
    "def preprocess_function(examples):\n",
    "    return tokenizer(examples[\"premise\"], examples[\"hypothesis\"], truncation=True)\n",
    "\n",
    "tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)\n",
    "model = AutoModelForSequenceClassification.from_pretrained(\n",
    "    model_checkpoint, num_labels=3\n",
    ")\n",
    "args = TrainingArguments(\n",
    "    \"distilbert-finetuned-mnli\",\n",
    "    evaluation_strategy=\"epoch\",\n",
    "    save_strategy=\"epoch\",\n",
    "    learning_rate=2e-5,\n",
    "    num_train_epochs=3,\n",
    "    weight_decay=0.01,\n",
    ")\n",
    "metric = load_metric(\"glue\", \"mnli\")\n",
    "\n",
    "def compute_metrics(eval_pred):\n",
    "    predictions, labels = eval_pred\n",
    "    predictions = np.argmax(predictions, axis=1)\n",
    "    return metric.compute(\n",
    "        predictions=predictions, references=labels\n",
    "    )\n",
    "\n",
    "trainer = Trainer(\n",
    "    model,\n",
    "    args,\n",
    "    train_dataset=tokenized_datasets[\"train\"],\n",
    "    eval_dataset=tokenized_datasets[\"validation_matched\"],\n",
    "    compute_metrics=compute_metrics,\n",
    "    tokenizer=tokenizer,\n",
    ")\n",
    "trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for batch in trainer.get_train_dataloader():\n",
    "    break\n",
    "\n",
    "outputs = trainer.model.cpu()(**batch)\n",
    "loss = outputs.loss\n",
    "loss.backward()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainer.create_optimizer()\n",
    "trainer.optimizer.step()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "name": "Debugging the Training Pipeline (PyTorch)",
   "provenance": []
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}