{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compile and Train a Hugging Face Transformer BERT Model with the SST Dataset using SageMaker Training Compiler"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "This notebook's CI test result for us-west-2 is as follows. CI test results in other regions can be found at the end of the notebook. \n",
    "\n",
    "![This us-west-2 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/us-west-2/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "---"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. [Overview](#Overview)\n",
    "2. [Introduction](#Introduction)  \n",
    "3. [Prepare SageMaker Environment and Permissions](#Prepare-SageMaker-Environment-and-Permissions)\n",
    "    1. [Install libraries](#Install-libraries)  \n",
    "    2. [SageMaker environment](#SageMaker-environment)  \n",
    "    3. [Permissions](#Permissions)\n",
    "4. [Loading a dataset](#Loading-a-dataset)   \n",
    "    1. [Tokenization](#Tokenization)  \n",
    "    2. [Uploading data to SageMaker_session_bucket](#Uploading-data-to-sagemaker_session_bucket)  \n",
    "5. [SageMaker Training Job](#Sagemaker-Training-Job)  \n",
    "    1. [Training a PyTorch Trainer without compiling](#Training-a-PyTorch-Trainer-without-compiling)  \n",
    "    2. [Training a PyTorch Trainer with SageMaker Training Compiler](#Training-a-PyTorch-Trainer-with-SageMaker-Training-Compiler) \n",
    "6. [Analysis and Results](#Analysis-and-Results)  "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SageMaker Training Compiler Overview\n",
    "\n",
    "SageMaker Training Compiler is a capability of SageMaker that makes these hard-to-implement optimizations to reduce training time on GPU instances. The compiler optimizes DL models to accelerate training by more efficiently using SageMaker machine learning (ML) GPU instances. SageMaker Training Compiler is available at no additional charge within SageMaker and can help reduce total billable time as it accelerates training. \n",
    "\n",
    "SageMaker Training Compiler is integrated into the AWS Deep Learning Containers (DLCs). Using the SageMaker Training Compiler enabled AWS DLCs, you can compile and optimize training jobs on GPU instances with minimal changes to your code. Bring your deep learning models to SageMaker and enable SageMaker Training Compiler to accelerate the speed of your training job on SageMaker ML instances for accelerated computing. \n",
    "\n",
    "For more information, see [SageMaker Training Compiler](https://docs.aws.amazon.com/sagemaker/latest/dg/training-compiler.html) in the *Amazon SageMaker Developer Guide*."
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introduction\n",
    "\n",
    "This notebook is an end-to-end binary text classification example. In this demo, we use the Hugging Face's transformers and datasets libraries with SageMaker Training Compiler to compile and fine-tune a pre-trained transformer for binary text classification. In particular, the pre-trained model will be fine-tuned using the Stanford Sentiment Treebank (SST) dataset. To get started, you need to set up the environment with a few prerequisite steps, for permissions, configurations, and so on. \n",
    "\n",
    "![image.png](attachment:image.png)\n",
    "\n",
    "**NOTE:** You can run this demo in SageMaker Studio, SageMaker notebook instances, or your local machine with AWS CLI set up. If using SageMaker Studio or SageMaker notebook instances, make sure you choose one of the PyTorch-based kernels, Python 3 (PyTorch x.y Python 3.x CPU Optimized) or conda_pytorch_p36 respectively.\n",
    "\n",
    "**NOTE:** This notebook uses two ml.p3.2xlarge instances that have single GPU. If you don't have enough quota, see [Request a service quota increase for SageMaker resources](https://docs.aws.amazon.com/sagemaker/latest/dg/regions-quotas.html#service-limit-increase-request-procedure). "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare SageMaker Environment and Permissions "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Installation\n",
    "\n",
    "This example notebook requires the **SageMaker Python SDK v2.108.0** and **transformers v4.21**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install \"sagemaker>=2.133.0\" botocore boto3 awscli s3fs typing-extensions \"torch==1.12.0\" --upgrade"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install transformers datasets --upgrade"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import botocore\n",
    "import boto3\n",
    "import sagemaker\n",
    "import transformers\n",
    "\n",
    "print(f\"sagemaker: {sagemaker.__version__}\")\n",
    "print(f\"transformers: {transformers.__version__}\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Copy and run the following code if you need to upgrade ipywidgets for datasets library and restart kernel. This is only needed when preprocessing is done in the notebook.\n",
    "\n",
    "```python\n",
    "%%capture\n",
    "import IPython\n",
    "!conda install -c conda-forge ipywidgets -y\n",
    "# has to restart kernel for the updates to be applied\n",
    "IPython.Application.instance().kernel.do_shutdown(True) \n",
    "```"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SageMaker environment "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note:** If you are going to use SageMaker in a local environment. You need access to an IAM Role with the required permissions for SageMaker. To learn more, see [SageMaker Roles](https://docs.aws.amazon.com/sagemaker/latest/dg/sagemaker-roles.html)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sagemaker\n",
    "\n",
    "sess = sagemaker.Session()\n",
    "# sagemaker session bucket -> used for uploading data, models and logs\n",
    "# sagemaker will automatically create this bucket if it not exists\n",
    "sagemaker_session_bucket = None\n",
    "if sagemaker_session_bucket is None and sess is not None:\n",
    "    # set to default bucket if a bucket name is not given\n",
    "    sagemaker_session_bucket = sess.default_bucket()\n",
    "\n",
    "role = sagemaker.get_execution_role()\n",
    "sess = sagemaker.Session(default_bucket=sagemaker_session_bucket)\n",
    "\n",
    "print(f\"sagemaker role arn: {role}\")\n",
    "print(f\"sagemaker bucket: {sess.default_bucket()}\")\n",
    "print(f\"sagemaker session region: {sess.boto_region_name}\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading the SST dataset\n",
    "\n",
    "When using the [🤗 Datasets library](https://github.com/huggingface/datasets), datasets can be downloaded directly with the following `datasets.load_dataset()` method:\n",
    "\n",
    "```python\n",
    "from datasets import load_dataset\n",
    "load_dataset('dataset_name')\n",
    "```\n",
    "\n",
    "If you'd like to try other training datasets later, you can simply use this method.\n",
    "\n",
    "For this example notebook, we prepared the [SST2 dataset](https://www.tensorflow.org/datasets/catalog/glue#gluesst2) in the public SageMaker sample S3 bucket. The following code cells show how you can directly load the dataset and convert to a HuggingFace DatasetDict."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tokenization "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset\n",
    "from transformers import AutoTokenizer\n",
    "from datasets import Dataset\n",
    "\n",
    "# tokenizer used in preprocessing\n",
    "tokenizer_name = \"bert-base-cased\"\n",
    "\n",
    "# dataset used\n",
    "dataset_name = \"sst\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# Read our data into pandas DataFrames to prepare it for training\n",
    "# Initially our dataset will contain one column called 'line'\n",
    "\n",
    "test_df = pd.read_csv(\n",
    "    f\"https://sagemaker-example-files-prod-{sess.boto_region_name}.s3.amazonaws.com/datasets/text/SST2/sst2.test\",\n",
    "    sep=\"delimiter\",\n",
    "    header=None,\n",
    "    engine=\"python\",\n",
    "    names=[\"line\"],\n",
    ")\n",
    "train_df = pd.read_csv(\n",
    "    f\"https://sagemaker-example-files-prod-{sess.boto_region_name}.s3.amazonaws.com/datasets/text/SST2/sst2.train\",\n",
    "    sep=\"delimiter\",\n",
    "    header=None,\n",
    "    engine=\"python\",\n",
    "    names=[\"line\"],\n",
    ")\n",
    "val_df = pd.read_csv(\n",
    "    f\"https://sagemaker-example-files-prod-{sess.boto_region_name}.s3.amazonaws.com/datasets/text/SST2/sst2.val\",\n",
    "    sep=\"delimiter\",\n",
    "    header=None,\n",
    "    engine=\"python\",\n",
    "    names=[\"line\"],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split each row into two columns one for the label and one for the text\n",
    "\n",
    "test_df[[\"label\", \"text\"]] = test_df[\"line\"].str.split(\" \", 1, expand=True)\n",
    "train_df[[\"label\", \"text\"]] = train_df[\"line\"].str.split(\" \", 1, expand=True)\n",
    "val_df[[\"label\", \"text\"]] = val_df[\"line\"].str.split(\" \", 1, expand=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Drop the line column as it is no longer needed since we split this column above\n",
    "\n",
    "test_df.drop(\"line\", axis=1, inplace=True)\n",
    "train_df.drop(\"line\", axis=1, inplace=True)\n",
    "val_df.drop(\"line\", axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert the label into numeric instead of object type to prepare binary labels for training\n",
    "# After conversion the labels will be 8 bit integers\n",
    "\n",
    "test_df[\"label\"] = pd.to_numeric(test_df[\"label\"], downcast=\"integer\")\n",
    "train_df[\"label\"] = pd.to_numeric(train_df[\"label\"], downcast=\"integer\")\n",
    "val_df[\"label\"] = pd.to_numeric(val_df[\"label\"], downcast=\"integer\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Combine test and val data into one df\n",
    "\n",
    "test_df = pd.concat([test_df, val_df], ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset = Dataset.from_pandas(train_df)\n",
    "test_dataset = Dataset.from_pandas(test_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Download tokenizer\n",
    "# The tokenizer is loaded from the AutoTokenizer class and we use the from_pretrained method\n",
    "# This allows us to instatiate a tokenizer based on a pretrained model\n",
    "tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)\n",
    "\n",
    "\n",
    "# Tokenizer helper function\n",
    "# This function specifies the input should be tokenized by padding to the max_length which is 512\n",
    "# Anything beyond this length will be truncated\n",
    "# This function will convert the 'text' column to a set of numeric input ids that can be used for\n",
    "# model training\n",
    "# For more information on tokenization see the Hugging Face documentation: https://huggingface.co/transformers/preprocessing.html\n",
    "def tokenize(batch):\n",
    "    return tokenizer(batch[\"text\"], padding=\"max_length\", truncation=True)\n",
    "\n",
    "\n",
    "# Tokenize dataset in batches\n",
    "# See Hugging Face documentation for more info on the map method: https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.map\n",
    "train_dataset = train_dataset.map(tokenize, batched=True)\n",
    "test_dataset = test_dataset.map(tokenize, batched=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset = train_dataset.rename_column(\"label\", \"labels\")\n",
    "train_dataset.set_format(\"torch\", columns=[\"input_ids\", \"attention_mask\", \"labels\"])\n",
    "test_dataset = test_dataset.rename_column(\"label\", \"labels\")\n",
    "test_dataset.set_format(\"torch\", columns=[\"input_ids\", \"attention_mask\", \"labels\"])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Uploading data to `sagemaker_session_bucket`\n",
    "\n",
    "After we processed the datasets we are going to use the new FileSystem [integration](https://huggingface.co/docs/datasets/filesystems.html) to upload our dataset to S3."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import botocore\n",
    "from datasets.filesystems import S3FileSystem\n",
    "\n",
    "s3 = S3FileSystem()\n",
    "\n",
    "# s3 key prefix for setting up the data channel for the current SageMaker session\n",
    "s3_prefix = \"samples/datasets/sst\"\n",
    "\n",
    "# save train_dataset to s3\n",
    "training_input_path = f\"s3://{sess.default_bucket()}/{s3_prefix}/train\"\n",
    "train_dataset.save_to_disk(training_input_path, fs=s3)\n",
    "\n",
    "# save test_dataset to s3\n",
    "test_input_path = f\"s3://{sess.default_bucket()}/{s3_prefix}/test\"\n",
    "test_dataset.save_to_disk(test_input_path, fs=s3)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SageMaker Training Job\n",
    "\n",
    "To create a SageMaker training job, we use a PyTorch estimator. Using the estimator, you can define which fine-tuning script should SageMaker use through entry_point, which instance_type to use for training, which hyperparameters to pass, and so on.\n",
    "\n",
    "When a SageMaker training job starts, SageMaker takes care of starting and managing all the required machine learning instances, picks up the PyTorch Deep Learning Container, uploads your training script, and downloads the data from sagemaker_session_bucket into the container at /opt/ml/input/data.\n",
    "\n",
    "In the following section, you learn how to set up two versions of the SageMaker PyTorch estimator, a native one without the compiler and an optimized one with the compiler."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pygmentize ./scripts/train.py"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training a PyTorch Trainer without compiling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sagemaker.pytorch import PyTorch\n",
    "\n",
    "hyperparameters = {\"epochs\": 4, \"train_batch_size\": 24, \"model_name\": \"bert-base-cased\"}\n",
    "\n",
    "# Scale the learning rate by batch size, as original LR was using batch size of 32\n",
    "hyperparameters[\"learning_rate\"] = float(\"5e-5\") / 32 * hyperparameters[\"train_batch_size\"]\n",
    "\n",
    "# Scale the volume size by number of epochs\n",
    "volume_size = 60 + 2 * hyperparameters[\"epochs\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# By setting the hyperparameters in the PyTorch Estimator below\n",
    "# and using the AutoModelForSequenceClassification class in the train.py script\n",
    "# we can fine-tune the bert-base-cased pretrained Transformer for sequence classification\n",
    "\n",
    "native_estimator = PyTorch(\n",
    "    entry_point=\"train.py\",\n",
    "    source_dir=\"./scripts\",\n",
    "    instance_type=\"ml.p3.2xlarge\",\n",
    "    instance_count=1,\n",
    "    role=role,\n",
    "    py_version=\"py39\",\n",
    "    base_job_name=\"native-sst-bert-base-cased-p3-2x-pytorch-190\",\n",
    "    volume_size=volume_size,\n",
    "    framework_version=\"1.13.1\",\n",
    "    hyperparameters=hyperparameters,\n",
    "    disable_profiler=True,\n",
    "    debugger_hook_config=False,\n",
    ")\n",
    "\n",
    "# starting the train job with our uploaded datasets as input\n",
    "native_estimator.fit({\"train\": training_input_path, \"test\": test_input_path}, wait=False)\n",
    "\n",
    "# The name of the training job. You might need to note this down in case your kernel crashes.\n",
    "native_estimator.latest_training_job.name"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training a PyTorch Trainer with SageMaker Training Compiler"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compilation through Training Compiler changes the memory footprint of the model. Most commonly, this manifests as a reduction in memory utilization and a consequent increase in the largest batch size that can fit on the GPU. Note that if you want to change the batch size, you must adjust the learning rate appropriately.\n",
    "\n",
    "**Note:** We recommend you to turn the SageMaker Debugger's profiling and debugging tools off when you use compilation to avoid additional overheads.\n",
    "\n",
    "We use the tested batch size that's provided at [Tested Models](https://docs.aws.amazon.com/sagemaker/latest/dg/training-compiler-support.html#training-compiler-tested-models) in the *SageMaker Training Compiler Developer Guide*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sagemaker.pytorch import PyTorch, TrainingCompilerConfig\n",
    "\n",
    "hyperparameters = {\n",
    "    \"epochs\": 4,\n",
    "    \"train_batch_size\": 36,\n",
    "    \"model_name\": \"bert-base-cased\",\n",
    "}\n",
    "\n",
    "# Scale the learning rate by batch size, as original LR was using batch size of 32\n",
    "hyperparameters[\"learning_rate\"] = float(\"5e-5\") / 32 * hyperparameters[\"train_batch_size\"]\n",
    "\n",
    "# Scale the volume size by number of epochs\n",
    "volume_size = 60 + 2 * hyperparameters[\"epochs\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# By setting the hyperparameters in the PyTorch Estimator below\n",
    "# and using the AutoModelForSequenceClassification class in the train.py script\n",
    "# the bert-base-cased pretrained Transformer is fine-tuned for sequence classification\n",
    "\n",
    "# Importantly, the TrainingCompilerConfig() is passed below to enable the SageMaker Training Compiler\n",
    "\n",
    "sm_training_compiler_estimator = PyTorch(\n",
    "    entry_point=\"train.py\",\n",
    "    source_dir=\"./scripts\",\n",
    "    instance_type=\"ml.p3.2xlarge\",\n",
    "    instance_count=1,\n",
    "    role=role,\n",
    "    py_version=\"py39\",\n",
    "    base_job_name=\"sm-compiled-sst-bert-base-cased-p3-2x-pytorch-190\",\n",
    "    volume_size=volume_size,\n",
    "    framework_version=\"1.13.1\",\n",
    "    compiler_config=TrainingCompilerConfig(),\n",
    "    hyperparameters=hyperparameters,\n",
    "    disable_profiler=True,\n",
    "    debugger_hook_config=False,\n",
    ")\n",
    "\n",
    "# starting the train job with our uploaded datasets as input\n",
    "sm_training_compiler_estimator.fit(\n",
    "    {\"train\": training_input_path, \"test\": test_input_path}, wait=False\n",
    ")\n",
    "\n",
    "# The name of the training job. You might need to note this down in case your kernel crashes.\n",
    "sm_training_compiler_estimator.latest_training_job.name"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Wait for the training jobs to complete"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "waiter = native_estimator.sagemaker_session.sagemaker_client.get_waiter(\n",
    "    \"training_job_completed_or_stopped\"\n",
    ")\n",
    "waiter.wait(TrainingJobName=native_estimator.latest_training_job.name)\n",
    "waiter = sm_training_compiler_estimator.sagemaker_session.sagemaker_client.get_waiter(\n",
    "    \"training_job_completed_or_stopped\"\n",
    ")\n",
    "waiter.wait(TrainingJobName=sm_training_compiler_estimator.latest_training_job.name)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Analysis and Results"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load information and logs of the training job *without* SageMaker Training Compiler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# container image used for native training job\n",
    "print(f\"container image used for training job: \\n{native_estimator.image_uri}\\n\")\n",
    "\n",
    "# s3 uri where the native trained model is located\n",
    "print(f\"s3 uri where the trained model is located: \\n{native_estimator.model_data}\\n\")\n",
    "\n",
    "# latest training job name for this estimator\n",
    "print(\n",
    "    f\"latest training job name for this estimator: \\n{native_estimator.latest_training_job.name}\\n\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "%%capture native\n",
    "\n",
    "# access the logs of the native training job\n",
    "native_estimator.sagemaker_session.logs_for_job(native_estimator.latest_training_job.name)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note:** If the estimator object is no longer available due to a kernel break or refresh, you need to directly use the training job name and manually attach the training job to a new PyTorch estimator. For example:\n",
    "```python\n",
    "native_estimator = PyTorch.attach(\"your_native_training_job_name\")\n",
    "```"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load information and logs of the training job *with* SageMaker Training Compiler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# container image used for optimized training job\n",
    "print(f\"container image used for training job: \\n{sm_training_compiler_estimator.image_uri}\\n\")\n",
    "\n",
    "# s3 uri where the optimized trained model is located\n",
    "print(f\"s3 uri where the trained model is located: \\n{sm_training_compiler_estimator.model_data}\\n\")\n",
    "\n",
    "# latest training job name for this estimator\n",
    "print(\n",
    "    f\"latest training job name for this estimator: \\n{sm_training_compiler_estimator.latest_training_job.name}\\n\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture optimized\n",
    "\n",
    "# access the logs of the optimized training job\n",
    "sm_training_compiler_estimator.sagemaker_session.logs_for_job(\n",
    "    sm_training_compiler_estimator.latest_training_job.name\n",
    ")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note:** If the estimator object is no longer available due to a kernel break or refresh, you need to directly use the training job name and manually attach the training job to a new PyTorch estimator. You may be able to retrieve the training job name from above where it was printed out, but the name can also be retrieved using the SageMaker Service Console, the SageMaker SDK, or the AWS CLI (https://docs.aws.amazon.com/cli/latest/reference/sagemaker/list-training-jobs.html)  \n",
    "\n",
    "For example:\n",
    "```python\n",
    "sm_training_compiler_estimator = PyTorch.attach(\"your_compiled_huggingface_training_job_name\")\n",
    "```"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create helper functions for analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from ast import literal_eval\n",
    "from collections import defaultdict\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "\n",
    "# Intermediary function for processing each line of stdout captured\n",
    "# Remove leading and trailing whitespace and append data in curly braces\n",
    "# to final list\n",
    "def _summarize(captured):\n",
    "    final = []\n",
    "    for line in captured.stdout.split(\"\\n\"):\n",
    "        cleaned = line.strip()\n",
    "        if \"{\" in cleaned and \"}\" in cleaned:\n",
    "            final.append(cleaned[cleaned.index(\"{\") : cleaned.index(\"}\") + 1])\n",
    "    return final\n",
    "\n",
    "\n",
    "# Check input with literal_eval\n",
    "# https://docs.python.org/3/library/ast.html\n",
    "def make_sense(string):\n",
    "    try:\n",
    "        return literal_eval(string)\n",
    "    except:\n",
    "        pass\n",
    "\n",
    "\n",
    "# Parse the stdout and organize by train, evaluation, and summary data\n",
    "def summarize(summary):\n",
    "    final = {\"train\": [], \"eval\": [], \"summary\": {}}\n",
    "    for line in summary:\n",
    "        interpretation = make_sense(line)\n",
    "        if interpretation:\n",
    "            if \"loss\" in interpretation:\n",
    "                final[\"train\"].append(interpretation)\n",
    "            elif \"eval_loss\" in interpretation:\n",
    "                final[\"eval\"].append(interpretation)\n",
    "            elif \"train_runtime\" in interpretation:\n",
    "                final[\"summary\"].update(interpretation)\n",
    "    return final"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot and compare throughput of compiled training and native training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "native_summary = summarize(_summarize(native))\n",
    "native_throughput = native_summary[\"summary\"][\"train_samples_per_second\"]\n",
    "\n",
    "optimized_summary = summarize(_summarize(optimized))\n",
    "optimized_throughput = optimized_summary[\"summary\"][\"train_samples_per_second\"]\n",
    "\n",
    "avg_speedup = f\"{round((optimized_throughput/native_throughput-1)*100)}%\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "native_summary[\"summary\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimized_summary[\"summary\"]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training Throughput Plot\n",
    "\n",
    "The following script creates a plot that compares the throughput (number_of_samples/second) of the two training jobs with and without SageMaker Training Compiler."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "plt.title(\"Training Throughput \\n (Higher is better)\")\n",
    "plt.ylabel(\"Samples/sec\")\n",
    "\n",
    "plt.bar(x=[1], height=native_throughput, label=\"Baseline PT\", width=0.35)\n",
    "plt.bar(x=[1.5], height=optimized_throughput, label=\"SM-Training-Compiler-enhanced PT\", width=0.35)\n",
    "\n",
    "plt.xlabel(\"  ====> {} SM-Training-Compiler savings <====\".format(avg_speedup))\n",
    "plt.xticks(ticks=[1, 1.5], labels=[\"Baseline PT\", \"SM-Training-Compiler-enhanced PT\"])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Convergence of Training Loss\n",
    "\n",
    "The following script creates a plot that compares the loss function of the two training jobs with and without SageMaker Training Compiler."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "vanilla_loss = [i[\"loss\"] for i in native_summary[\"train\"]]\n",
    "vanilla_epochs = [i[\"epoch\"] for i in native_summary[\"train\"]]\n",
    "optimized_loss = [i[\"loss\"] for i in optimized_summary[\"train\"]]\n",
    "optimized_epochs = [i[\"epoch\"] for i in optimized_summary[\"train\"]]\n",
    "\n",
    "plt.title(\"Plot of Training Loss\")\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.ylabel(\"Training Loss\")\n",
    "plt.plot(vanilla_epochs, vanilla_loss, label=\"Baseline PT\")\n",
    "plt.plot(optimized_epochs, optimized_loss, label=\"SM-Training-Compiler-enhanced PT\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "attachments": {
    "trainingloss.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Convergence Example Plot\n",
    "\n",
    "![trainingloss.png](attachment:trainingloss.png)\n",
    "\n",
    "**Note:** For this example, due to the larger batch size that can be accommodated by the SageMaker Training Compiler, the initial decrease in training loss is greater when the SageMaker Training Compiler is enabled"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create table of summary results including loss, accuracy, f1 score, precision, recall, and steps per second\n",
    "# for both native and optimized training jobs\n",
    "\n",
    "table = pd.DataFrame(\n",
    "    [native_summary[\"eval\"][-1], optimized_summary[\"eval\"][-1]],\n",
    "    index=[\"Baseline PT\", \"SM-Training-Compiler-enhanced PT\"],\n",
    ")\n",
    "table.drop(columns=[\"eval_runtime\", \"eval_samples_per_second\", \"epoch\"])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clean up\n",
    "\n",
    "Stop all training jobs launched if the jobs are still running."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import boto3\n",
    "\n",
    "sm = boto3.client(\"sagemaker\")\n",
    "\n",
    "\n",
    "def stop_training_job(name):\n",
    "    status = sm.describe_training_job(TrainingJobName=name)[\"TrainingJobStatus\"]\n",
    "    if status == \"InProgress\":\n",
    "        sm.stop_training_job(TrainingJobName=name)\n",
    "\n",
    "\n",
    "stop_training_job(native_estimator.latest_training_job.name)\n",
    "stop_training_job(sm_training_compiler_estimator.latest_training_job.name)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Also, to find instructions on cleaning up resources, see [Clean Up](https://docs.aws.amazon.com/sagemaker/latest/dg/ex1-cleanup.html) in the *Amazon SageMaker Developer Guide*."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Notebook CI Test Results\n",
    "\n",
    "This notebook was tested in multiple regions. The test results are as follows, except for us-west-2 which is shown at the top of the notebook.\n",
    "\n",
    "![This us-east-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/us-east-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This us-east-2 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/us-east-2/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This us-west-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/us-west-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ca-central-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ca-central-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This sa-east-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/sa-east-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This eu-west-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/eu-west-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This eu-west-2 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/eu-west-2/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This eu-west-3 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/eu-west-3/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This eu-central-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/eu-central-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This eu-north-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/eu-north-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ap-southeast-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ap-southeast-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ap-southeast-2 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ap-southeast-2/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ap-northeast-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ap-northeast-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ap-northeast-2 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ap-northeast-2/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n",
    "\n",
    "![This ap-south-1 badge failed to load. Check your device's internet connectivity, otherwise the service is currently unavailable](https://h75twx4l60.execute-api.us-west-2.amazonaws.com/sagemaker-nb/ap-south-1/sagemaker-training-compiler|huggingface|pytorch_single_gpu_single_node|bert-base-cased|bert-base-cased-single-node-single-gpu.ipynb)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (Data Science 3.0)",
   "language": "python",
   "name": "python3__SAGEMAKER_INTERNAL__arn:aws:sagemaker:us-east-1:081325390199:image/sagemaker-data-science-310-v1"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.6"
  },
  "vscode": {
   "interpreter": {
    "hash": "11bd1a5e22b55030e7dafa6146ff58da0a8819c60346681265e3e4937faaddb3"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}