sagemaker-huggingface-inference-toolkit 2.3.0

Open source library for running inference workload with Hugging Face Deep Learning Containers on Amazon SageMaker.

SageMaker Hugging Face Inference Toolkit

SageMaker Hugging Face Inference Toolkit is an open-source library for serving 🤗 Transformers and Diffusers models on Amazon SageMaker. This library provides default pre-processing, predict and postprocessing for certain 🤗 Transformers and Diffusers models and tasks. It utilizes the SageMaker Inference Toolkit for starting up the model server, which is responsible for handling inference requests.

For Training, see Run training on Amazon SageMaker.

For the Dockerfiles used for building SageMaker Hugging Face Containers, see AWS Deep Learning Containers.

For information on running Hugging Face jobs on Amazon SageMaker, please refer to the 🤗 Transformers documentation.

For notebook examples: SageMaker Notebook Examples.

---

💻 Getting Started with 🤗 Inference Toolkit

needs to be adjusted -> currently pseudo code

Install Amazon SageMaker Python SDK

pip install sagemaker --upgrade

Create a Amazon SageMaker endpoint with a trained model.

from sagemaker.huggingface import HuggingFaceModel

# create Hugging Face Model Class
huggingface_model = HuggingFaceModel(
    transformers_version='4.6',
    pytorch_version='1.7',
    py_version='py36',
    model_data='s3://my-trained-model/artifacts/model.tar.gz',
    role=role,
)
# deploy model to SageMaker Inference
huggingface_model.deploy(initial_instance_count=1,instance_type="ml.m5.xlarge")

Create a Amazon SageMaker endpoint with a model from the 🤗 Hub.
note: This is an experimental feature, where the model will be loaded after the endpoint is created. Not all sagemaker features are supported, e.g. MME

from sagemaker.huggingface import HuggingFaceModel
# Hub Model configuration. https://huggingface.co/models
hub = {
  'HF_MODEL_ID':'distilbert-base-uncased-distilled-squad',
  'HF_TASK':'question-answering'
}
# create Hugging Face Model Class
huggingface_model = HuggingFaceModel(
    transformers_version='4.6',
    pytorch_version='1.7',
    py_version='py36',
    env=hub,
    role=role,
)
# deploy model to SageMaker Inference
huggingface_model.deploy(initial_instance_count=1,instance_type="ml.m5.xlarge")

---

🛠️ Environment variables

The SageMaker Hugging Face Inference Toolkit implements various additional environment variables to simplify your deployment experience. A full list of environment variables is given below.

HF_TASK

The HF_TASK environment variable defines the task for the used 🤗 Transformers pipeline. A full list of tasks can be find here.

HF_TASK="question-answering"

HF_MODEL_ID

The HF_MODEL_ID environment variable defines the model id, which will be automatically loaded from huggingface.co/models when creating or SageMaker Endpoint. The 🤗 Hub provides +10 000 models all available through this environment variable.

HF_MODEL_ID="distilbert-base-uncased-finetuned-sst-2-english"

HF_MODEL_REVISION

The HF_MODEL_REVISION is an extension to HF_MODEL_ID and allows you to define/pin a revision of the model to make sure you always load the same model on your SageMaker Endpoint.

HF_MODEL_REVISION="03b4d196c19d0a73c7e0322684e97db1ec397613"

HF_API_TOKEN

The HF_API_TOKEN environment variable defines the your Hugging Face authorization token. The HF_API_TOKEN is used as a HTTP bearer authorization for remote files, like private models. You can find your token at your settings page.

HF_API_TOKEN="api_XXXXXXXXXXXXXXXXXXXXXXXXXXXXX"

HF_TRUST_REMOTE_CODE

The HF_TRUST_REMOTE_CODE environment variable defines wether or not to allow for custom models defined on the Hub in their own modeling files. Allowed values are "True" and "False"

HF_TRUST_REMOTE_CODE="True"

HF_OPTIMUM_BATCH_SIZE

The HF_OPTIMUM_BATCH_SIZE environment variable defines the batch size, which is used when compiling the model to Neuron. The default value is 1. Not required when model is already converted.

HF_OPTIMUM_BATCH_SIZE="1"

HF_OPTIMUM_SEQUENCE_LENGTH

The HF_OPTIMUM_SEQUENCE_LENGTH environment variable defines the sequence length, which is used when compiling the model to Neuron. There is no default value. Not required when model is already converted.

HF_OPTIMUM_SEQUENCE_LENGTH="128"

---

🧑🏻‍💻 User defined code/modules

The Hugging Face Inference Toolkit allows user to override the default methods of the HuggingFaceHandlerService. Therefore, they need to create a folder named code/ with an inference.py file in it. You can find an example for it in sagemaker/17_customer_inference_script. For example:

model.tar.gz/
|- pytorch_model.bin
|- ....
|- code/
  |- inference.py
  |- requirements.txt 

In this example, pytorch_model.bin is the model file saved from training, inference.py is the custom inference module, and requirements.txt is a requirements file to add additional dependencies. The custom module can override the following methods:

• model_fn(model_dir): overrides the default method for loading the model, the return value model will be used in the predict() for predicitions. It receives argument the model_dir, the path to your unzipped model.tar.gz.
• transform_fn(model, data, content_type, accept_type): overrides the default transform function with a custom implementation. Customers using this would have to implement preprocess, predict and postprocess steps in the transform_fn. NOTE: This method can't be combined with input_fn, predict_fn or output_fn mentioned below.
• input_fn(input_data, content_type): overrides the default method for preprocessing, the return value data will be used in the predict() method for predicitions. The input is input_data, the raw body of your request and content_type, the content type form the request Header.
• predict_fn(processed_data, model): overrides the default method for predictions, the return value predictions will be used in the postprocess() method. The input is processed_data, the result of the preprocess() method.
• output_fn(prediction, accept): overrides the default method for postprocessing, the return value result will be the respond of your request(e.g.JSON). The inputs are predictions, the result of the predict() method and accept the return accept type from the HTTP Request, e.g. application/json

---

🤝 Contributing

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

---

📜 License

SageMaker Hugging Face Inference Toolkit is licensed under the Apache 2.0 License.

---

🧑🏻‍💻 Development Environment

Install all test and development packages with

pip3 install -e ".[test,dev]"

Run Model Locally

1. manually change MMS_CONFIG_FILE

wget -O sagemaker-mms.properties https://raw.githubusercontent.com/aws/deep-learning-containers/master/huggingface/build_artifacts/inference/config.properties

2. Run Container, e.g. text-to-image

HF_MODEL_ID="stabilityai/stable-diffusion-xl-base-1.0" HF_TASK="text-to-image" python src/sagemaker_huggingface_inference_toolkit/serving.py

3. Adjust handler_service.py and comment out if content_type in content_types.UTF8_TYPES: thats needed for SageMaker but cannot be used locally

4. Send request

curl --request POST \
  --url http://localhost:8080/invocations \
  --header 'Accept: image/png' \
  --header 'Content-Type: application/json' \
  --data '"{\"inputs\": \"Camera\"}" \
  --output image.png