LLaVA Model Card

Model details

This is a fork from origianl liuhaotian/llava-v1.5-7b. This repo added code/inference.py and code/requirements.txt to provide customize inference script and environment for SageMaker deployment.

Model type: LLaVA is an open-source chatbot trained by fine-tuning LLaMA/Vicuna on GPT-generated multimodal instruction-following data. It is an auto-regressive language model, based on the transformer architecture.

Model date: LLaVA-v1.5-7B was trained in September 2023.

Paper or resources for more information: https://llava-vl.github.io/

How to Deploy on SageMaker

Following deploy_llava.ipynb (full tutorial here) , bundle llava model weights and code into a model.tar.gz and upload to S3:

from sagemaker.s3 import S3Uploader

# upload model.tar.gz to s3
s3_model_uri = S3Uploader.upload(local_path="./model.tar.gz", desired_s3_uri=f"s3://{sess.default_bucket()}/llava-v1.5-7b")

print(f"model uploaded to: {s3_model_uri}")

Then use HuggingfaceModel to deploy our real-time inference endpoint on SageMaker:

from sagemaker.huggingface.model import HuggingFaceModel

# create Hugging Face Model Class
huggingface_model = HuggingFaceModel(
   model_data=s3_model_uri,      # path to your model and script
   role=role,                    # iam role with permissions to create an Endpoint
   transformers_version="4.28.1",  # transformers version used
   pytorch_version="2.0.0",       # pytorch version used
   py_version='py310',            # python version used
   model_server_workers=1
)

# deploy the endpoint endpoint
predictor = huggingface_model.deploy(
    initial_instance_count=1,
    instance_type="ml.g5.xlarge",
)

Inference on SageMaker

Default conv_mode for llava-1.5 is setup as llava_v1 to process raw_prompt into meaningful prompt. You can also setup conv_mode as raw to directly use raw_prompt.

data = {
    "image" : 'https://raw.githubusercontent.com/haotian-liu/LLaVA/main/images/llava_logo.png', 
    "question" : "Describe the image and color details.",
    # "max_new_tokens" : 1024,
    # "temperature" : 0.2,
    # "conv_mode" : "llava_v1"
}
output = predictor.predict(data)
print(output)

Or use SageMakerRuntime to setup endpoint invoking client.

License

Where to send questions or comments about the model: https://github.com/haotian-liu/LLaVA/issues

Intended use

Primary intended uses: The primary use of LLaVA is research on large multimodal models and chatbots.

Primary intended users: The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.

Training dataset

558K filtered image-text pairs from LAION/CC/SBU, captioned by BLIP.
158K GPT-generated multimodal instruction-following data.
450K academic-task-oriented VQA data mixture.
40K ShareGPT data.

Evaluation dataset

A collection of 12 benchmarks, including 5 academic VQA benchmarks and 7 recent benchmarks specifically proposed for instruction-following LMMs.