LLaVA Training Scripts for SageMaker

Build the training image. The training image can be built on SageMaker Code Editor with instance type ml.g5.xlarge or on g5.xlarge EC2. For ECR usage, your can refer to https://docs.aws.amazon.com/AmazonECR/latest/userguide/docker-push-ecr-image.html
Prepare the training data and SageMaker Code Editor working directory (see llava-lora-finetuning-sagemaker.ipynb)
Run the llava-full-finetuning-sagemaker.ipynb or llava-lora-finetuning-sagemaker.ipynb to get the training job running on SageMaker

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
data		data
.gitignore		.gitignore
Dockerfile		Dockerfile
Dockerfile.bak		Dockerfile.bak
README.md		README.md
llava-full-finetuning-sagemaker.ipynb		llava-full-finetuning-sagemaker.ipynb
llava-lora-finetuning-sagemaker.ipynb		llava-lora-finetuning-sagemaker.ipynb
requirements.txt		requirements.txt

Provide feedback