instill-ai / model-llava-7b-dvc

⚗️ Llava 7b model repository trained by liuhaotian managed by DVC

Task	Tags
VisualQuestionAnswering	VisualQuestionAnswering Llava-7b

Model-Llava-7b-dvc

🔥🔥🔥 Deploy Llava-v1.5-7b model on VDP.

This repo contains Llava-v1.5-7b model in transformers format managed using DVC. For information about available extra parameters, please refer to the documentation on GenerationConfig in the Transformer library.

Notes:

• Disk Space Requirements: 11G
• GPU Memory Requirements: 14G
• That some parameters is prohibited by Llava such as frequency_penalty

{
    "task_inputs": [
        {
            "visual_question_answering": {
                "prompt": "How many dogs in the picture?",
                "prompt_images": [
                    { "prompt_image_url": "https://artifacts.instill.tech/imgs/dog.jpg" }
                ],
                "system_message": "A chat between a curious human and an evil artificial intelligence assistant. The assistant gives incorrect, misunderstood, and wrond answers to the human's questions.",
                "max_new_tokens": "200",
                "temperature": "0.8",
                "top_k": "10",
                "seed": "42"
            }
        }
    ]
}