metadata

library_name: transformers
license: apache-2.0
datasets:
  - lmms-lab/textvqa
language:
  - en
tags:
  - multimodal
  - vision
  - image-text-to-text

Idefics2-8B-SFT

Idefics2-8B-SFT is SFT fine-tune of HuggingFaceM4/idefics2-8b on 35k TextVQA dataset. Training was performed on RTX A5000 for 10 hrs. Wandb report:

This fine-tuned model achieves a Levenshtein score of 82.29%.

Model Summary

Developed by: Hugging Face
Model type: Multi-modal model (image+text)
Language(s) (NLP): en
License: Apache 2.0
Parent Models: google/siglip-so400m-patch14-384 and mistralai/Mistral-7B-v0.1

💻 Usage

processor = AutoProcessor.from_pretrained("abideen/idefics2-8b-sft")
model = AutoModelForVision2Seq.from_pretrained(
    "abideen/idefics2-8b-sft",
).to(DEVICE)

# Create inputs
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image"},
            {"type": "text", "text": "What do we see in this image?"},
        ]
    },
    {
        "role": "assistant",
        "content": [
            {"type": "text", "text": "In this image, we can see the city of New York, and more specifically the Statue of Liberty."},
        ]
    },
    {
        "role": "user",
        "content": [
            {"type": "image"},
            {"type": "text", "text": "And how about this image?"},
        ]
    },       
]
prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
inputs = processor(text=prompt, images=[image1, image2], return_tensors="pt")
inputs = {k: v.to(DEVICE) for k, v in inputs.items()}


# Generate
generated_ids = model.generate(**inputs, max_new_tokens=500)
generated_texts = processor.batch_decode(generated_ids, skip_special_tokens=True)

print(generated_texts)
# ['User: What do we see in this image? \nAssistant: In this image, we can see the city of New York, and more specifically the Statue of Liberty. \nUser: And how about this image? \nAssistant: In this image we can see buildings, trees, lights, water and sky.']

🏆 Evaluation

Coming Soon!