unsloth-vision

What problem does it solve?

This Skill streamlines the process of fine-tuning large multimodal vision-language models, enabling users to adapt them for specialized visual tasks without requiring extensive computational resources.

Core Features & Use Cases

• Optimized Fine-Tuning: Efficiently fine-tune models like Llama 3.2 Vision and Qwen2.5 VL.
• Selective Layer Training: Control which parts of the model (vision, language, or specific components) are updated.
• Image Handling: Includes specialized data collators to manage varying image sizes and optimize training.
• Use Case: Adapt a vision model to accurately describe medical images by fine-tuning its vision layers on a curated dataset.

Quick Start

Use the unsloth-vision skill to load and configure a vision model for fine-tuning.