gray311 / UnifiedMultimodalInstructionTuning

🐙 Octopus:

Octopus: Unifiying Multimodal Instruction Format for Enhancing In-context Learning. Codes and details will be released soon.

• Foundation Model: OpenFlamingo
• Visual Encoder: ViT-L/14
• Pre-trained LLM: MPT-7B

Demo: Demo link