Can I read Mm-Lima: A 200-Instruction Paradigm For Fine-Tuning Minigpt-4 on EtoBox?

Mm-Lima: A 200-Instruction Paradigm For Fine-Tuning Minigpt-4 by Anirudha Vishvakarma is a document available to read on EtoBox.

What is Mm-Lima: A 200-Instruction Paradigm For Fine-Tuning Minigpt-4 about?

The paper introduces MM-LIMA, a multimodal large language model fine-tuned on a limited dataset of 200 high-quality instruction examples, which is only 6% of the data used for MiniGPT-4. It emphasizes the importance of data quality over quantity in training, demonstrating that MM-LIMA outperforms MiniGPT-4 across various multimodal tasks. The study also presents new metrics for assessing the quality of multimodal instruction data and a data selector to filter low-quality instances effectively.

Author

Anirudha Vishvakarma

Language

EN