MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915002295779328 |
|---|---|
| author | Zhang, Renrui Wei, Xinyu Jiang, Dongzhi Guo, Ziyu Li, Shicheng Zhang, Yichi Tong, Chengzhuo Liu, Jiaming Zhou, Aojun Wei, Bin Zhang, Shanghang Gao, Peng Li, Chunyuan Li, Hongsheng |
| author_facet | Zhang, Renrui Wei, Xinyu Jiang, Dongzhi Guo, Ziyu Li, Shicheng Zhang, Yichi Tong, Chengzhuo Liu, Jiaming Zhou, Aojun Wei, Bin Zhang, Shanghang Gao, Peng Li, Chunyuan Li, Hongsheng |
| contents | The mathematical capabilities of Multi-modal Large Language Models (MLLMs) remain under-explored with three areas to be improved: visual encoding of math diagrams, diagram-language alignment, and chain-of-thought (CoT) reasoning. This draws forth an urgent demand for an effective training paradigm and a large-scale, comprehensive dataset with detailed CoT rationales, which is challenging to collect and costly to annotate manually. To tackle this issue, we propose MAVIS, a MAthematical VISual instruction tuning pipeline for MLLMs, featuring an automatic data engine to efficiently create mathematical visual datasets. We design the data generation process to be entirely independent of human intervention or GPT API usage, while ensuring the diagram-caption correspondence, question-answer correctness, and CoT reasoning quality. With this approach, we curate two datasets, MAVIS-Caption (558K diagram-caption pairs) and MAVIS-Instruct (834K visual math problems with CoT rationales), and propose four progressive stages for training MLLMs from scratch. First, we utilize MAVIS-Caption to fine-tune a math-specific vision encoder (CLIP-Math) through contrastive learning, tailored for improved diagram visual encoding. Second, we also leverage MAVIS-Caption to align the CLIP-Math with a large language model (LLM) by a projection layer, enhancing vision-language alignment in mathematical domains. Third, we adopt MAVIS-Instruct to perform the instruction tuning for robust problem-solving skills, and term the resulting model as MAVIS-7B. Fourth, we apply Direct Preference Optimization (DPO) to enhance the CoT capabilities of our model, further refining its step-wise reasoning performance. Code and data will be released at https://github.com/ZrrSkywalker/MAVIS |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_08739 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine Zhang, Renrui Wei, Xinyu Jiang, Dongzhi Guo, Ziyu Li, Shicheng Zhang, Yichi Tong, Chengzhuo Liu, Jiaming Zhou, Aojun Wei, Bin Zhang, Shanghang Gao, Peng Li, Chunyuan Li, Hongsheng Computer Vision and Pattern Recognition The mathematical capabilities of Multi-modal Large Language Models (MLLMs) remain under-explored with three areas to be improved: visual encoding of math diagrams, diagram-language alignment, and chain-of-thought (CoT) reasoning. This draws forth an urgent demand for an effective training paradigm and a large-scale, comprehensive dataset with detailed CoT rationales, which is challenging to collect and costly to annotate manually. To tackle this issue, we propose MAVIS, a MAthematical VISual instruction tuning pipeline for MLLMs, featuring an automatic data engine to efficiently create mathematical visual datasets. We design the data generation process to be entirely independent of human intervention or GPT API usage, while ensuring the diagram-caption correspondence, question-answer correctness, and CoT reasoning quality. With this approach, we curate two datasets, MAVIS-Caption (558K diagram-caption pairs) and MAVIS-Instruct (834K visual math problems with CoT rationales), and propose four progressive stages for training MLLMs from scratch. First, we utilize MAVIS-Caption to fine-tune a math-specific vision encoder (CLIP-Math) through contrastive learning, tailored for improved diagram visual encoding. Second, we also leverage MAVIS-Caption to align the CLIP-Math with a large language model (LLM) by a projection layer, enhancing vision-language alignment in mathematical domains. Third, we adopt MAVIS-Instruct to perform the instruction tuning for robust problem-solving skills, and term the resulting model as MAVIS-7B. Fourth, we apply Direct Preference Optimization (DPO) to enhance the CoT capabilities of our model, further refining its step-wise reasoning performance. Code and data will be released at https://github.com/ZrrSkywalker/MAVIS |
| title | MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2407.08739 |