VISCO: Benchmarking Fine-Grained Critique and Correction Towards Self-Improvement in Visual Reasoning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909540349378560 |
|---|---|
| author | Wu, Xueqing Ding, Yuheng Li, Bingxuan Lu, Pan Yin, Da Chang, Kai-Wei Peng, Nanyun |
| author_facet | Wu, Xueqing Ding, Yuheng Li, Bingxuan Lu, Pan Yin, Da Chang, Kai-Wei Peng, Nanyun |
| contents | The ability of large vision-language models (LVLMs) to critique and correct their reasoning is an essential building block towards their self-improvement. However, a systematic analysis of such capabilities in LVLMs is still lacking. We propose VISCO, the first benchmark to extensively analyze the fine-grained critique and correction capabilities of LVLMs. Compared to existing work that uses a single scalar value to critique the entire reasoning [4], VISCO features dense and fine-grained critique, requiring LVLMs to evaluate the correctness of each step in the chain-of-thought and provide natural language explanations to support their judgments. Extensive evaluation of 24 LVLMs demonstrates that human-written critiques significantly enhance the performance after correction, showcasing the potential of the self-improvement strategy. However, the model-generated critiques are less helpful and sometimes detrimental to the performance, suggesting that critique is the crucial bottleneck. We identified three common patterns in critique failures: failure to critique visual perception, reluctance to "say no", and exaggerated assumption of error propagation. To address these issues, we propose an effective LookBack strategy that revisits the image to verify each piece of information in the initial reasoning. LookBack significantly improves critique and correction performance by up to 13.5%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_02172 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | VISCO: Benchmarking Fine-Grained Critique and Correction Towards Self-Improvement in Visual Reasoning Wu, Xueqing Ding, Yuheng Li, Bingxuan Lu, Pan Yin, Da Chang, Kai-Wei Peng, Nanyun Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language The ability of large vision-language models (LVLMs) to critique and correct their reasoning is an essential building block towards their self-improvement. However, a systematic analysis of such capabilities in LVLMs is still lacking. We propose VISCO, the first benchmark to extensively analyze the fine-grained critique and correction capabilities of LVLMs. Compared to existing work that uses a single scalar value to critique the entire reasoning [4], VISCO features dense and fine-grained critique, requiring LVLMs to evaluate the correctness of each step in the chain-of-thought and provide natural language explanations to support their judgments. Extensive evaluation of 24 LVLMs demonstrates that human-written critiques significantly enhance the performance after correction, showcasing the potential of the self-improvement strategy. However, the model-generated critiques are less helpful and sometimes detrimental to the performance, suggesting that critique is the crucial bottleneck. We identified three common patterns in critique failures: failure to critique visual perception, reluctance to "say no", and exaggerated assumption of error propagation. To address these issues, we propose an effective LookBack strategy that revisits the image to verify each piece of information in the initial reasoning. LookBack significantly improves critique and correction performance by up to 13.5%. |
| title | VISCO: Benchmarking Fine-Grained Critique and Correction Towards Self-Improvement in Visual Reasoning |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2412.02172 |