Visual Question Answering 벤치마크
Visual Question Answering on PlotQA-D2
1:1 Accuracy
- 2022-12-19 — MatCha: 1:1 Accuracy 90.7
- 2024-01-01 — MatCha4096 + LaMenDa: 1:1 Accuracy 91.84
| Rank | Model | 1:1 Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | MatCha4096 + LaMenDa | 91.84 | Synthesize Step-by-Step: Tools Templates and LLMs as Data Generators for Reasoning-Based Chart VQA | 2024 | |
| 2 | MatCha | 90.7 | MatCha: Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering | huggingface/transformers | 2022 |
| 3 | MatCha4096 + LaMenDa | 91.84 | Synthesize Step-by-Step: Tools Templates and LLMs as Data Generators for Reasoning-Based Chart VQA | 2024 | |
| 4 | MatCha | 90.7 | MatCha: Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering | huggingface/transformers | 2022 |