Numerical Parsing on PyraMathBench multimodal (test)
2.8Numerical Reasoning (NR)LLaVA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LLaVASize=13B2026.06 | 2.8 | 8.5 | 7 | |
| GPT-4o2026.06 | 12.3 | 6.6 | 1.9 | |
| GPT-4o mini2026.06 | 14.6 | 16.7 | 0 |
Loading the SOTA2 catalog…
| Method | Links | |||
|---|---|---|---|---|
| LLaVASize=13B2026.06 | 2.8 | 8.5 | 7 | |
| GPT-4o2026.06 | 12.3 | 6.6 | 1.9 | |
| GPT-4o mini2026.06 | 14.6 | 16.7 | 0 |