Character motivation recognition on CroSS-MR
68.24AccuracyClaude Sonnet 4.5
Evaluation results
| Method | Links | |
|---|---|---|
| Claude Sonnet 4.52026.01 | 68.24 | |
| HumanLLM-32BParameters=32B2026.01 | 64.27 | |
| Qwen3-32BParameters=32B2026.01 | 63.37 | |
| GPT-52026.01 | 62.25 | |
| Qwen3-8B (OT+CoSER)Parameters=8B, Training Data=OpenThoughts + CoSER2026.01 | 54.98 | |
| HumanLLM-8BParameters=8B2026.01 | 54.23 | |
| Qwen3-8BParameters=8B2026.01 | 53.26 |