Prompt Adherence (test)
69.5Exact MatchDB-3DME
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DB-3DMEtraining_status=fine-tuned, base_model=Qwen-2.5-VL-7B2026.06 | 69.5 | 96.9 | 0.583 | |
| ChatGPT-5-hightraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 67.9 | 96.9 | 0.542 | |
| ChatGPT-5-lowtraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 66.2 | 95 | 0.489 | |
| ChatGPT-5-mediumtraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 65.4 | 96.4 | 0.499 | |
| Gemini-2.5-protraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 65.1 | 95.2 | 0.49 | |
| ChatGPT-4otraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 64.7 | 95.4 | 0.463 | |
| ChatGPT-4.1training_status=pre-trained, system_prompt=best of V1/V22026.06 | 60.8 | 95.6 | 0.445 | |
| Qwen-2.5-VL-7Btraining_status=pre-trained, system_prompt=best of V1/V22026.06 | 50.4 | 86.8 | 0.207 |