Autonomous LLM Fine-tuning on oMeBench
48.4oMeScoreTREX
Evaluation results
| Method | Links | |
|---|---|---|
| TREXResearcher Backend=Gemini 3 Pro, Base Model Scale=Qwen3-1.7B2026.04 | 48.4 | |
| TREXResearcher Backend=Qwen3-Next-80B-Thinking, Base Model Scale=Qwen3-1.7B2026.04 | 39.2 | |
| Qwen3-235B-2507Model Category=Ref Model2026.04 | 28.3 | |
| Qwen3-1.7BModel Category=Base Model2026.04 | 19.8 |