Autonomous Exploration on TextCraft
8.7StepsQwen2.5-7B+GRPO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Qwen2.5-7B+GRPOModel Category=Open-Source Models, Training=Task-oriented GRPO2026.05 | 8.7 | 11.3 | 2.1 | |
| Qwen3-4B+GRPOModel Category=Open-Source Models, Training=Task-oriented GRPO2026.05 | 14.5 | 10.8 | 0.2 | |
| Qwen3-4BModel Category=Open-Source Models2026.05 | 21.9 | 20.6 | 3.4 | |
| GPT-4.1Model Category=Closed-Source Models2026.05 | 31.4 | 57.6 | 4.3 | |
| Qwen2.5-7BModel Category=Open-Source Models2026.05 | 50.8 | 15.2 | 1.1 | |
| LLaMA3.1-8BModel Category=Open-Source Models2026.05 | 65.9 | 22.1 | 1.5 | |
| Claude-Opus-4.5Model Category=Closed-Source Models2026.05 | 97.3 | 82.5 | 7.8 |