Next-state Prediction Accuracy on ALFWorld (AW)
99.87EM AccuracyQwen2.5-7B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2.5-7BEvaluation Protocol=SFT2025.12 | 99.87 | |
| Llama3.1-8BEvaluation Protocol=SFT2025.12 | 99.71 | |
| Claude-sonnet-4.5Evaluation Protocol=Few-shot (3 shot)2025.12 | 77.04 | |
| GPT-5Evaluation Protocol=Few-shot (3 shot)2025.12 | 67.13 | |
| Claude-sonnet-4.5Evaluation Protocol=Zero-shot2025.12 | 64.73 | |
| GPT-4o-miniEvaluation Protocol=Few-shot (3 shot)2025.12 | 63.79 | |
| GPT-4.1Evaluation Protocol=Few-shot (3 shot)2025.12 | 63.37 | |
| GPT-4-turboEvaluation Protocol=Few-shot (3 shot)2025.12 | 62.56 | |
| Gemini-2.5-flashEvaluation Protocol=Few-shot (3 shot)2025.12 | 61.85 | |
| GPT-4oEvaluation Protocol=Few-shot (3 shot)2025.12 | 56.88 | |
| Gemini-2.5-flashEvaluation Protocol=Zero-shot2025.12 | 50 | |
| GPT-4o-miniEvaluation Protocol=Zero-shot2025.12 | 45.2 | |
| GPT-4oEvaluation Protocol=Zero-shot2025.12 | 44.45 | |
| GPT-4.1Evaluation Protocol=Zero-shot2025.12 | 43.56 | |
| GPT-4-turboEvaluation Protocol=Zero-shot2025.12 | 42.64 | |
| GPT-5Evaluation Protocol=Zero-shot2025.12 | 35.09 |