Precursor Prediction on Inorganic material synthesis dataset (Split 1)
71.02Top-1 AccuracyMSP-LLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MSP-LLMBackbone=Qwen-2.5-7B2026.02 | 71.02 | 85.37 | 90.49 | 94.06 | |
| MSP-LLMBackbone=LLaMA-3.1-8B2026.02 | 70.75 | 85.56 | 91.68 | 94.97 | |
| GPT-5.1Shots=20-shot2026.02 | 64.72 | 77.42 | 80.53 | 83.27 | |
| GPT-5.1Shots=40-shot2026.02 | 64.08 | 76.33 | 79.62 | 83.46 | |
| GPT-4oShots=40-shot2026.02 | 61.79 | 70.48 | 73.4 | 76.32 | |
| Retrieval-Retro2026.02 | 60.88 | 69.38 | 70.93 | 74.13 | |
| GPT-4oShots=20-shot2026.02 | 60.05 | 68.56 | 71.76 | 75.05 | |
| Claude Haiku 4.5Shots=40-shot2026.02 | 57.31 | 68.74 | 74.31 | 78.43 | |
| GPT-4o-miniShots=20-shot2026.02 | 56.58 | 65.36 | 67.82 | 70.66 | |
| GPT-4o-miniShots=40-shot2026.02 | 55.21 | 63.44 | 66.27 | 69.1 | |
| Claude Haiku 4.5Shots=20-shot2026.02 | 55.03 | 68.37 | 74.5 | 78.88 | |
| He et al.2026.02 | 53.11 | 60.05 | 61.06 | 63.07 |