Chemistry on USPTO
64.5AccuracyES + TT-MV
Evaluation Results
| Method | Links | |
|---|---|---|
| ES + TT-MVModel=Llama3.1-8B-Inst2026.03 | 64.5 | |
| ESModel=Qwen2.5-3B-Inst2026.03 | 52.9 | |
| GRPOModel=OLMo3-7B-Inst2026.03 | 51 | |
| PPOModel=OLMo3-7B2026.03 | 49.8 | |
| GRPOModel=Qwen2.5-3B-Inst2026.03 | 49.7 | |
| ESModel=OLMo3-7B2026.03 | 48.1 | |
| GRPOModel=OLMo3-7B2026.03 | 48 | |
| ES + TT-MVModel=OLMo3-7B-Inst2026.03 | 46.3 | |
| ESModel=OLMo3-7B-Inst2026.03 | 45.2 | |
| PPOModel=Qwen2.5-3B-Inst2026.03 | 44.7 | |
| Best-of-N‡Model=Qwen2.5-3B-Inst2026.03 | 44.3 | |
| RandOptModel=OLMo3-7B2026.03 | 44.3 | |
| TT-MV†Model=Qwen2.5-3B-Inst2026.03 | 43.2 | |
| RandOptModel=OLMo3-7B-Inst2026.03 | 43 | |
| RandOptModel=Qwen2.5-3B-Inst2026.03 | 42.3 | |
| RandOptModel=Llama3.1-8B-Inst2026.03 | 41 | |
| PPOModel=OLMo3-7B-Inst2026.03 | 40.2 | |
| ES + TT-MVModel=Qwen2.5-3B-Inst2026.03 | 39.8 | |
| BaseModel=Qwen2.5-3B-Inst2026.03 | 38.5 | |
| ES + TT-MVModel=OLMo3-7B2026.03 | 38.1 | |
| Best-of-N‡Model=OLMo3-7B2026.03 | 38 | |
| TT-MV†Model=OLMo3-7B2026.03 | 35.4 | |
| GRPOModel=Llama3.1-8B-Inst2026.03 | 35 | |
| RandOptModel=Qwen2.5-1.5B-Inst2026.03 | 34.3 | |
| ESModel=Llama3.1-8B-Inst2026.03 | 34.1 | |
| TT-MV†Model=Qwen2.5-1.5B-Inst2026.03 | 33 | |
| PPOModel=Llama3.1-8B-Inst2026.03 | 32.9 | |
| Best-of-N‡Model=Qwen2.5-1.5B-Inst2026.03 | 32.7 | |
| ES + TT-MVModel=Qwen2.5-0.5B-Inst2026.03 | 32.5 | |
| RandOptModel=Qwen2.5-0.5B-Inst2026.03 | 32.2 | |
| ES + TT-MVModel=Qwen2.5-1.5B-Inst2026.03 | 32.2 | |
| ESModel=Qwen2.5-1.5B-Inst2026.03 | 32 | |
| Best-of-N‡Model=OLMo3-7B-Inst2026.03 | 32 | |
| PPOModel=Qwen2.5-1.5B-Inst2026.03 | 31.9 | |
| GRPOModel=Qwen2.5-1.5B-Inst2026.03 | 31.8 | |
| GRPOModel=Qwen2.5-0.5B-Inst2026.03 | 31.7 | |
| ESModel=Qwen2.5-0.5B-Inst2026.03 | 31.6 | |
| PPOModel=Qwen2.5-0.5B-Inst2026.03 | 31.2 | |
| TT-MV†Model=Qwen2.5-0.5B-Inst2026.03 | 30.2 | |
| BaseModel=Qwen2.5-1.5B-Inst2026.03 | 30.2 | |
| TT-MV†Model=OLMo3-7B-Inst2026.03 | 30.2 | |
| BaseModel=Qwen2.5-0.5B-Inst2026.03 | 29.9 | |
| BaseModel=OLMo3-7B2026.03 | 29 | |
| Best-of-N‡Model=Llama3.1-8B-Inst2026.03 | 28 | |
| BaseModel=OLMo3-7B-Inst2026.03 | 27.2 | |
| Best-of-N‡Model=Qwen2.5-0.5B-Inst2026.03 | 26.8 | |
| TT-MV†Model=Llama3.1-8B-Inst2026.03 | 25.4 | |
| BaseModel=Llama3.1-8B-Inst2026.03 | 19.5 |