Mathematical Reasoning on GSM-Hard (Pass@1, #L)
15.83Pass@1Latent-GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Latent-GRPOBase Model=LLaMA-3.2-1B-Instruct, Reasoning Mode=Latent Reasoning (No Sampling)2026.04 | 15.83 | 28 | |
| GRPOBase Model=LLaMA-3.2-1B-Instruct, Reasoning Mode=Explicit Reasoning2026.04 | 14.7 | 130.8 | |
| SFTBase Model=LLaMA-3.2-1B-Instruct, Reasoning Mode=Explicit Reasoning2026.04 | 11.67 | 85.78 | |
| Soft-GRPOBase Model=LLaMA-3.2-1B-Instruct, Reasoning Mode=Latent Reasoning (No Sampling)2026.04 | 10.3 | 25.6 | |
| Latent-SFTBase Model=LLaMA-3.2-1B-Instruct, Reasoning Mode=Latent Reasoning (No Sampling)2026.04 | 10.25 | 25.5 |