Mathematical Reasoning on MultiArith (Det., Cov@8, Cov@16)
93.4Detection RateLatent-SFT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Latent-SFTBackbone=Llama-3.2-1B, Sampling=MC-dropout2025.10 | 93.4 | 96.2 | 96.7 |
Loading the SOTA2 catalog…
| Method | Links | |||
|---|---|---|---|---|
| Latent-SFTBackbone=Llama-3.2-1B, Sampling=MC-dropout2025.10 | 93.4 | 96.2 | 96.7 |