Derivation on OneReason-Bench
28.6I2I ScoreOneReason RFT thinking
Evaluation Results
| Method | Links | |
|---|---|---|
| OneReason RFT thinkingCategory=Itemic Token-Based, Thinking Mode=thinking, Stage=RFT2026.06 | 28.6 | |
| OneReason SFT non-thinkingCategory=Itemic Token-Based, Thinking Mode=non-thinking, Stage=SFT2026.06 | 28.36 | |
| LC-Rec-PT-SFT-8BCategory=Itemic Token-Based2026.06 | 25.54 | |
| OneReason RFT non-thinkingCategory=Itemic Token-Based, Thinking Mode=non-thinking, Stage=RFT2026.06 | 23.99 | |
| OneReason SFT thinkingCategory=Itemic Token-Based, Thinking Mode=thinking, Stage=SFT2026.06 | 23.88 | |
| LC-Rec-SFT-Only-14BCategory=Itemic Token-Based2026.06 | 16.21 | |
| LC-Rec-SFT-Only-8BCategory=Itemic Token-Based2026.06 | 3.43 |