Train Qwen2.5-1.5B on Knights & Knaves logic puzzles via SFT → DPO → GRPO.
See design spec and implementation plan.
Status: in development. Final results table will replace this stub.
| Name | Name | Last commit date | ||
|---|---|---|---|---|
Train Qwen2.5-1.5B on Knights & Knaves logic puzzles via SFT → DPO → GRPO.
See design spec and implementation plan.
Status: in development. Final results table will replace this stub.