feat: add SFT entropy logging and validation loss monitoring#1925

Open

none0663 wants to merge 3 commits into

none0663:feature/sft-entropy-and-val-loss

Commits on May 19, 2026

feat: add SFT entropy logging and validation loss monitoring

none0663
and
claude
committed
fix: correct CP>1 val/loss underestimation and resume baseline step

none0663
and
claude
committed
fix: handle small val dataset and baseline step collision edge cases

none0663
and
claude
committed