Commit History
2 days ago
-
output classifier trainer (value head, omega ramp)
Verified signature808f43d -
calibration + red-team eval with rubric judge
Verified signaturee8fd040 -
dataset builder: seeds + vLLM augmentation
Verified signaturece87bb4 -
input classifier trainer (TRL SFT, NTP verdict)
Verified signature966092b