Commit Graph

6 Commits (c972c4e09018be0646ce1abd110239cd03fc0b24)

Author SHA1 Message Date
xiao-baia c972c4e090
Fix DPO loss_mask boundary (include first assistant token)
Fix off-by-one in generate_loss_mask so the first assistant token contributes to loss in DPO.
2026-01-07 21:00:46 +08:00
jingyaogong 7641985d14 [update] simplify loader 2026-01-06 01:20:52 +08:00
jingyaogong 4a5c9f5ece [feat] stream load data 2025-12-28 16:58:52 +08:00
jingyaogong fe24501602 [feat] adjust seq length 2025-12-14 20:41:58 +08:00
jingyaogong 7d877db79b [feat] minimind-2510 2025-10-21 21:19:47 +08:00
jingyaogong 274483cb1b 250426 2025-04-26 10:07:55 +08:00