-
Update DFlash recommendation (#4)
- Update DFlash recommendation (f685cd148bcd8a2061e5dfd448198e8783738713)
-
Promote RC2->main: INT4/BF16 hybrid release (1M context); card cleanup (context 1M, sampling defers to generation_config, new-checkpoint note)
-
int4: spinquantless routed-only weights, 256K
-
Enable thinking by default, preserve reasoning; drop max_new_tokens cap
-
Mark </assistant> (token 24) as special in tokenizer.json
-
Mark </assistant> (token 24) as special to match internal serving
-
Laguna S 2.1 initial release

joerowell +1: varunrandery Co-authored-by: varunrandery <varunrandery@users.noreply.huggingface.co>