|
recipe.toml
|
6310b23019
ling-31-flash-fp4: fix KV OOM — weights are 86.43 GiB/rank (replicated embeddings/lm_head/dense/MLA on top of 291.7/4), GMU 0.78 left 0.29 GiB KV (65536 needs 0.77, engine init refused). Raise GMU 0.86, drop max-model-len 32768 + max-num-batched-tokens 4096 (bench scenarios short)
|
6 hari lalu |