[Quantization] Rework quantization_config to use QuantKey and allow for activation override (#41566)

This commit is contained in:
Michael Goin
2026-05-13 16:58:32 -04:00
committed by GitHub
parent cca32d55a2
commit 8efd508204
16 changed files with 621 additions and 322 deletions
+5
View File
@@ -214,4 +214,9 @@ gpt_oss_20b = ModelFusionInfo(
sequence_parallel=n_layers * 2 + 1,
async_tp=n_layers * 2,
),
model_kwargs=(
{"quantization_config": {"moe": {"activation": "mxfp8"}}}
if is_blackwell()
else {}
),
)