Remove weight_scale.T special case for SM90 Block FP8 CUTLASS kernel (#28431)

Signed-off-by: mgoin <mgoin64@gmail.com>
2025-11-11 09:46:04 -07:00
parent 287bbbeb06
commit f9a4087182
5 changed files with 36 additions and 36 deletions
--- a/vllm/model_executor/layers/quantization/fp8.py
+++ b/vllm/model_executor/layers/quantization/fp8.py
@@ -540,7 +540,7 @@ class Fp8LinearMethod(LinearMethodBase):
            return

        if self.block_quant:
-            maybe_post_process_fp8_weight_block(layer, self.cutlass_block_fp8_supported)
+            maybe_post_process_fp8_weight_block(layer)

    def apply(
        self,