[ROCm][FEAT] Integrate aiter gemm w8a8 ptpc (#33773)
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
@@ -175,6 +175,8 @@ class FBGEMMFp8LinearMethod(LinearMethodBase):
|
||||
# Activations not quantized for marlin.
|
||||
del layer.input_scale_ub
|
||||
|
||||
self.fp8_linear.process_weights_after_loading(layer)
|
||||
|
||||
def apply(
|
||||
self,
|
||||
layer: torch.nn.Module,
|
||||
|
||||
Reference in New Issue
Block a user