[Kernel] Support deep_gemm for linear methods (#19085)
Signed-off-by: artetaout <lulala341@gmail.com>
This commit is contained in:
@@ -402,6 +402,7 @@ class Fp8LinearMethod(LinearMethodBase):
|
||||
|
||||
if self.block_quant:
|
||||
assert self.quant_config.weight_block_size is not None
|
||||
|
||||
return torch.ops.vllm.apply_w8a8_block_fp8_linear(
|
||||
input=x,
|
||||
weight=layer.weight,
|
||||
|
||||
Reference in New Issue
Block a user