[Quantization][FP8] Adding support for fp8 gemm layer input in fp8 (#14578)
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com> Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
This commit is contained in:
co-authored by
Luka Govedič
parent
e7f720ea56
commit
4d0ec37267
@@ -22,6 +22,7 @@ class QuarkW8A8Fp8(QuarkScheme):
|
||||
self.qscheme = qscheme
|
||||
self.is_static_input_scheme = is_static_input_scheme
|
||||
self.fp8_linear = Fp8LinearOp(use_per_token_if_dynamic=True)
|
||||
self.out_dtype = torch.get_default_dtype()
|
||||
|
||||
@classmethod
|
||||
def get_min_capability(cls) -> int:
|
||||
@@ -134,5 +135,6 @@ class QuarkW8A8Fp8(QuarkScheme):
|
||||
return self.fp8_linear.apply(input=x,
|
||||
weight=layer.weight,
|
||||
weight_scale=layer.weight_scale,
|
||||
out_dtype=self.out_dtype,
|
||||
input_scale=layer.input_scale,
|
||||
bias=bias)
|
||||
|
||||
Reference in New Issue
Block a user