[Bugfix] Fix quant RMS norm fusion for quantization with TMA-aligned scales (#33255)
Signed-off-by: ElizaWszola <ewszola@redhat.com> Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
This commit is contained in:
co-authored by
Luka Govedič
parent
a49ea5a58f
commit
a88b3be7c4
+3
-1
@@ -315,7 +315,9 @@ void silu_and_mul_scaled_fp4_experts_quant(
|
||||
void per_token_group_quant_fp8(const torch::Tensor& input,
|
||||
torch::Tensor& output_q, torch::Tensor& output_s,
|
||||
int64_t group_size, double eps, double fp8_min,
|
||||
double fp8_max, bool scale_ue8m0);
|
||||
double fp8_max, bool scale_ue8m0,
|
||||
bool dummy_is_scale_transposed,
|
||||
bool dummy_is_tma_aligned);
|
||||
|
||||
void per_token_group_quant_int8(const torch::Tensor& input,
|
||||
torch::Tensor& output_q,
|
||||
|
||||
Reference in New Issue
Block a user