Remove QServe and FBGEMM FP8 quantization (#31109)
This commit is contained in:
@@ -61,8 +61,6 @@ else:
|
||||
gptq_gemm,
|
||||
gptq_shuffle,
|
||||
int8_scaled_mm,
|
||||
qserve_w4a8_per_chn_gemm,
|
||||
qserve_w4a8_per_group_gemm,
|
||||
sgl_per_token_group_quant_8bit,
|
||||
sgl_per_token_group_quant_fp8,
|
||||
sgl_per_token_group_quant_int8,
|
||||
@@ -192,8 +190,6 @@ else:
|
||||
"moe_sum",
|
||||
"moe_sum_reduce",
|
||||
"prepare_moe_input",
|
||||
"qserve_w4a8_per_chn_gemm",
|
||||
"qserve_w4a8_per_group_gemm",
|
||||
"reconstruct_indices_from_tree_mask",
|
||||
"rmsnorm",
|
||||
"rotary_embedding",
|
||||
|
||||
Reference in New Issue
Block a user