Remove QServe and FBGEMM FP8 quantization (#31109)

This commit is contained in:
Brayden Zhong
2026-07-17 17:10:34 -07:00
committed by GitHub
parent f926c30c57
commit 238b2b2c9c
18 changed files with 1 additions and 2566 deletions
-4
View File
@@ -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",