Add awq dequantize kernel to sgl with 1x to 3x speedup (#4104)

2025-03-12 00:10:02 -07:00
parent e0917e6bd0
commit 07f944631e
8 changed files with 324 additions and 0 deletions
--- a/sgl-kernel/python/sgl_kernel/init.py
+++ b/sgl-kernel/python/sgl_kernel/init.py
@@ -23,6 +23,7 @@ from sgl_kernel.elementwise import (
    silu_and_mul,
 )
 from sgl_kernel.gemm import (
+    awq_dequantize,
    bmm_fp8,
    cublas_grouped_gemm,
    fp8_blockwise_scaled_mm,