Support not officially supported high sgl-kernel version with low srt version (#11786)

2025-10-19 16:11:59 +08:00
parent ce399e154c
commit a27825ae01
2 changed files with 7 additions and 0 deletions
--- a/sgl-kernel/python/sgl_kernel/init.py
+++ b/sgl-kernel/python/sgl_kernel/init.py
@@ -265,6 +265,8 @@ from sgl_kernel.gemm import (
    scaled_fp4_quant,
    sgl_per_tensor_quant_fp8,
    sgl_per_token_group_quant_8bit,
+    sgl_per_token_group_quant_fp8,
+    sgl_per_token_group_quant_int8,
    sgl_per_token_quant_fp8,
    shuffle_rows,
    silu_and_mul_scaled_fp4_grouped_quant,
--- a/sgl-kernel/python/sgl_kernel/gemm.py
+++ b/sgl-kernel/python/sgl_kernel/gemm.py
@@ -137,6 +137,11 @@ def sgl_per_token_group_quant_8bit(
    )


+# For legacy usage
+sgl_per_token_group_quant_fp8 = sgl_per_token_group_quant_8bit
+sgl_per_token_group_quant_int8 = sgl_per_token_group_quant_8bit
+
+
 def sgl_per_tensor_quant_fp8(
    input: torch.Tensor,
    output_q: torch.Tensor,