Refactor fp8 nextn layer for DeepSeek nvfp4 checkpoint (#15353)
This commit is contained in:
@@ -253,6 +253,7 @@ class Envs:
|
||||
SGLANG_MOE_NVFP4_DISPATCH = EnvBool(False)
|
||||
SGLANG_NVFP4_CKPT_FP8_GEMM_IN_ATTN = EnvBool(False)
|
||||
SGLANG_PER_TOKEN_GROUP_QUANT_8BIT_V2 = EnvBool(False)
|
||||
SGLANG_NVFP4_CKPT_FP8_NEXTN_MOE = EnvBool(False)
|
||||
|
||||
# Flashinfer
|
||||
SGLANG_IS_FLASHINFER_AVAILABLE = EnvBool(True)
|
||||
|
||||
Reference in New Issue
Block a user