[AMD] Gate DFLASH non-greedy verify on the target-only kernel being registered (#33694)
Co-authored-by: HaiShaw <hixiao@gmail.com>
This commit is contained in:
co-authored by
HaiShaw
parent
5d1a0c7129
commit
cf7923615c
@@ -53,8 +53,7 @@ elif is_hip():
|
||||
from sglang.kernels.ops.sampling.renorm_triton import (
|
||||
top_p_renorm_probs_triton as top_p_renorm_prob,
|
||||
)
|
||||
|
||||
_DFLASH_SAMPLING_VERIFY_AVAILABLE = True
|
||||
tree_speculative_sampling_target_only = None
|
||||
else:
|
||||
top_k_renorm_prob = None
|
||||
top_p_renorm_prob = None
|
||||
|
||||
Reference in New Issue
Block a user