EngineX/xc-llm-kunlun
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[Bug] Fix no apply_top_k_top_p issue. (#101)

Browse Source
This commit is contained in:
Jin Hanyu
2026-01-12 16:38:03 +08:00
committed by GitHub
parent 18fc1c006e
commit bd90350968
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm_kunlun/v1/sample/ops/topk_topp_sampler.py
View File

@@ -11,6 +11,7 @@ from vllm import envs
from vllm.logger import init_logger from vllm.logger import init_logger
from vllm.platforms import current_platform from vllm.platforms import current_platform
import xtorch_ops import xtorch_ops
import os
logger = init_logger(__name__) logger = init_logger(__name__)
@@ -28,6 +29,7 @@ class TopKTopPSampler(nn.Module):
logger.info_once( logger.info_once(
"Using FlashInfer for top-p & top-k sampling.") "Using FlashInfer for top-p & top-k sampling.")
self.forward = self.forward_kunlun self.forward = self.forward_kunlun
self.apply_top_k_top_p = apply_top_k_top_p
def forward_native( def forward_native(
self, self,