apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[BugFix] fix TPDP mix parallel infer (#4583)

Browse Source 
Co-authored-by: Yuanle Liu <yuanlehome@163.com>
This commit is contained in:
lizhenyun01
2025-10-28 16:58:20 +08:00
committed by GitHub
parent c63361fd1d
commit 4d2f478d53
3 changed files with 13 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/config.py
View File

@@ -1588,10 +1588,6 @@ class FDConfig:
f"be less than or equal to max_num_partial_prefills: {self.max_num_partial_prefills}"
)
assert self.scheduler_config.splitwise_role in ["mixed", "prefill", "decode"]
# TODO(@wufeisheng): TP and EP need to be supported simultaneously.
assert (self.parallel_config.tensor_parallel_size == 1 and self.parallel_config.expert_parallel_size >= 1) or (
self.parallel_config.tensor_parallel_size >= 1 and self.parallel_config.expert_parallel_size == 1
), "TP and EP cannot be enabled at the same time"
if not self.cache_config.enable_chunked_prefill:
if not envs.ENABLE_V1_KVCACHE_SCHEDULER: