apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-09-26 20:41:53 +08:00
Code Issues Actions 6 Packages Projects Releases Wiki Activity

fix mtp in rl (#4234)
Some checks failed
CE Compile Job / ce_job_pre_check (push) Has been cancelled
Details
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
Details
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
Details
CE Compile Job / Show Code Archive Output (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
Details
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Details

Browse Source
This commit is contained in:
freeliuzc
2025-09-24 16:59:24 +08:00
committed by GitHub
parent 8a506500f3
commit e2b68b33c9
2 changed files with 7 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
fastdeploy/engine/args_utils.py
View File

@@ -1028,7 +1028,10 @@ class EngineArgs:
if paddle.is_compiled_with_xpu():
self.max_num_batched_tokens = self.max_model_len
else:
self.max_num_batched_tokens = 8192 # if set to max_model_len, it's easy to be OOM
if speculative_cfg is not None and speculative_cfg.method is not None:
self.max_num_batched_tokens = self.max_model_len
else:
self.max_num_batched_tokens = 8192 # if set to max_model_len, it's easy to be OOM
else:
if self.enable_chunked_prefill:
self.max_num_batched_tokens = 2048

3
fastdeploy/output/token_processor.py
View File

@@ -332,6 +332,9 @@ class TokenProcessor:
+ accept_num[i]
].tolist()
if len(token_ids) == 0 or token_ids[-1] <= 0:
if envs.ENABLE_V1_KVCACHE_SCHEDULER:
if task_id in self.resource_manager.to_be_rescheduled_request_id_set:
self.resource_manager.reschedule_preempt_task(task_id)
continue
else:
token_id = int(tokens[i, 0])