apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-04 16:22:57 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[fix] qwen output inconsistency when top_p=0 (#3634)

Browse Source 
* [fix] qwen output inconsistency when top_p=0

* [fix] remove decode pre_id code
This commit is contained in:
李泳桦
2025-08-27 17:16:23 +08:00
committed by GitHub
parent 1265f6c192
commit b2afdf4fc6
3 changed files with 3 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
fastdeploy/worker/gpu_model_runner.py
View File

@@ -288,6 +288,7 @@ class GPUModelRunner(ModelRunnerBase):
self.share_inputs["step_idx"][idx : idx + 1] = (
len(request.output_token_ids) if prefill_end_index >= len(input_ids) else 0
)
self.share_inputs["pre_ids"][idx : idx + 1] = -1
has_prefill_task = True
elif request.task_type.value == RequestType.DECODE.value: # decode task
logger.debug(f"Handle decode request {request} at idx {idx}")

1
fastdeploy/worker/metax_model_runner.py
View File

@@ -276,6 +276,7 @@ class MetaxModelRunner(ModelRunnerBase):
self.share_inputs["step_idx"][idx : idx + 1] = (
len(request.output_token_ids) if prefill_end_index >= len(input_ids) else 0
)
self.share_inputs["pre_ids"][idx : idx + 1] = -1
has_prefill_task = True
elif request.task_type.value == RequestType.DECODE.value: # decode task
logger.debug(f"Handle decode request {request} at idx {idx}")

1
fastdeploy/worker/xpu_model_runner.py
View File

@@ -411,6 +411,7 @@ class XPUModelRunner(ModelRunnerBase):
self.share_inputs["step_idx"][idx : idx + 1] = (
len(request.output_token_ids) if prefill_end_index >= len(input_ids) else 0
)
self.share_inputs["pre_ids"][idx : idx + 1] = -1
has_prefill_task = True
elif request.task_type.value == RequestType.DECODE.value: # decode task
logger.debug(f"Handle decode request {request} at idx {idx}")