apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-11-01 04:12:58 +08:00
Code Issues Actions 6 Packages Projects Releases Wiki Activity

Release/2.1 (#3414)

Browse Source 
* Pre ce modified (#3335) (#3360)

* Pre ce modified (#3335)

* update

* update

* fix

* fix

* update

* update

* update

* fix

* update

* update

* update

* add ut fix pr(3367)

* [Bug Fix] Fix V1 video bug (#3387)

* fix stopseq error info (#3342)

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>

* [BugFix] Fix default log level of paddleformers (#3377)

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>

* [Polish Code] Remove useless notes

* feat(log):add_request_and_response_log (#3392)

* Optimize CI execution workflow. (#3371) (#3384)

* fix

* [BugFix] fix control signal release failed (#3374)

* [BugFix]

* [BugFix]

* [BugFix]

* [BugFix]

* fix

* fix

---------

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>

* Revert "Merge branch 'feature/online/vs_think_20250813' into release/2.1"

This reverts commit 02596fc537, reversing
changes made to 03347626a6.

* [XPU] Fixed the issue of performance degradation caused by enabling ENABLE_V1_KVCACHE_SCHEDULER (#3393)

* fix v1 schedule oom bug

* fix v1 schedule oom bug

* [BugFix] fix ErnieProcessor not set raw_prediction (#3401)

* [Doc]Release fastdeploy-xpu 2.1.0 (#3407)

* fix v1 schedule oom bug

* fix v1 schedule oom bug

* update release note

* [Doc]Release fastdeploy-xpu 2.0.3  (#3408)

* fix v1 schedule oom bug

* fix v1 schedule oom bug

* update release note

* update info

---------

Co-authored-by: YUNSHEN XIE <1084314248@qq.com>
Co-authored-by: ming1753 <61511741+ming1753@users.noreply.github.com>
Co-authored-by: JYChen <zoooo0820@qq.com>
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>
Co-authored-by: Jiang-Jia-Jun <jiangjiajun@baidu.com>
Co-authored-by: xiaolei373 <zley373@gmail.com>
Co-authored-by: ltd0924 <32387785+ltd0924@users.noreply.github.com>
Co-authored-by: yinwei <yinwei_hust@163.com>
Co-authored-by: memoryCoderC <1137889088@qq.com>
This commit is contained in:
luukunn
2025-08-14 20:53:47 +08:00
committed by GitHub
parent e11331927f
commit 132a8ef425
30 changed files with 132 additions and 1068 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
fastdeploy/worker/xpu_model_runner.py
View File

@@ -383,15 +383,18 @@ class XPUModelRunner(ModelRunnerBase):
req_len = len(req_dicts)
has_prefill_task = False
has_decode_task = False
for i in range(req_len):
request = req_dicts[i]
idx = request.idx
if request.task_type.value == RequestType.PREFILL.value: # prefill task
logger.debug(f"Handle prefill request {request} at idx {idx}")
prefill_start_index = request.prefill_start_index
prefill_end_index = request.prefill_end_index
length = prefill_end_index - prefill_start_index
input_ids = request.prompt_token_ids + request.output_token_ids
logger.debug(
f"Handle prefill request {request} at idx {idx} prefill_start_index {prefill_start_index} prefill_end_index {prefill_end_index} need_prefilled_token_num {len(input_ids)}"
)
self.share_inputs["input_ids"][idx : idx + 1, :length] = np.array(
input_ids[prefill_start_index:prefill_end_index]
)
@@ -420,6 +423,8 @@ class XPUModelRunner(ModelRunnerBase):
self.share_inputs["block_tables"][idx : idx + 1, :encoder_block_num] = np.array(
request.block_tables, dtype="int32"
)
if self.share_inputs["is_block_step"][idx]: # has tasks to continue to decode
has_decode_task = True
continue
else: # preempted task
logger.debug(f"Handle preempted request {request} at idx {idx}")
@@ -460,7 +465,7 @@ class XPUModelRunner(ModelRunnerBase):
self.share_inputs["stop_seqs"][:stop_seqs_num, : len(request.get("stop_token_ids")[0])] = np.array(
request.get("stop_token_ids"), dtype="int64"
)
if has_prefill_task:
if has_prefill_task or has_decode_task:
self.share_inputs["not_need_stop"][0] = True
def process_prefill_inputs(self, req_dicts: List[Request]):