apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Feature] support seed parameter (#3161)

Browse Source 
* support seed

* fix

* add SamplingMetadata seed test

* The next_tokens values are inconsistent!

* add air and rejection seed test

* fix

* add SamplingParams seed test

* fix seed=0

* Default to defualt

* fix

* fix args_utils

* fix review

* fix review

* fix

* fix

* add xpu,gcu,iluvatar support seed

* fix
This commit is contained in:
lizexu123
2025-08-06 15:20:47 +08:00
committed by GitHub
parent 20839abccf
commit afff4d37ea
15 changed files with 386 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/worker/gpu_worker.py
View File

@@ -27,7 +27,7 @@ from fastdeploy.config import FDConfig
from fastdeploy.engine.request import Request
from fastdeploy.platforms import current_platform
from fastdeploy.plugins.model_runner import load_model_runner_plugins
from fastdeploy.utils import get_logger
from fastdeploy.utils import get_logger, set_random_seed
from fastdeploy.worker.model_runner_base import ModelRunnerBase
from fastdeploy.worker.output import ModelRunnerOutput
from fastdeploy.worker.worker_base import WorkerBase
@@ -75,6 +75,7 @@ class GpuWorker(WorkerBase):
else:
raise RuntimeError(f"Not support device type: {self.device_config.device}")
set_random_seed(self.fd_config.model_config.seed)
# Construct model runner
self.model_runner: ModelRunnerBase = ModelRunner(
fd_config=self.fd_config,
@@ -129,6 +130,7 @@ class GpuWorker(WorkerBase):
# 2. Profile run
self.model_runner.profile_run()
set_random_seed(self.fd_config.model_config.seed)
# 3. Statistical memory information
paddle_reserved_mem_after_run = paddle.device.cuda.max_memory_reserved(local_rank)