apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-09-28 21:32:29 +08:00
Code Issues Actions 6 Packages Projects Releases Wiki Activity

[Executor] Adjust signal sending order in RL training (#3773) (#4066) (#4178)
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 
* Adjust processing order

* fix bug

* fix update_parameters bug

* refine code
This commit is contained in:
RAM
2025-09-22 14:31:36 +08:00
committed by GitHub
parent 7bdc6f41e5
commit 01f6934162
3 changed files with 20 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
fastdeploy/rl/dynamic_weight_manager.py
View File

@@ -45,6 +45,7 @@ class DynamicWeightManager:
self.model: nn.Layer = model
self._capture_model_state()
self.update_parameters()
self.finalize_update()
logger.info(
f"✅ DynamicLoad model built successfully by {self.load_config.load_strategy}, "
@@ -81,8 +82,6 @@ class DynamicWeightManager:
logger.info(f"Update parameters in {time.perf_counter()-start_time:.2f}s")
self._finalize_update(pid)
def _update_ipc_snapshot(self):
"""Update using IPC snapshot strategy for elastic recovery."""
model_path = os.path.join(
@@ -146,7 +145,7 @@ class DynamicWeightManager:
if src.shape != dst.shape:
raise ValueError(f"Shape mismatch for {name}: {src.shape} vs {dst.shape}")
def _finalize_update(self, pid: int):
def finalize_update(self, pid: int = 0):
"""Finalize update process with verification."""
self._verify_parameters("update")
if self.parallel_config.tensor_parallel_size > 1: