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

[NewFeture]add ep rollout model init and update/clear ep buffer (#4039)

Browse Source 
* fix gid

* merge

* fix test

* fix bug

* fix

* fix ci
This commit is contained in:
gaoziyuan
2025-09-17 20:24:53 +08:00
committed by GitHub
parent 0d3a57a2c6
commit 896e3bb606
12 changed files with 348 additions and 293 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/rl/rollout_config.py
View File

@@ -26,13 +26,13 @@ class RolloutModelConfig:
max_model_len: int = 32768,
tensor_parallel_size: int = 4,
dynamic_load_weight: bool = True,
load_strategy: str = "ipc_snapshot",
load_strategy: str = "meta",
enable_mm: bool = False,
# Default values for all other parameters
max_num_seqs: int = 34,
total_block_num: int = 2000,
block_size: int = 64,
engine_worker_queue_port: int = 9923,
engine_worker_queue_port: str = "8002",
device_ids: str = "0",
dtype: str = "bfloat16",
enc_dec_block_num: int = 1,