[NewFeture]add ep rollout model init and update/clear ep buffer (#3927)

* add ep rollout model init && add deep update/clear

* fix test
This commit is contained in:
gaoziyuan
2025-09-12 14:15:13 +08:00
committed by GitHub
parent c64ceac34d
commit 10768a4d79
13 changed files with 364 additions and 304 deletions

View File

@@ -24,13 +24,13 @@ 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,