This commit is contained in:
Yuanle Liu
2025-07-16 15:33:10 +08:00
committed by GitHub
parent a83a3eea5f
commit dda4a9f848
10 changed files with 26 additions and 131 deletions

View File

@@ -14,6 +14,7 @@
# limitations under the License.
"""
from fastdeploy.worker.worker_process import initialize_fd_config
@@ -24,7 +25,7 @@ class RolloutModelConfig:
max_model_len: int = 32768,
tensor_parallel_size: int = 4,
dynamic_load_weight: bool = True,
load_strategy: str = "meta",
load_strategy: str = "ipc_snapshot",
enable_mm: bool = False,
# Default values for all other parameters
max_num_seqs: int = 34,