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 custom all-reduce (#2758)

Browse Source 
* [Feature] support custom all-reduce

* add vllm adapted
This commit is contained in:
zhink
2025-07-09 16:00:27 +08:00
committed by GitHub
parent be21ef5047
commit b89180f1cd
16 changed files with 1194 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/worker/worker_process.py
View File

@@ -507,6 +507,9 @@ def parse_args():
parser.add_argument("--enable_prefix_caching",
action='store_true',
help="enable prefix cache")
parser.add_argument("--enable-custom-all-reduce",
action='store_true',
help="enable custom all-reduce")
parser.add_argument("--splitwise_role",
type=str,
default="mixed",
@@ -659,6 +662,7 @@ def initialize_fd_config(config_or_args) -> FDConfig:
parallel_config.enable_chunked_prefill = getattr(config_or_args, 'enable_chunked_prefill', False)
parallel_config.max_num_batched_tokens = getattr(config_or_args, 'max_num_batched_tokens', 0)
parallel_config.enable_prefix_caching = getattr(config_or_args, 'enable_prefix_caching', False)
parallel_config.enable_custom_all_reduce = getattr(config_or_args, 'enable_custom_all_reduce', False)
parallel_config.use_ep = getattr(config_or_args, 'enable_expert_parallell', False)
parallel_config.tensor_parallel_degree = getattr(config_or_args, 'tensor_parallel_size', 1)
parallel_config.expert_parallel_degree = getattr(config_or_args, 'expert_parallel_size', 1)