EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[Feature] Integrate DeepEP into SGLang (#4232)

Browse Source 
Co-authored-by: Cheng Wan <cwan39@gatech.edu>
Co-authored-by: Xuting Zhou <xutingz@nvidia.com>
This commit is contained in:
Jinyan Chen
2025-03-19 23:16:31 +08:00
committed by GitHub
parent f9c53cbb42
commit f44db16c8e
12 changed files with 1228 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
docs/backend/server_arguments.md
View File

@@ -89,6 +89,7 @@ Please consult the documentation below to learn more about the parameters you ma
### Expert parallelism
* `enable_ep_moe`: Enables expert parallelism that distributes the experts onto multiple GPUs for MoE models.
* `ep_size`: The size of EP. Please shard the model weights with `tp_size=ep_size`, for detailed benchmarking refer to [this PR](https://github.com/sgl-project/sglang/pull/2203). If not set, `ep_size` will be automatically set to `tp_size`.
* `enable_deepep_moe`: Enables expert parallelism that distributes the experts onto multiple GPUs for DeepSeek-V3 model based on deepseek-ai/DeepEP. Currently DeepEP is bind to DP Attention. Please set `--enable-dp-attention --enable-deepep-moe`, perfer `tp_size=dp_size=ep_size`.
## Memory and scheduling