Skip to content

[NVIDIA] Change to use num_local_experts #8453

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ch-wan merged 2 commits into from
Jul 28, 2025
Merged

[NVIDIA] Change to use num_local_experts #8453

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b286e19
change to num_local_experts
kaixih Jul 28, 2025
722821e
Add doc
kaixih Jul 28, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion docs/backend/server_arguments.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -214,7 +214,8 @@ Please consult the documentation below and [server_args.py](https://github.com/s
| `--ep-size` | The expert parallelism size. | 1 |
| `--enable-ep-moe` | Enabling expert parallelism for moe. The ep size is equal to the tp size. | False |
| `--enable-deepep-moe` | Enabling DeepEP MoE implementation for EP MoE. | False |
| `--enable-flashinfer-moe` | Enabling Flashinfer MoE implementation. | False |
| `--enable-flashinfer-cutlass-moe` | Enabling Flashinfer Cutlass MoE implementation for high throughput. | False |
| `--enable-flashinfer-trtllm-moe` | Enabling Flashinfer Trtllm MoE implementation for low latency. | False |
| `--deepep-mode` | Select the mode when enable DeepEP MoE, could be `normal`, `low_latency` or `auto`. Default is `auto`, which means `low_latency` for decode batch and `normal` for prefill batch. | auto |
| `--ep-num-redundant-experts` | Allocate this number of redundant experts in expert parallel. | 0 |
| `--ep-dispatch-algorithm` | The algorithm to choose ranks for redundant experts in expert parallel. | None |
Expand Down
2 changes: 1 addition & 1 deletion python/sglang/srt/layers/moe/ep_moe/layer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1268,7 +1268,7 @@ def forward(self, hidden_states: torch.Tensor, router_logits: torch.Tensor):
topk_group=self.topk_group,
intermediate_size=self.w2_weight.shape[2],
local_expert_offset=self.start_expert_id,
local_num_experts=self.num_experts_per_partition,
local_num_experts=self.num_local_experts,
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Jul 28, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The code was referencing self.num_experts_per_partition which does not exist. It should be self.num_local_experts to correctly pass the number of local experts to the kernel.

local_num_experts=self.num_local_experts

routed_scaling_factor=self.routed_scaling_factor,
tile_tokens_dim=_get_tile_tokens_dim(
hidden_states.shape[0], self.top_k, self.num_experts
Expand Down
Loading