Skip to content

[Reland] Clean up deprecated tile_tokens_dim for next flashinfer #13535

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

Closed
Fridge003 wants to merge 5 commits into from
Closed

[Reland] Clean up deprecated tile_tokens_dim for next flashinfer #13535

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
fab3e61
fix
Fridge003 Nov 18, 2025
b18940a
upd
Fridge003 Nov 18, 2025
96798d6
Merge branch 'main' into baizhou/patch-2
Fridge003 Nov 24, 2025
eb60cc5
Merge branch 'main' into baizhou/patch-2
Fridge003 Dec 5, 2025
68ffa50
Merge branch 'main' into baizhou/patch-2
Fridge003 Dec 18, 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
1 change: 0 additions & 1 deletion python/sglang/srt/layers/moe/fused_moe_triton/layer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1226,7 +1226,6 @@ def forward(self, hidden_states: torch.Tensor, topk_output: TopKOutput):
local_expert_offset=self.moe_ep_rank * self.num_local_experts,
local_num_experts=self.num_local_experts,
routed_scaling_factor=self.moe_runner_config.routed_scaling_factor,
tile_tokens_dim=None,
routing_method_type=routing_method_type,
do_finalize=True,
tune_max_num_tokens=next_power_of_2(hs_fp4.shape[0]),
Expand Down
1 change: 0 additions & 1 deletion python/sglang/srt/layers/quantization/fp8.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1382,7 +1382,6 @@ def apply_with_router_logits(
if routed_scaling_factor is not None
else 1.0
),
tile_tokens_dim=None,
routing_method_type=routing_method_type,
use_shuffled_weight=False,
tune_max_num_tokens=next_power_of_2(a_q.shape[0]),
Expand Down
1 change: 0 additions & 1 deletion python/sglang/srt/layers/quantization/modelopt_quant.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -785,7 +785,6 @@ def apply(
else 1.0
),
use_routing_scales_on_input=use_routing_scales_on_input,
tile_tokens_dim=None,
routing_method_type=routing_method_type,
tune_max_num_tokens=next_power_of_2(x.shape[0]),
)
Expand Down
1 change: 0 additions & 1 deletion python/sglang/srt/layers/quantization/mxfp4.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -682,7 +682,6 @@ def apply(
layer.moe_ep_rank * layer.num_local_experts, # local_expert_offset
layer.num_local_experts, # local num experts
None,
None, # tile_tokens_dim
1, # routing_method_type, renormalize
True, # do finalize
tune_max_num_tokens=next_power_of_2(x_quant.shape[0]),
Expand Down
Loading