Skip to content

Fix MXFP4 scale placeholder initialization #33500

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
Fridge003 merged 1 commit into from
Aug 6, 2026
Merged

Fix MXFP4 scale placeholder initialization #33500

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
26 changes: 18 additions & 8 deletions python/sglang/srt/layers/quantization/mxfp4.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,6 +70,10 @@

has_triton_kernels = is_triton_kernels_available()

# Serialized MXFP4 scales use raw UE8M0 bytes. Keep fresh parameters valid for
# post-load transforms and dummy initialization; 127 is the neutral scale (1.0).
_UE8M0_ONE = 127


if is_flashinfer_available():
from flashinfer import (
Expand Down Expand Up @@ -473,10 +477,13 @@ def create_weights(
set_weight_attrs(w13_weight, extra_weight_attrs)

w13_weight_scale = torch.nn.Parameter(
torch.zeros(
layer.num_local_experts,
2 * intermediate_size_per_partition_after_pad,
hidden_size // mxfp4_block,
torch.full(
(
layer.num_local_experts,
2 * intermediate_size_per_partition_after_pad,
hidden_size // mxfp4_block,
),
fill_value=_UE8M0_ONE,
dtype=scale_dtype,
),
requires_grad=False,
Expand Down Expand Up @@ -512,10 +519,13 @@ def create_weights(
set_weight_attrs(w2_weight, extra_weight_attrs)

w2_weight_scale = torch.nn.Parameter(
torch.zeros(
layer.num_local_experts,
hidden_size,
intermediate_size_per_partition_after_pad // mxfp4_block,
torch.full(
(
layer.num_local_experts,
hidden_size,
intermediate_size_per_partition_after_pad // mxfp4_block,
),
fill_value=_UE8M0_ONE,
dtype=scale_dtype,
),
requires_grad=False,
Expand Down
Loading