Skip to content

Compute DeepSeek-V4 router logits in fp32 #5849

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

Open
AusarYao28 wants to merge 1 commit into
base: main
Choose a base branch
from
+2 −0
Open

Compute DeepSeek-V4 router logits in fp32 #5849

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
2 changes: 2 additions & 0 deletions src/maxtext/configs/models/deepseek4-284b.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,8 @@ routed_bias_update_rate: 0.001
load_balance_loss_weight: 0.0001
log_moe_bias_norms: false
routed_scaling_factor: 1.5
# Reference Gate computes router logits in fp32 (linear(x.float(), weight.float())) regardless of model dtype.
float32_gate_logits: true

@github-actions github-actions Bot Oct 19, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

🟡 The new config setting is unvalidated by automated regression tests in CI. While the PR description cites tests/unit/configs_test.py and tests/unit/deepseek_v4_routed_bias_test.py, deepseek4-284b.yml is omitted from DEEPSEEK_CONFIGS in configs_test.py, and deepseek_v4_routed_bias_test.py sets float32_gate_logits explicitly via CLI args rather than validating model config defaults.



# --- Attention configuration ---
Expand Down
Loading