b10083

ggml-org/llama.cppb10083Jul 22, 2026by github-actions[bot]same day as b10082

What changed since b10082

low risk

b10083 adds a sqrt_softplus activation to the CUDA top-k MoE path for dsv4 models, with no other changes noted.

New features

  • CUDA top-k MoE now supports a sqrt_softplus activation function, targeted at dsv4 models.

Upgrade notes

This is a single-commit increment over b10082, so no config or migration steps are needed. Only users running dsv4-style MoE models on CUDA benefit; other backends and models are unaffected. The KleidiAI missing-kernel warning from b10082 remains in place.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Adds a sqrt_softplus function to the top-k Mixture of Experts (MoE) implementation for dsv4, enhancing activation function capabilities.

Key Highlights

  • Adds sqrt_softplus to topk-moe
  • Targeted improvement for dsv4 model

Release Notes (excerpt)

<details open> cuda: add sqrt_softplus in topk-moe for dsv4 (#25896) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (ROCm 7.2) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16) **Android:**…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories