Change8

b9156

📦 llama-cppView on GitHub →
1 features🐛 4 fixes🔧 1 symbols

Summary

This release enables NVIDIA self-hosted CI for ggml-webgpu, addresses several precision and placement issues within WebGPU builds, and provides extensive pre-compiled binaries across multiple operating systems and hardware configurations.

✨ New Features

  • Enabled NVIDIA self-hosted CI for ggml-webgpu builds.

🐛 Bug Fixes

  • Addressed precision issues related to ggml-webgpu.
  • Fixed placement issues in ggml-webgpu.
  • Relaxed constraints on more set_rows and div operations.
  • Attempted to relax all f16 operations.

Affected Symbols