b9832

ggml-org/llama.cppb9832Jun 28, 2026by github-actions[bot]same day as b9831

What changed since b9831

low risk

Adds a `--dump-prog` debugging flag to the Jinja template runtime; no other changes between b9831 and b9832.

New features

  • New `--dump-prog` flag in the Jinja runtime that prints the generated program for template troubleshooting.

Improvements

  • Minor cleanup in common/jinja/runtime.cpp alongside the new debug flag.

Upgrade notes

Drop-in replacement with no config or model format changes; the new flag is opt-in and only useful when diagnosing Jinja chat template issues. Existing DFlash v2 and sliding-window behavior from b9831 is unchanged.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Enhances the Jinja template engine with a debugging flag to output the generated program for troubleshooting.

Key Highlights

  • Added `--dump-prog` flag to the Jinja runtime
  • Allows users to debug Jinja templates by dumping the generated program
  • Updated common/jinja/runtime.cpp

Release Notes (excerpt)

<details open> jinja: add --dump-prog for debugging (#25086) * jinja: add --dump-prog for debugging * Update common/jinja/runtime.cpp Co-authored-by: Sigbjørn Skjæret <1629204+CISC@users.noreply.github.com> --------- Co-authored-by: Sigbjørn Skjæret <1629204+CISC@users.noreply.github.com> </details> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories