7 Commits
Author SHA1 Message Date
SamareshSinghandGitHub 5cbaa5e69e docker : add OCI image labels for version and build date (#21653)
* docker: add OCI image labels to all published images

* docker: propagate OCI labels as manifest and index annotations

* docker: drop hardcoded org URL and revert accidental intel version bump

The OCI image url and source are now driven by build args with a sensible default. The workflow passes the actual repository url so fork builds get labels pointing at the fork instead of upstream. Also restores the IGC, compute runtime, and IGDGMM versions in the intel Dockerfile labeled stage which I accidentally bumped in the first commit.

* docker: add skip_s390x workflow_dispatch input for fast test runs

Lets maintainers and PR authors trigger the docker workflow without the s390x build target, which depends on the IBM Z runner and is by far the slowest job in the matrix. The flag filters the s390x row out of the build matrix before merge_matrix is derived, so the merge job sees a consistent shape too.

Signed-off-by: Samaresh Kumar Singh <ssam3003@gmail.com>

---------

Signed-off-by: Samaresh Kumar Singh <ssam3003@gmail.com>
2026-05-18 22:14:45 +02:00
81df3f7cfa fix: GLM-DSA crash in llama-tokenize when using vocab_only (#22102)
* llama: fix crash in print_info for GLM-DSA when vocab_only is set

* addressed code review comments

* cont : simplify

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2026-04-20 10:32:46 +03:00
SamareshSinghandGitHub 59accc8863 ggml-backend-meta: add multi-segment read support in get_tensor (#22063) 2026-04-18 10:04:51 +02:00
SamareshSinghandGitHub 650bf14eb9 llama-model: read final_logit_softcapping for Gemma 4 (#21390) 2026-04-04 13:05:10 +02:00
SamareshSinghandGitHub 0fac87b157 imatrix : fix crash when using --show-statistics with zero counts (#19532)
* imatrix: fix crash when using --show-statistics with zero counts

Fixes division by zero that caused floating point exceptions when processing imatrix files with zero count values. Added checks to skip zero counts and handle empty activation vectors.

Fix for the bug #19190

* imatrix: lower log level for zero-count skip message to DBG
2026-03-26 08:14:36 +01:00
cb8f4fa3f8 Fix locale-dependent float printing in GGUF metadata (#17331)
* Set C locale for consistent float formatting across all binaries.

* Add C locale setting to all tools binaries

Add std::setlocale(LC_NUMERIC, "C") to all 16 binaries in the tools/
directory to ensure consistent floating-point formatting.

* Apply suggestion from @JohannesGaessler

---------

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2026-03-04 09:30:40 +01:00
SamareshSinghandGitHub 3a00c98584 cmake : fix KleidiAI install target failure with EXCLUDE_FROM_ALL (#19581)
* cmake: fix KleidiAI install target failure with EXCLUDE_FROM_ALL

Fix for the bug #19501 by adding EXCLUDE_FROM_ALL to FetchContent_Declare. This properly excludes KleidiAI from both build and install targets, preventing install failures when GGML_CPU_KLEIDIAI=ON is used.

The KleidiAI source files are still compiled into libggml-cpu.so, preserving all functionality.

* addressed code review comments
2026-02-15 06:22:53 +01:00