]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
llama: end-to-end tests (#19802)
authorJohannes Gäßler <redacted>
Sun, 8 Mar 2026 11:30:21 +0000 (12:30 +0100)
committerGitHub <redacted>
Sun, 8 Mar 2026 11:30:21 +0000 (12:30 +0100)
commita976ff081b4657b67f48295bbefc030d9d899b17
treec98c6f7ee614e1965aa01b0aec291d7d9b8bdba4
parenta95047979a1671be970398a7c8073159ac71013e
llama: end-to-end tests (#19802)

* tests: add end-to-end tests per model architecture

* fixup for rebase

* fix use-after-free in llama-model-loader.cpp

* fix CI

* fix WebGPU

* fix CI

* disable CI for macOS-latest-cmake-arm64

* use expert_weights_scale only if != 0.0f

* comments
33 files changed:
.github/workflows/build.yml
common/arg.cpp
common/common.h
include/llama.h
scripts/git-bisect-run.sh [new file with mode: 0755]
scripts/git-bisect.sh [new file with mode: 0755]
src/llama-arch.cpp
src/llama-arch.h
src/llama-context.cpp
src/llama-graph.cpp
src/llama-model-loader.cpp
src/llama-model-loader.h
src/llama-model-saver.cpp
src/llama-model-saver.h
src/llama-model.cpp
src/llama-quant.cpp
src/llama-vocab.cpp
src/llama.cpp
src/models/baichuan.cpp
src/models/bailingmoe2.cpp
src/models/dots1.cpp
src/models/exaone-moe.cpp
src/models/glm4-moe.cpp
src/models/kimi-linear.cpp
src/models/mamba-base.cpp
src/models/nemotron-h.cpp
src/models/plamo2.cpp
tests/CMakeLists.txt
tests/test-llama-archs.cpp [new file with mode: 0644]
tools/CMakeLists.txt
tools/results/CMakeLists.txt [new file with mode: 0644]
tools/results/README.md [new file with mode: 0644]
tools/results/results.cpp [new file with mode: 0644]