From d01bccde9f759b24449fdaa16306b406a50eb367 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Tue, 18 Jul 2023 14:24:43 +0300 Subject: ci : integrate with ggml-org/ci (#2250) * ci : run ctest ggml-ci * ci : add open llama 3B-v2 tests ggml-ci * ci : disable wget progress output ggml-ci * ci : add open llama 3B-v2 tg tests for q4 and q5 quantizations ggml-ci * tests : try to fix tail free sampling test ggml-ci * ci : add K-quants ggml-ci * ci : add short perplexity tests ggml-ci * ci : add README.md * ppl : add --chunks argument to limit max number of chunks ggml-ci * ci : update README --- ci/README.md | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) create mode 100644 ci/README.md (limited to 'ci/README.md') diff --git a/ci/README.md b/ci/README.md new file mode 100644 index 0000000..6c74c81 --- /dev/null +++ b/ci/README.md @@ -0,0 +1,20 @@ +# CI + +In addition to [Github Actions](https://github.com/ggerganov/llama.cpp/actions) `llama.cpp` uses a custom CI framework: + +https://github.com/ggml-org/ci + +It monitors the `master` branch for new commits and runs the +[ci/run.sh](https://github.com/ggerganov/llama.cpp/blob/master/ci/run.sh) script on dedicated cloud instances. This allows us +to execute heavier workloads compared to just using Github Actions. Also with time, the cloud instances will be scaled +to cover various hardware architectures, including GPU and Apple Silicon instances. + +Collaborators can optionally trigger the CI run by adding the `ggml-ci` keyword to their commit message. +Only the branches of this repo are monitored for this keyword. + +It is a good practice, before publishing changes to execute the full CI locally on your machine: + +```bash +mkdir tmp +bash ./ci/run.sh ./tmp/results ./tmp/mnt +``` -- cgit v1.2.3 From 5d500e8ccf5eee3de3ae66685cc3be75e43e08b9 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Sat, 22 Jul 2023 11:48:22 +0300 Subject: ci : add 7B CUDA tests (#2319) * ci : add 7B CUDA tests ggml-ci * ci : add Q2_K to the tests * ci : bump CUDA ppl chunks ggml-ci * ci : increase CUDA TG len + add --ignore-eos * ci : reduce CUDA ppl cunks down to 4 to save time --- ci/README.md | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'ci/README.md') diff --git a/ci/README.md b/ci/README.md index 6c74c81..65cfe63 100644 --- a/ci/README.md +++ b/ci/README.md @@ -16,5 +16,10 @@ It is a good practice, before publishing changes to execute the full CI locally ```bash mkdir tmp + +# CPU-only build bash ./ci/run.sh ./tmp/results ./tmp/mnt + +# with CUDA support +GG_BUILD_CUDA=1 bash ./ci/run.sh ./tmp/results ./tmp/mnt ``` -- cgit v1.2.3