Description
The ggml-ci is an extremely valuable resource that allows to run heavy workloads across a wider-range of hardware. It works by utilizing self-hosted nodes in the cloud and on-premise. For example, here is the current list of workflows running for llama.cpp:

The ggml-ci
workflows are defined via a very simple Bash script. For more information, check the ggml-ci repository. Here are a couple of examples scripts that have been used successfully for a very long time:
- https://github.com/ggerganov/llama.cpp/tree/master/ci
- https://github.com/ggerganov/ggml/tree/master/ci
The goal of this issue is to add a similar ggml-ci
workflow for whisper.cpp
that would run on all commits on the master
branch and for upcoming pull requests from collaborators. This would allow us to verify integrity of the results across different hardware and would improve the development process significantly.
Metadata
Metadata
Assignees
Type
Projects
Status