-
Story
-
Resolution: Done
-
Undefined
-
None
-
None
-
None
-
Inference, RHOAI
-
False
-
False
-
User Story:
The AI Safety Working Group in MLCommons has a goal of advancing the state of the art for evaluating AI Safety. This could be useful for the granite models and we have access to the benchmarks so we can do early testing as a beta version of the production release, MLSafe v1.0 becomes available.
The MLCommons AI Safety Working Group wrote this paper, Introducing v0.5 of the AI Safety Benchmark from MLCommons, to introduce first version of the benchmark which addresses LLM safety.
Keep group within Red Hat informed regarding what is going in the AI Safety WG and make sure the connections are being made and collaboration happening appropriately within RH (and maybe IBM as well).
Acceptance criteria: