Skip to content

Enable MMLU tests to be run on an already served model #50

Description

@alimaredia

Right now the inferencing done during MMLU is done within the lm-eval-harness library. Lm-eval-harness has the ability to inference with openai-api compatible servers similar to how models are being served via ilab serve:

https://github.com/EleutherAI/lm-evaluation-harness?tab=readme-ov-file#model-apis-and-inference-servers

A fix for this issue would involve serving a model on an endpoint with either vLLM or llama-cpp and being able to pass that endpoint into MMLUEvaluator.run() or the MMLUEvaluator class to run MMLU tests.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or requestmmluPertains to MMLUmmlubranchPertains to MMLUBranch

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions