LLM as a Judge Sequential Chaining

[TheWelcomer]

Describe the feature or potential improvement

I wanted to propose supporting the ability to feed selected output of an LLM as a judge evaluation into other LLM as a judge evaluations within the same experiment or live run. I would find this feature useful for my use case, and in the general case where output from one judge would be useful in another judge's evaluation

Additional information

No response

[marliessophie]

Hey, thanks for sharing. Directionally agree this would be helpful! Cannot commit any support for the next few weeks though. Will keep this open for the minute to collect feedback and upvotes. Hope this helps.