Yoav Katz comments

Results 74 comments of


                                            Yoav Katz

Using cached metrics in consecutive test_card runs can be faulty

@antonioan showed this to me. Basically, you just need to have a variable in the metrics (even a count of processed instances variables). If you run two cards, the value...

For issue 1575: Eliminating Manual Class Registration in Unitxt, replaced by Import Paths

What is the status of this PR? I think it's an important change.

Add prediction variable name customization to LLM as Judge

I think the main point is whether it makes a difference in performance (because it complicates the API and the interface).

Eliminating Manual Class Registration in Unitxt with Import Paths

This is still worked on .Right?

Eliminating Manual Class Registration in Unitxt with Import Paths

Still is still important.

Eliminating Manual Class Registration in Unitxt with Import Paths

Still worked on in #1713

Eliminating Manual Class Registration in Unitxt with Import Paths

This is still an important issue that people struggle with.

Added calls to prepare() to ensure requirement checks are done early

This PR tries to deal with the fact that dependency checks were not performed early enough , and when running examples with required installed packages returned "module was not found"...

Added calls to prepare() to ensure requirement checks are done early

@elronbandel - the problem is that watsonx_ai requires python 3.10. What do you suggest we do?

Fix missing deep copy in MapInstanceValues

@dafnapension Thanks for making the changes and adding the test. Looks good. I can not approve it (only you or @elronbandel ).