Creating a New Project with a bad key results in null values

[cfrenchi]

Describe the bug
After creating a New Project with a bad key for OpenAI GPT-3.5-Turbo, the results are returned as Null. I was expecting to get a bad key response or error. Perhaps rather than"null", the error code or reason is what is captured from the request? Either an invalid_api_key or if/when rate limits are hit a 429?

To Reproduce
Steps to reproduce the behavior:

1. Create a New Project and fill in all the fields and add a jsonl file for your dataset
2. Select an OpenAI LLM for Eval
3. Enter in the key sk (Should there be a key-length check here and warning?)
4. Click Next
5. Select any/all metrics
6. Click Finish
7. The Evaluations screen shows "null" for your metric score

Should scores be null based on a key failure? Should there be an error reason shown? Null is vague. Is my key wrong or is my system?

Expected behavior
If a bad/expired key is used, the evaluation to reflect that there is a key problem and not present "null"

Screenshots
If applicable, add screenshots to help explain your problem.

Seeing null for a bad key is confusing. Where is the error actually happening?

Additional context
Entering in a bad key or having an expired key might lead to confusion when "null" starts to present itself. Should an error code or reason be presented instead?

[Dominastorm]

Thanks for raising this. This is currently in our pipeline, and should be fixed in the next dashboard release.