README + website updates#10
Conversation
| @@ -90,22 +79,22 @@ | |||
| - `GOOGLE_APPLICATION_CREDENTIALS`: Gemini via Vertex AI (audio judge metrics) | |||
| - `AWS_ACCESS_KEY_ID` + `AWS_SECRET_ACCESS_KEY`: Claude via Bedrock (faithfulness metric) | |||
There was a problem hiding this comment.
No i think this complicates things unnecessarily. also i think if people want to they can easily just do this on their own. I think lots of people (including us) just auto replace the judges whenever they clone new benchmarks to match what they have
There was a problem hiding this comment.
Ok! Do we still want to mention that JUDGE_MODEL exists and can be overwritten?
There was a problem hiding this comment.
i dont think it does exist anymore though
| @@ -69,19 +67,10 @@ | |||
| ``` | |||
|
|
|||
There was a problem hiding this comment.
Should we keep the header? Good job on the rest though, it's true it was duplicated
Co-authored-by: Joseph Marinier <Joseph.Marinier@gmail.com>
…a into pr/tara/website_and_readme
| │ │ └── validation/ # Quality control metrics | ||
| │ └── utils/ # Utilities (LLM client, log processing) | ||
| ├── scripts/ # Utility scripts | ||
| │ ├── run_benchmark.py # Benchmark runner |
There was a problem hiding this comment.
This was moved since then.
gabegma
left a comment
gabegma
left a comment
There was a problem hiding this comment.
Tiny comment and then this is good to go!
3 participants
No description provided.