How to eval the accuracy/quality of a LLM?
Options
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-28-2023 04:35 PM
extremely subjective (human responses on likert scale don't suffice and hard to quantify with one metric on accuracy). curious how others have approached this
Options
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-28-2023 05:32 PM
There aren't exact metrics to evaluate how LLM is accurate right now. Just make test questions have a large coverage could be a tip. I expect many people will work on it!
