Open ishaan-jaff opened 3 months ago
bug in adding models, it seems to iterate through the list. i believe we've had a fix for this and it's come back. some unit test here, would help
cost tracking bug - it shows spend for models where in the models tab there's no input / output price - so i'd assume it's not being tracked.
The Feature
Tracking List of Items to improve Admin UI / Proxy
[x] Teams support isn't bug-free yet
[x] Key-owner/End-User identification needs to be cleaner
[ ] When clicking through different views, I sometimes have to refresh the UI multiple times before the data appears correctly. (Look at the Usage Panel, Try adding a new model and see if the model instantly added)
[ ] Pagination appears to be broken in some places (Users page)
[ ] Reporting / Alerting needs to be thought out further
[x] Configurable alerting types (so that users can be notified about specific categories of issues versus everything)
[ ] Delivered via Slack or Email
[x] Move Models Metrics to Usage Panel
[x] Model Metrics - sort by latency and make sure slowest models are visible
[ ] Show hanging requests on graph
[x] fix teams error where 'all-proxy-models' isn't respected
[ ] show which models a team has access to on the page
[x] Show which keys belong to Key Owner on Key Owner / End-user page
[x] Show user email if known for key owner on Key Owner / End-user page
[ ] Have Users and API Keys live in the same region on the left nav (makes relationship clearer to understand)
[x] Alerting: show the langfuse id + clickable link on slack
[x] Alerting: show model name passed in for 'Message: LLM API call failed' messages
[x] Alerting: show
api_base
onRequests are hanging
requests[ ] have a search box on models view that allows user to see filtered rows of results
Motivation, pitch
-
Twitter / LinkedIn details
No response