feat: model selection error handling #73
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
| @@ -68,6 +72,29 @@ export const SettingsModal = memo(function SettingsModal({ | |||
| } | |||
| }; | |||
|
|
|||
| const handleChangeModel = async (v: string) => { | |||
| if (await checkModelAccess(v)) { | |||
There was a problem hiding this comment.
should we add a loading state with the <Spinner /> when calling handleChangeModel? User experience feels less snappy when selecting new model with small delay.
There was a problem hiding this comment.
hmmm if we could do with a small diff that might be nice but its probably fine as is imo
transmissions11
changed the title
|
hey sorry for the delay here — this code looks great! however... the whitelist checking just doesn't seem to work. i have both gpt-4 and 32k access and when i use the vercel preview for this PR i cant use either of them? |
|
I just opened #83, which has some overlap with this (but fetches model list on app launch / API key change, and offers all available gpt chat models to the user). |
|
@adietrichs correct me if im wrong but #83 basically replaces this PR, i dont see a reason for them to coexist — and #83 is probably the ideal solution |
|
feel free to correct me again if im missing something but going to close this PR for now as i feel #83 supersedes it |
Motivation
Requested in #26 since GPT-4 nor GPT-4-32k isn't available for everyone yet and it would be a better experience to see something pop up instead of not being sure what happened.
Solution
On model change in settings:
Checklist
PR-Codex overview
This PR adds a new function
checkModelAccessto validate access to OpenAI models and updates theSettingsModalcomponent to use it when changing the selected model.Detailed summary
checkModelAccessfunction to validate access to OpenAI modelsSettingsModalcomponent to usecheckModelAccesswhen changing the selected modeluseToasthook to display error messages when changing to an unsupported model