-
Notifications
You must be signed in to change notification settings - Fork 69
It should be possible to use InferencePool without ext_proc #660
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Agreed, the idea behind the EndpointPickerConfig is to provide a one-of config, currently the only option is ExtensionRef, and we were planning to relax the Do you have something concrete in mind to spec as a second option to |
I think have an "implementation specific" option that is just up to the provider is probably a good start. I'm not sure what else would be standardized |
Agreed. 👍 The API needs to be more generally applicable, and not coupled so tightly with a specific implementation. Some kind of resolution here would be ideal as a requirement for a GA release. |
+1 Having the |
Note that the current API is designed with that in mind as I mentioned above, the If there is agreement on the above, the next step is to propose a second option within |
+1 to optional |
Agreed |
What would you like to be added:
The InferencePool current hardcodes that a user MUST specify and Endpoint picker.
gateway-api-inference-extension/api/v1alpha2/inferencepool_types.go
Line 69 in 6058b09
As a vendor agnostic API, we should instead let the API drive behavior not implementation. An implementation should be free to chose it's own implementation of the semantics of InferencePool
The text was updated successfully, but these errors were encountered: