Aaron Pham

Will ping once it is ready, discussing with Yixin atm to clear up some confusion

@WoosukKwon the initial implementation should be ready for first round of eyes now. also cc @russellb whenever you have a chance, and @mgoin if you are interested

cc @mmoskal probably also want your input on this as well.

Ah doc should be fixed now

> I'm not sure the added complexity is worth it, IDEs can follow the strings anyway: > >  Yeah, let's just close this PR then?

> * The proposed tokens need to be checked for validity according to the matcher. For each draft token, it should either be restricted to be sampled from the guided...

Oh this is added in 3.13. I will take a look for backport on this then

Hmm, then maybe similar implementation as the builtins behaviour of 3.13 would suffice?

@hmellor should be ready for another round of review

example https://github.com/yuchanns/avante_bedrock.nvim We will need to cosnider how to support different model with providers such as bedrock. for example you can run mistral, claude, command-r, and even jamba on there.