Josh Leverette
Josh Leverette
Both of the examples you provide seem very nice to me. It's true that supporting every possible HTTP endpoint is difficult, but supporting a useful subset _somehow_ seems immensely worthwhile...
Considering that any implementation of this concept of services would be highly opinionated, if it gets implemented, it might be worthwhile to include an option in `reproto` to not emit...
This looks awesome! I'm glad to see progress happening here!
Semi-related, but isn't k-quant the newer/better quantization method? I have found it confusing that ollama defaults to the non-K quants, but maybe I'm confused about which method is better.
The underlying library `llama.cpp` [does not support Phi-3-128k yet](https://github.com/ggerganov/llama.cpp/issues/6849#issuecomment-2074899603), so there's nothing `ollama` can do to support it yet.
@dhiltgen are you talking about Phi3, or Phi3-128k? ollama mentions nothing about the 128k context model: https://ollama.com/library/phi3/tags
Personally, I think it is better to allow interior whitespace, but limit it to up to one new line and up to a small number of spaces between fields. Papers...
I also wish we could get a maintainer to look at these changes… it has been awhile. @jmorganca ?
Actually, I went ahead and started playing with it. The Free Angle Autorouting is really cool, but it doesn't even come close to being able to create initial routes for...
It's a commercial property, so I can't just publish it at the moment, but if you send me an email to the account (my username) on the email provider Gmail,...