John
John
I am trying to repair functionality of Copilot. Using IDE scripting console I can call most things I need (util = com.github.copilot.util.ApplicationUtil) but the javascript does not offer any docs...
``` public static IObservable KeyUpObservable(this IKeyboardEvents source) { return Observable .FromEventPattern(source, "KeyUp") .Select(ep => ep.EventArgs.KeyCode); } ``` It uses "KeyDown" there which is a typo from copy/paste of the Keydown...
The readme makes it sound very simple: "Replace bert with xphonebert" Looking a bit closer looks like it's quite a feat to make StyleTTS2 talk in non-english languages (https://github.com/yl4579/StyleTTS2/issues/28) StyleTTS2...
I've been comparing OpenBuddy and OpenAssistant using Falcon 40B, in both cases the latest finetunes available. I noticed OpenBuddy to deliver much better responses, it sticks a lot better to...
I noticed the readme mentions Falcon is not working on ggml, the bug has been solved (https://github.com/cmp-nct/ggllm.cpp/issues/68) There are a couple issues remaining (no prompt cache and no batched processing)...
A couple things I noticed that would be great to have: 1) CSS tweak - the background is hard to read for some people. My dad can't read the text...
Latest dev version: Input: cache.buf.resize(2u*n_elements*ggml_type_size(wtype) + 2u*MB); Output: cache.buf.resize(2un_elementsggml_type_size(wtype) + 2u*MB); This happens with the prompt as well as when Bing answers, I'm quite sure the answer contains the correct...
I think a good part of potential users will not be able to get PandoraAI running. It requires quite a bit of dedication, reading and knowhow to work. My suggestion...
I'm stuck with other work, I recently pushed half finished branch, containing a ton of fixes and changes but not finished. Also moved from falcon_main to "ggfalcon" which is meant...
I plan to PR today, though it depends on final progress. The computation speed is slow because we currently have no mulmat kernel with interleaving broadcast support yet, so tests...