setzer22

Implementation of prompt caching

2

This was simpler than I expected, so time to sneak in a little PR :) This implements the same feature described in https://github.com/ggerganov/llama.cpp/issues/64 Basically, this adds an API to access...

Good ideas from llama.cpp

8

I've been tracking the `llama.cpp` repo. I'll use this issue to list any good ideas / things we should be aware of to keep up with in Rust land: -...

Use the HuggingFace llama Tokenizer

13

The [`tokenizers`](https://github.com/huggingface/tokenizers) crate by HuggingFace should give us a more correct tokenizer implementation than the one we're currently using. Looks like a LLaMA implementation already landed there https://github.com/huggingface/transformers/pull/21955, and then...

Embedding extraction

4

Implements #56. I ported the llama.cpp code to allow extracting word embeddings and logits from a call to `evaluate`. I validated this using an `ad_hoc_test` (currently hard-coded in `main`) and...

Swap strategy for infinite output

2

As discussed in https://github.com/ggerganov/llama.cpp/issues/71#issuecomment-1483907574 The idea is to achieve a naive implementation for infinite output generation using a strategy that simply clears the context window (you can keep the original...

The recently released [`lox`](https://github.com/LukasKalbertodt/lox) crate looks like a fast and usable library for geometry processing that could potentially replace the custom `HalfEdgeMesh` in blackjack :eyes: One particular similarity between the...

## Checklist - CI Checked: - [x] `cargo fmt` has been ran - [x] `cargo clippy` reports no issues - [x] `cargo test` succeeds - [x] `cargo rend3-doc` has no...

Error list keeps showing old cached results after the first render

5

I've updated my emacs packages, and I noticed this issue. It seems to be related (or identical) to #149, but I've been able to reproduce this in other language servers...

Firefox (and Thunderbird) pop-up windows immediately disappear

9

This is an issue that I've been dealing with for some time in `exwm`. I only experience this issue with Firefox and Thunderbird, no other program does this. What happens...

When using some of the arrow macros, like `-`, it seems type hinting does not work. Consider the following two examples: ``` (as-> "Hello" (.toUpperCase ^String )) (- "Hello" (.toUpperCase...