homarp

31 ms·

571.
by homarp 3y ago
discussed here https://news.ycombinator.com/item?id=39978717
572.
by homarp 3y ago
CodeGemma comes in three flavors: A 2B base model specialized in infilling and open-ended generation. A 7B base model trained with both code infilling and natural language. A 7B instruct model a user can chat with about code.
573.
CodeGemma – an official Google release for code LLMs(huggingface.co)
4 points by homarp 3y ago|2 comments
574.
by homarp 3y ago
Rewrite in (safe) Rust, I am sure many will enjoy to compare how it runs.
575.
by homarp 3y ago
it is AGPL
576.
by homarp 3y ago
I don't think anything is "pretty much guaranteed": things evolve
577.
by homarp 3y ago
Here are links to the most recent pull requests sent https://github.com/ggerganov/llama.cpp/pull/6414 https://github.com/ggerganov/llama.cpp/pull/6412
578.
by homarp 3y ago
Author compared various whisper implementation "We found that WhisperX is the best framework for transcribing long audio files efficiently and accurately. It’s much better than using the standard openai-whisper library." https:&#
579.
SOTA ASR Tooling: Long-Form Transcription(amgadhasan.substack.com)
3 points by homarp 3y ago|2 comments
580.
Towards 1-bit Machine Learning Models(mobiusml.github.io)
351 points by homarp 3y ago|157 comments
581.
by homarp 3y ago
it is because GPL is viral https://www.gnu.org/licenses/gpl-faq.html#WhatDoesCompatMean
582.
by homarp 3y ago
meli last commit is yesterday. Last release 3 months ago
583.
by homarp 3y ago
or use tvm
584.
13th Edition of Basic 10 Liner Contest(gkanold.wixsite.com)
1 points by homarp 3y ago|0 comments
585.
Thunder, Source to source compiler for PyTorch(github.com)
4 points by homarp 3y ago|0 comments
586.
Cosmopedia: How to create large-scale synthetic data for pre-training(huggingface.co)
2 points by homarp 3y ago|0 comments
587.
by homarp 3y ago
https://github.com/squirrelinhell is the author
588.
by homarp 3y ago
more discussion at https://news.ycombinator.com/item?id=39722498
589.
Using Godot for GUI App Development(popcar.bearblog.dev)
3 points by homarp 3y ago|0 comments
590.
A little guide to building Large Language Models in 2024(docs.google.com)
5 points by homarp 3y ago|0 comments
591.
by homarp 3y ago
code from the paper https://arxiv.org/abs/2401.18059 It introduces a novel approach of recursively embedding, clustering, and summarizing chunks of text, constructing a tree with differing levels of summarization from
592.
Raptor: Recursive Abstractive Processing for Tree-Organized Retrieval(github.com)
5 points by homarp 3y ago|1 comments
593.
Ace of Aces: Why you should do maths as a game designer(paxsims.wordpress.com)
223 points by homarp 3y ago|42 comments
594.
by homarp 3y ago
In Rust, nice implementation of Mixtral 8x7B
595.
Like grep but for natural language questions. Based on Mixtral 8x7B(github.com)
2 points by homarp 3y ago|1 comments
596.
by homarp 3y ago
see also https://old.reddit.com/r/LocalLLaMA/comments/1b1q88w/selfext... who could help with code because of huge context.
597.
by homarp 3y ago
This is the sequel of StarCoder and The Stack (v1) - https://news.ycombinator.com/item?id=35954481 StarCoder2 has 3 models, biggest one is StarCoder2-15B, trained on 600+ programming languages from The Stack v2 on 4+ trilli
598.
by homarp 3y ago
see Pingora, https://news.ycombinator.com/item?id=39540594
599.
by homarp 3y ago
Pingora is a Rust framework we have been using to build services that power a significant portion of the traffic on Cloudflare. Pingora is released under the Apache License version 2.0. see also https://news.ycombinator.com/
600.
MicroZig: Unified abstraction layer and HAL for Zig on several microcontrollers(github.com)
249 points by homarp 3y ago|60 comments

More ›