The ML with PyTorch one? Agreed. I hope to get to it in the next 1-2 years!
OPEN SOURCE
-
GQA Implementation Guide for Llama Model Conversion
By
–
Nice work! In case you get stuck, I've implemented GQA as part of the GPT -> Llama conversion guides: https://
github.com/rasbt/LLMs-fro
m-scratch/tree/main/ch05/07_gpt_to_llama
…
But since you already implemented MQA, GQA should be pretty smooth sailing 🙂 -
Windsurf Wave 2: AI for Developers
By
–
Loads of updates from Windsurf this month: Web search, memory, Deepseek R1 and o3-mini additions, plus much more 👀
— 🚨 AI News | TestingCatalog (@testingcatalog) 9 février 2025
"Windsurf wave 2: When AI actually gets how developers work" documented by @kernelkook on TestingCatalog 👇 https://t.co/ECOFP4EQVd pic.twitter.com/O41cxdtxjdLoads of updates from Windsurf this month: Web search, memory, Deepseek R1 and o3-mini additions, plus much more "Windsurf wave 2: When AI actually gets how developers work" documented by @kernelkook on TestingCatalog
-
Lightweight Python-based tool coming soon via pip
By
–
never used it. mine is python-based, light weight. will release soon and anyone can install via pip
-
Using vLLM with Llama 8B GGUF for Local Backend
By
–
backend can be anything. im using vllm with llama 8b gguf locally
-
Open-sourcing AI from the inside strategy
By
–
We don’t manage to make them open-source enough from the outside (IMO they could be #1 in AI) so we’re trying to do it from the inside now haha
-
Deepseek Distilled Qwen: Smaller Model with Good Performance
By
–
Try deepseek distilled Qwen, much smaller and reasonably good.
-
Deep Research Agent: Automate Research with Open-Source AI
By
–
Day 9 – 25 Days of AI for ALL in 2025 Revolutionize research with AI! Manual research? Too slow.
Sifting through data? Inefficient.
AI-driven research? Fast, smart, and scalable.
Introducing: Deep Research Agent – an open-source AI tool that automates deep research, -
Why Everyone Needs Offline AI Models as Backup
By
–
Everyone should have a few offline models downloaded as backup.