🦙Meta’s Llama 3 Dominates Open Source LLMs

Robby Boney
Short Bits
Published in
Sent as a

Newsletter

3 min readApr 19, 2024

Meta releases Llama 3

Meta finally released their much awaited Llama 3 Large Language Model in 8B and 70B parameter model sizes. Mark Zuckerberg also stated in an interview with Dwarkesh Patel that a 400B parameter model version is also still training. Llama 3 is showing itself to perform near similar sized models and even come close to a GPT-4 level experience 🤯

See More: Introducing Meta Llama 3: The most capable openly available LLM to date

Google Next

Google had their Cloud Next 2024 conference and announced a huge list of things. One thats particularly interesting and shows Google wants to take a lead, is their Agent Builder product on Vertex AI. This makes agents easy to build, manage and deploy with a simple interface.

See More:

2 Papers on Infinite Length Context Windows

A Paper from a team at Meta and another one from Google are showing ways that LLMs may be able to have an unlimited context window, which would far surpase the largest windows today such as Gemini 1.5’s 1 million token context window.

See More:

--

--

Robby Boney
Short Bits

Director of Product Development @ Interject Data Systems. I write Software for Science, Data & Enterprise…https://robbyboney.notion.site/