Issue #2: LLMs, LLMs, LLMs!
Issue #2: LLMs, LLMs, LLMs!
Google introduces two new variants for developers and researchers, Meta confirms LLama 3 launch timeline
Welcome to Issue #2 of One Minute AI, your daily AI news companion. This issue will cover two major updates in the world of LLMs from Google and Meta.
Google adds new models to the Gemma family
Google announced two new additions to the Gemma family of Large Language Models (LLMs) catered to developers and researchers: CodeGemma and RecurrentGemma.
CodeGemma brings powerful yet lightweight coding capabilities to the community. CodeGemma models include a 7B pre-trained variant that specializes in code completion and code generation tasks, a 7B instruction-tuned variant for code chat and instruction following, and a 2B pre-trained variant for fast code completion that fits on your local computer.
RecurrentGemma enables efficient, faster inference at higher batch sizes for researchers. It leverages recurrent neural networks and local attention to improve memory efficiency. While achieving similar benchmark performance to the Gemma 2B model, RecurrentGemma’s unique architecture brings several pros, such as reduced memory usage and higher throughput.
Meta confirms that Llama 3 is coming next month
At the Meta AI Day in London, Nick Clegg, Meta’s President of Global Affairs, shared that they hope to roll out Llama 3, the next generation of their LLMs for GenAI assistants, within the next month.
While the specifics are unknown, it is expected that Llama 3 with have about 140 billion parameters (in comparison, the largest model in Llama 2 had 70 billion parameters). Meta intends to integrate Llama 3 across their diverse product ecosystem, with their long-term goal being a Llama-powered Meta AI.
Want to help?
If you liked this issue, help spread the word and share One Minute AI with your peers and community.
You can also share feedback with us, as well as news from the AI world that you’d like to see featured by joining our chat on Substack.
