We are delighted to introduce the two latest additions to our party, Meta: Llama 2 70 B (beta) and Google PaLM 2 Bison!
These two new LLMs are extremely capable – here’s what you need to know:
Summarization Superpowers
Llama-2 matches GPT-4’s factual accuracy in experiments, making it perfect for high-integrity NLP. This means you can trust their summaries to be on point!
Serious Speed
These two new members of our community are incredibly quick! Prepare to be impressed.
Diverse Training Data
PaLM 2’s pre-training corpus includes a myriad of sources, such as web documents, books, code, and conversational data! Also, it’s multilingual, being trained in over 100 languages, giving it an exceptional contextual understanding and translation capabilities.
The Unfiltered Experience
One striking feature about these two is that they’re unfiltered! This means they have no moderation layers, and you get the AI, raw and undiluted. See the attached examples (apologies if they touch on NSFW topics, but it’s the clearest way to demonstrate their potential! )
We can’t wait to see what you’ll do with these new capabilities!
