Newly introduced LLM Inflection-2 claims to outperform Google's flagship PaLM 2, only 2nd to GPT4

Inflection AI has announced the release of Inflection-2, a new large language model (LLM) that is the second most capable LLM in the world today after GPT4 by OpenAI.

What are LLMs?

Large language models (LLMs) are computer programs that can generate text, translate languages, write different kinds of creative content, and answer questions informatively.

Inflection-2 was trained on a massive dataset of text and code. This dataset was so large that it required 5,000 NVIDIA H100 GPUs to process. ~10²? FLOPs, to be precise, almost. Inflection claims it outperforms Google’s flagship PaLM 2 Large model on most standard AI performance benchmarks.

In addition to its technical prowess, Inflection-2 is also designed with serving efficiency in mind. This means that Inflection-2 can be used to power real-time applications, such as chatbots and virtual assistants. Inflection-2 will soon be powering Pi, a new personal AI device from Inflection AI.

I have used different AI-powered assistants, including ChatGPT and Bard, but I found Pi to be the most cozy and welcoming. Its warm color, font, and uncluttered screen create a sense of comfort that I didn’t feel with the others. Additionally, Pi has a dedicated safety team ensures its AI models are rigorously evaluated and aligned with human values. Inflection AI was the first company to sign up to the White House’s voluntary commitments for AI safety and alignment in July 2023.

Overall, I am excited to see how Inflection-2 creates new and innovative products and services.