Artificial Intelligence #226

Hey, in this issue: a new series of open-weight Phi-3 models; AlphaFold 3 predicts the structure and interactions of all of life’s molecules; Llama3 implemented from scratch; using ideas from game theory to improve the reliability of language models; wiley shuts 19 scholarly journals amid AI paper mill problems; an interpretable ML package; and more.

The sponsors of this issue are 42 Questions and Nebius.

• Wiley shuts 19 scholarly journals amid AI paper mill problems
• [Microsoft] A new series of open-weight Phi-3 models (3.8B, 7B and 14B)
• [MIT] Using ideas from game theory to improve the reliability of language models
• [Sponsored] Join the 42 Questions waiting list. Your colleagues will thank you

Imagine your best customer service agent and AI had a baby. Well, 42 Questions is better than that, making it easy to use AI to answer questions at scale. Join us on our journey to make knowledge more accessible. See for yourself for free until 2025. Launching June 2024. Join the waiting list.

• [Google] AlphaFold 3 predicts the structure and interactions of all of life’s molecules
• Accelerating Llama3 FP8 inference with triton kernels
• [Tool] imodels: Interpretable ML package for concise, transparent, and accurate predictive modeling
• [Sponsored] Fight for the best price on your H100 GPU cluster. Special offer from Nebius AI for those training large models.
• [Project] Llama3 implemented from scratch
• Microsoft offers cloud customers AMD alternative to Nvidia AI processors

More than 800,000 subscribers are reading this newsletter. If you are building an AI or a data product or service, you can become a sponsor of one of the future newsletter issues and get your business featured in the newsletter. Feel free to reach out to hello@truepositive.ca for more details on sponsorships.

Enjoy the newsletter? Please help us make it bigger and better by sharing it with colleagues and friends.