Artificial Intelligence #214

Hey, in this issue: the era of 1-bit LLMs; OpenAI accuses NYT of hacking ChatGPT; why people spent 60 years building chatbots; a framework for prompt tuning using intent-based prompt calibration; LoRA+ is 2 times faster compared to LoRA; RLHF in 2024; and more.

The sponsors of this issue are Tecton and Galileo.

• [ArXiv] The era of 1-bit LLMs: all large language models are in 1.58 bits
• [WP] The future of AI video after Sora is impressive — and flawed
• [ArsTechnica] OpenAI accuses NYT of hacking ChatGPT to set up copyright suit
• [Sponsored] [Virtual] apply() 2024: Mastering AI and ML at Production Scale

Join the next apply() conference on April 3, for a free event that brings together the engineering community to master AI and ML in production. Past speakers include engineers and tech leads from Uber, Pinterest, Databricks, Riot Games - and we’re finalizing an equally fantastic lineup for April!

• [Discovery] LoRA+ is 2 times faster compared to LoRA and provides an increase in quality
• [The Verge] From Eliza to ChatGPT: why people spent 60 years building chatbots
• [Tool] AutoPrompt: A framework for prompt tuning using intent-based prompt calibration
• [Sponsored] Learn from top GenAI experts on how to build enterprise GenAI. Register now for GenAI Productionize 2024 for FREE!
• Don't mock machine learning models in unit tests
• [Tutorial] RLHF in 2024 with DPO & Hugging Face

More than 750,000 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 [email protected] for more details on sponsorships.

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