Artificial Intelligence #219

Hey, in this issue: Mamba explained; ~40% of LLM layers can be removed; Jamba: a first large competitive SSM-Transformer model; how OpenAI Sora works; matching 7B model performance with 1/3 activated parameters; and more.

The sponsors of this issue are Synaptiq and MyScale.

• [Explained] Mamba explained
• [ArXiv] ~40% of LLM layers can be removed with a minimum decrease in quality
• Jamba: a first large competitive SSM-Transformer model
• [Sponsored] > CLICK HERE TO GET YOUR ORGANIZATION'S AIQ? SCORE <

Visualize where to invest in AI with our scoring algorithm & analysis. Learn what AI technologies might be a good fit - Machine Learning, Intelligent Automation, Natural Language Processing, Machine Vision - based on your workflows, data, business needs, and how you should budget for AI solutions.

• [Video] The tech behind the magic: how OpenAI Sora works
• [Time] Nobody knows how to safety-test AI
• [Open LLM] Qwen1.5-MoE: Matching 7B model performance with 1/3 activated parameters
• [Sponsored] MyScaleDB, the Revolutionary SQL Vector Database, Goes Open-Source
• [Tutorial] Perform the image generation inference with just 6.4GB of VRAM
• [Project] SWE-agent turns LLMs into software engineering agents that can fix bugs and issues in real GitHub repositories

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 [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.