Data Machina #251

Data Machina · 2 years ago

This newsletter curates six AI activities for a long weekend, including generating comics with ByteDance's StoryDiffusion model, learning to build AI agents using CrewAI and Groq, playing the AI Town game locally with Llama-3, reading research on in-context learning versus fine-tuning, reviewing Scale AI's 2024 readiness report surveying 1,800 practitioners, and studying a primer on differentiable programming for neural networks. Key concrete detail: Scale AI's research team interviewed 1,800 AI/ML practitioners for their 2024 State of AI Readiness Report. The article signals growing interest in practical AI agent development and debate within the research community about whether in-context learning with long-context models can replace fine-tuning for domain-specific accuracy.