- LLms guess what functions to call, such as running a calculation in Python or a function to do a web search.
- Efficient Tool Use with Chain-of-Abstraction Reasoning, Gao et al. (2024)
Microsoft Absorbs Inflection, Nvidia’s New GPUs, Managing AI Bio Risk, and more
Four AI Agent Strategies That Improve GPT-4 and GPT-3.5 Performance
- AI agenic workflows can help AI achieve 95% code quality vs zero shot.
Agentic Design Patterns Part 2: Reflection
- Here’s code intended for task X: [previously generated code] Check the code carefully for correctness, style, and efficiency, and give constructive criticism for how to improve it.
- Next, we can prompt the LLM with context including (i) the previously generated code and the constructive feedback and (ii) ask it to use the feedback to rewrite the code
Agentic Design Patterns Part 3: Tool Use
FirstMark | 2024 MAD (ML/AI/Data) Landscape - there are too many tools for any one person or organization to know what to do with. It’s a full time job just to understand what’s out there.
https://www.linkedin.com/posts/luke-b_data-nerds-what-technology-should-you-learn-activity-7181698703722729472-N03S?utm_source=share&utm_medium=member_ios
Gecko: Versatile Text Embeddings Distilled from Large Language Models - Google DeepMind
- gecko is a text embedding model that can do more with 256 tokens than most can with 700+
_________________________
Bryan lives somewhere at the intersection of faith, fatherhood, and futurism and writes about tech, books, Christianity, gratitude, and whatever’s on his mind. If you liked reading, perhaps you’ll also like subscribing: