Repeated sampling from instruction-tuned models can yield low-diversity, stereotyped output. For creative tasks, k-shot prompting base models will improve creativity (measured e.g. as vocabulary size), though generally with worse adherence to the (now implicit) instructions.
GENERATIVE AI
-
LLM prompting failure fixed with k-shot examples using Claude
By
–
In these screenshots our (illustrative, toy) task is to write a one-line description of a cleaning product. A base LLM will not respect instructions to this effect. The task must be communicated by repeating k-shot examples, which we generate using Claude.
-
Every AI Model Since GPT-4 Looks Identical
By
–
Update: On this graph, literally every model that has come out since GPT-4 still looks pretty much like … GPT-4.
-
CEO AI Strategy: Task-Based Analysis Approach for Boards
By
–
Almost every CEO is being asked by their board of directors “what is your game plan for generative AI?” Many of them are implementing an approach based on what we call the task-based analysis.
-
GenAI Engineering Basics Course on Databricks in 2 Hours
By
–
Learn the basics of GenAI engineering in just 2 hours! This self-paced course will help you get started with GenAI on Databricks. Learn how to build a compound AI system, fine-tune models, drive AI governance, and more: https://
bit.ly/4fPUk6S -
LLMs for Coding: Productivity and Educational Value Considerations
By
–
I'm not against LLMs for coding. I think there's a range of situations where they save time and have educational value. However, in my own workflows I have not found them to lead to any productivity boost. Mind you, I was also virtually never using SO in the pre-LLM days, whereas
-
LLM Limitations in Coding: Expert Skepticism Grows
By
–
Coding used to seem like the best use for LLMs, but in recent days @burkov and @fchollet have both come out strongly against even that use case. Both tried; both have been left disappointed.
-
AI Breakthroughs Driving Productivity Boom Insights
By
–
And part is based on the recent breakthroughs in AI: https://
brookings.edu/articles/machi
nes-of-mind-the-case-for-an-ai-powered-productivity-boom/
… -
Productivity J-curve framework for emerging General Purpose Technologies
By
–
Part of my reasoning is based on the Productivity J-curve for powerful new General Purpose Technologies (or simply GPTs as we once got to call them).
-
AI-Powered Productivity Boom: What’s Next?
By
–
Are we on the verge of an AI-powered productivity boom? @BrianCAlbrecht makes the case over at @Noahpinion
's blog.