MCP is a powerful new tool for software builders. Reading this post from a security expert should make builders and users a bit more nervous about it. “In just the past few weeks, the Model Context Protocol (MCP) has rapidly […]
How do you get a dumb LLM to do your bidding? You have to write a very good system prompt! Nobody wants theirs exposed as it’s intellectual property, but it seems that they inevitably leak because LLMs are inevitably jailbroken. […]
You know how every research agency and big company likes to public an annual trends report? Imagine taking 233 of those and feeding them into an LLM to extract the key points by industry. Someone did that, then summarized it […]
This is a fun nerdy explanation of embeddings, a sort of math function that’s behind our favorite new LLM technology. ~ learn more
This is a fun nerdy explanation of embeddings, a sort of math function that’s behind our favorite new LLM technology. ~ learn more
Not really. “The response itself depends on the amount of layers offloaded onto the GPU, and your drivers. Basically you can only get the same output with the exact same input with exactly the same setup, that’s consistent, but change […]
“It’s not a SQL statement that would work today in a cloud data warehouse. But an LLM would understand it : summarize the book Moby Dick in two sentences.” ~ learn more
This research makes the claim that using “Chain of Thought” prompting for LLMs does not generalize. I consider this evidence of LLM limitations (at least for now?). “This spotlights drawbacks of chain of thought, especially because of the sharp tradeoff […]
“For the analysis, I used an LLM to embed the companies’ value propositions in a vector space, ran a clustering algorithm on top of the embeddings, manually analyzed the results, and used ChatGPT to edit the draft. The conclusions are […]
This starts out basic and gets complex toward the end. “In this paper, we introduce core concepts, advanced techniques like Chain-of-Thought and Reflection, and the principles behind building LLM-based agents. Finally, we provide a survey of tools for prompt engineers.” ~ learn […]