Hypothesis: When an LLM agent doesn't do what you want or expect (stuck in a loop, irrelevant directions, not getting into practical details), it's not because it has inherent limitations but because it's trying to guess what you want it to do and hasn't had enough guidance from
@marek_rosa
-
Hackathon with continued development planned
By
–
Jj, na hackathone by ste sa hodili Dobra sprava je ze tento hackathon asi bude mat pokracovania.
-
Design Roadmap Development for Improved User Experience
By
–
We may eventually make it OS… now we want to look on the design again, make a roadmap and release something that will be easy to use and show the benefits/potential.
-
Multi-agent LLM system wins third place at Prague AI Hackathon
By
–
My team finished in 3rd place at this weekend's Prague AI Hackathon Project title: Multi-agent AGI We built a prototype of a multi-agent system with seven LLM agents connected in a hierarchical structure. Initially generalists, these agents adapt and specialize over time as
-
LTM Library Release: Long-Term Memory for AI Systems
By
–
We will release our LTM (long-term memory) library in a week or two. It's a layer above the vector database, helping you store and retrieve memories and pre-process and post-process them to maximize retrieval quality. We have focused on dialogue memories but plan to look at
-
LLMs for High Level Planning and Context Assumption
By
–
Yes, we are using LLMs for high level planning, like that “bring me coffee” example. Especially because LLMs seem to assume the right context even if we don’t specify it. But here I am not talking about the edge cases when the assumptions fail.
-
Building Games with LLM-Driven Non-Player Characters
By
–
We are actually building a game with LLM driven NPCs https://t.co/LDaBjAdTbt
— Marek Rosa | European🇪🇺 | South African🇿🇦 (@marek_rosa) 16 avril 2023We are actually building a game with LLM driven NPCs
-
Agents Building Themselves: Self-Improving AI Systems
By
–
That’s the point Use agents to build themselves.
-
Current AI Systems Safe, Future Iterations Need Preparation
By
–
This is good point, there’s definitely a lot that can be studied on today’s systems for years. On the other hand, current systems as they are are not dangerous. Only one of their next iterations could be dangerous and we can prepare for it only if we have its predecessor.
-
AI Agents Moving Too Fast for Human Monitoring
By
–
Yeah, this is a good answer. AI agents will be too fast for humans to monitor and adapt to.