Video yükleniyor...
Video Yüklenemedi
Why the memory layer will become more and more important and how Instinct crushed it “One thing about agents that is rare today but will be incredibly common in the next few years is this memory layer. If you look at Instinct or Muse, one of the things that... show more
31,158 görüntüleme • 9 gün önce •via X (Twitter)
14 Yorum
i completely agree (and respect) @hwchase17!
In enterprise, the hard part of the memory layer is permissions. An agent that remembers everything will eventually tell one user what only another was allowed to see.
That’s what makes an agent feel like yours
memory is the part users feel first. nobody churns over model quality, they churn because it forgot them
For small business use cases memory is the whole game. A support agent that forgets the customer complained last week does more damage than no agent at all.
The real unlock is when agents stop starting from zero every conversation.
if instinct remembers something wrong about you, can you actually see what it stored and edit it? or does the memory only show up in how it acts
memory is the difference between a tool you use and a coworker you trust
i still nudge my agents on the same stuff every session, memory would help
Totally agree with this
memory is the part that makes agents feel useful day to day without it you're re explaining your whole life every session
Memory layer is the real moat. @mozlyxy has been making a similar case for how this plays out in public markets.
本质还是上下文的组织问题，确实是还是非常早期的阶段， context window 类比是CPU寄存器缓存，内存是可以存储更为海量的知识的，如何协作就是一个有意思的研究方向
Were you not using OpenClaw since February. Muse it OpenClaw but 10x earlier for eveyone.
![The most interesting part for me is where Andrej Karpathy describes why LLMs aren't able to learn like humans. As you would expect, he comes up with a wonderfully evocative phrase to describe RL: “sucking supervision bits through a straw.” A single end reward gets broadcast across every token in a successful trajectory, upweighting even wrong or irrelevant turns that lead to the right answer. > “Humans don't use reinforcement learning, as I've said before. I think they do something different. Reinforcement learning is a lot worse than the average person thinks. Reinforcement learning is terrible. It just so happens that everything that we had before is much worse.” So what do humans do instead? > “The book I’m reading is a set of prompts for me to do synthetic data generation. It's by manipulating that information that you actually gain that knowledge. We have no equivalent of that with LLMs; they don't really do that.” > “I'd love to see during pretraining some kind of a stage where the model thinks through the material and tries to reconcile it with what it already knows. There's no equivalent of any of this. This is all research.” Why can’t we just add this training to LLMs today? > “There are very subtle, hard to understand reasons why it's not trivial. If I just give synthetic generation of the model thinking about a book, you look at it and you're like, 'This looks great. Why can't I train on it?' You could try, but the model will actually get much worse if you continue trying.” > “Say we have a chapter of a book and I ask an LLM to think about it. It will give you something that looks very reasonable. But if I ask it 10 times, you'll notice that all of them are the same.” > “You're not getting the richness and the diversity and the entropy from these models as you would get from humans. How do you get synthetic data generation to work despite the collapse and while maintaining the entropy? It is a research problem.” How do humans get around model collapse? > “These analogies are surprisingly good. Humans collapse during the course of their lives. Children haven't overfit yet. They will say stuff that will shock you. Because they're not yet collapsed. But we [adults] are collapsed. We end up revisiting the same thoughts, we end up saying more and more of the same stuff, the learning rates go down, the collapse continues to get worse, and then everything deteriorates.” In fact, there’s an interesting paper arguing that dreaming evolved to assist generalization, and resist overfitting to daily learning - look up The Overfitted Brain by Erik Hoel. I asked Karpathy: Isn’t it interesting that humans learn best at a part of their lives (childhood) whose actual details they completely forget, adults still learn really well but have terrible memory about the particulars of the things they read or watch, and LLMs can memorize arbitrary details about text that no human could but are currently pretty bad at generalization? > “[Fallible human memory] is a feature, not a bug, because it forces you to only learn the generalizable components. LLMs are distracted by all the memory that they have of the pre-trained documents. That's why when I talk about the cognitive core, I actually want to remove the memory. I'd love to have them have less memory so that they have to look things up and they only maintain the algorithms for thought, and the idea of an experiment, and all this cognitive glue for acting.”](https://image.24vids.com/tw-1979259041013731752/amplify_video_thumb/1979257571786543104/img/OqUh56yit9gUQWnb.jpg)