Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Claude Memory

anthropic.com

Post reply on HN
Print view View original

311–320 of 326 posts

Re: Claude Memory

#311

Post by skeeter2020 »

brookst wrote: post #172
corry wrote: post #27

Earlier quoted context omitted.

Strong agree. For every time that I'd get a better answer if the LLM had a bit more context on me (that I didn't think to provide, but it 'knew') there seems to be a multiple of that where the 'memory' was either actually confounding or possibly confounding the best response. I'm sure OpenAI and Antropic look at the data, and I'm sure it says that for new / unsophisticated users who don't know how to prompt, that thi…

I'm pretty deep in this stuff and I find memory super useful. For instance, I can ask "what windshield wipers should I buy" and Claude (and ChatGPT and others) will remember where I live, what winter's like, the make, model, and year of my car, and give me a part number. Sure, there's more control in re-typing those details every single time. But there is also value in not having to.

until you ask it why you have trouble seeing when driving at night and it focuses on you need to buy replacement wiper blades.

Re: Claude Memory

#312

Post by xpe »

deadbabe wrote: post #211
xpe wrote: post #188

Earlier quoted context omitted.

What are we debating? Does anyone know? One claim seems to be “people should cease using any anthropocentric language when describing LLMs”? Most of the other claims seem either uncontested or a matter of one’s preferred definitions. My point is more of a suggestion: if you understand what someone means, that’s enough. Maybe your true concerns lie elsewhere, such as: “Humanity is special. If the results of our thinki…

If people think LLMs and humans are equal, people will treat humans the way they treat LLMs.

Looking over the comment chain as a whole, I still have some questions. Is it fair to say this is your main point?...

> Also, Claude doesn’t “think” anything, I wish they’d stop with the anthropomorphizations.

Parsing they above leads to some ambiguity: who do you wish would stop? Anthropic? People who write about LLMs?

If the first (meaning you wish Claude was trained/tuned to not speak anthropomorphically and not to refer to itself in human-like ways), can you give an example (some specific language hopefully) of what you think would be better? I suspect there isn't language that is both concise and clear that won't run afoul of your concerns. But I'd be interested to see if I'm missing something.

If the second, can you point to some examples of where researchers or writers do it more to your taste? I'd like to see what that looks like.

Re: Claude Memory

#313

Post by ToDougie »

DiskoHexyl wrote: post #150

CC barely manages to follow all of the instructions within a single session in a single well-defined repo. 'You are totally right, it's been 2 whole messages since the last reminder, and I totally forgot that first rule in claude.md, repeated twice and surrounded by a wall of exclamation marks'. Would be wary to trust its memories over several projects

Yep -- every message I send includes a requirement that CC read my non-negotiables, repeat them back to me, execute tasks, and then review output for compliance with my non-negotiables.

Re: Claude Memory

#314

Post by scottyah »

sheepscreek wrote: post #285
brulard wrote: post #177

Earlier quoted context omitted.

I would say these are two distinct use cases - one is the assistant that remembers my preferences. The other use case is the clean intelligent blackbox that knows nothing about previous sessions and I can manage the context in fine detail. Both are useful, but for very different problems.

Good point. I almost wish for an anonymous mode with chat history.

Well you're in luck! They have that feature and talk about it in the article

Re: Claude Memory

#315

Post by scottyah »

skeeter2020 wrote: post #311
brookst wrote: post #172

Earlier quoted context omitted.

I'm pretty deep in this stuff and I find memory super useful. For instance, I can ask "what windshield wipers should I buy" and Claude (and ChatGPT and others) will remember where I live, what winter's like, the make, model, and year of my car, and give me a part number. Sure, there's more control in re-typing those details every single time. But there is also value in not having to.

until you ask it why you have trouble seeing when driving at night and it focuses on you need to buy replacement wiper blades.

Claude, at least in my use in the last couple weeks, is loads better than any other LLMs at being able to take feedback and not focus on a method. They must have some anti-ADHD meds for it ;)

Re: Claude Memory

#316

Post by typpilol »

Sammi wrote: post #272
typpilol wrote: post #206

Earlier quoted context omitted.

My experience is with copilot and it uses various models, but the sweet spot is between 60 and 120 lines. With psuedo xml tags between sections Might be different across platforms due to how stuff is setup though.

My AGENTS.md is 845 lines and it only started getting good once it got that long. I'm still wanting to add much more... I'm thinking maybe I need a folder of short doc files and an index in AGENTS.md describing the different doc files and when to use them instead.

I know copilot supports nested agent files per folder.

Re: Claude Memory

#317

Post by srmatto »

morsecodist wrote: post #296

I am pretty skeptical of how useful "memory" is for these models. I often need to start over with fresh context to get LLMs out of a rut. Depending on what I am working on I often find ChatGPT's memory system has made answers worse because it sometimes assumes certain tasks are related when they aren't and I have not really gotten much value out of it. I am even more skeptical on a conceptual level. The LLM memories…

That is my experience as well. This memory feature strikes me as beneficial for Anthropic but not for end users.

Re: Claude Memory

#318

Post by labrador »

bongodongobob wrote: post #239
labrador wrote: post #96

Earlier quoted context omitted.

Parent is not chatting though. Parent is crafting a precise prompt. I agree, in that case you don't want memory to introduce global state. I see the distinction between two workflows: one where you need deterministic control and one where you want emergent, exploratory conversation.

Yes, you still craft an initial prompt with exploratory chats. I feel like I'm talking to a bot right now tbh.

The first sentence is mine. The second I adapted from Claude after it helped me understand why someone called my original reply insane. Turns out we're talking about different approaches to using LLMs.

Re: Claude Memory

#319

Post by labrador »

mnhnthrow34 wrote: post #309
labrador wrote: post #193

Earlier quoted context omitted.

The training data contains all kinds of truths. Say I told Claude I was a Christian at some point and then later on I told it I was thinking of stealing office supplies and quitting to start my own business. If Claude said "thou shalt not steal," wouldn't that be true?

Not necessarily. You know that it's true that stealing is against the ten commandments, so when the LLM says something to that effect based on the internal processing of your input in relation to its training data, YOU can determine the truth of that. > The training data contains all kinds of truths. There is also noise, fiction, satire, and lies in the training data. And the recombination of true data can lead to fa…

> Personifying the LLM as being capable of knowing truths seems like a risky pattern to me.

I can see why I got downvoted now. People must think I'm a Blake Lemoine at Google saying LLMs are sentient.

> If you find truth in what the LLM says, that comes from YOU, it's not because the LLM in some way can knows what is true

I thought that goes without saying. I assign the truthiness of LLM output according to my educational background and experience. What I'm saying is that sometimes it helps to take a good hard look in the mirror. I didn't think that would controversial when talking about LLMs, with people rushing to remind me that the mirror is not sentient. It feels like an insecurity on the part of many.

Re: Claude Memory

#320

Post by kromem »

tezza wrote: post #66

Main problem for me is that the quality tails off on chats and you need to start afresh I worry that the garbage at the end will become part of the memory. How many of your chats do you end… “that was rubbish/incorrect, i’m starting a new chat!”

So a thing with claude.ai chats is that after long enough they add a long context injection on every single turn after a while.

That injection (for various reasons) will essentially eat up a massive amount of the model's attention budget and most of the extended thinking trace if present.

I haven't really seen lower quality of responses with modern Claudes with long context for the models themselves, but in the web/app with the LCR injections the conversation goes to shit very quickly.

And yeah, LCRs becoming part of the memory is one (of several) things that's probably going to bite Anthropic in the ass with the implementation here.

Post reply on HN