Hacker Newsnew | past | comments | ask | show | jobs | submit | eurekin's commentslogin

I personally started using AI with a very rudimentary "memory" system, and for first weeks corrected it ad nauseam, until, in the small domain I was responsible for, it started doing work end to end fully autonomously. Takes a lot of time and effort to bootstrap, but once steered well enough, it just clicks and less and less is needed to check manually


Yeah a memory system per domain, or per module or even per file works amazing, we have it too. Took a ton of time to improve the code and list down all the tribal knowledge, but now that it's done, it works super well.


That's my personal suspicion, why there's no real coding tutorials like that on YouTube. Having so many eyes to basically rip to shreds your output is probably the reason.


Except now that no one cares about handwritten code, maybe instead people will celebrate the diversity of ways one can write code to do things. Might see a resurgence in little artisanal videos like these


> I'm sure I would have loved to have a real phone as a toddler, but my sister and I got a lot of mileage out of our Fisher Price phone, making lots of pretend calls, which was probably better for our developing brains than a real phone.

I was promised a PC at around 11 years. I think for months (or even more than a year) I fantasized about, what I could do with it. Read some books, that I had access to. Understood practically nothing, but that didn't stop me from dreaming about the bright future. The book might have been "Amiga E programming reference" and it had two chapters dedicated to the blitter and sprite programming. I reread those chapters many times, trying to imagine how to make a simple game with it.

The promised PC came, but it was a work decommissioned 486 at the time, when everyone else was upgrading to Pentiums 200 to play the latest Lara Croft. My sister was bummed out that no software from friends worked on it.

I somehow was able to make a very simple arkanoid game anyway, in C and OpenGL software renderer. Pretty sure, that the main reason I managed to do anything, was all the time spent expecting the PC and imagining possibilities.


> ~1000 bytes per context token per user

Where's that figure comning from? Last time I checked (could be the 3.6 Qwen 27b) single token needed 32kb


The smallest I have seen is DeepSeek 4.1 flash at 890 bytes per token.


I suspect that commercial labs have unpublished work lowering this further. It's the obvious thing to optimise for when RAM is so expensive and you have a lot of users.


I still can't fathom my company application security decision. Found pretty damning requests in our logs. Escalated. Expected it would result in at least reporting the TOS break from the originating place (one of cloud providers). Instead of that, they just went: "yeah, but we don't have logs". Provided them. "Yeah, but that ip doesn't resolve". I matched real ones from the load balancer. "There could just be many of them". There was one. When I had all the evidence gathered, they looked at me and finally told:

- It's just an Independent Security Researcher.

- So that's it? You will do no action?

- Correct


It is? I still found it invaluable, when having a lot of externally managed knowledge. Basically the only gate before it goes on a certain failure, when dealing with proprietary libraries, tools and services


Brain fog? This might be a legitimate symptom worth seeing a doctor for


Have you ever gotten an IQ test?


Yes, my mom has never told me the score, but it led to me being put in gifted classes. I took the SAT at 12 and did well enough to make JHU CTY's SET program.

So in terms of raw brainpower it's there, but I've never really been able to apply it consistently for whatever reason and have ended up very average. Still haven't really come to terms with that in my late 20s.

I've got a perfectly good life but I have this vague "what if" about if I had applied myself more.


Supposedly having over 125 (130-ish) can affect "normal functioning", it's very counterintuitive, but I'd recommend to just go through common psychology checks for that case.


Care to share any details? I'm about to check the 2.6b lfm on document editing.


I found that they didn't want to call tools in the same way as other models, and it led to bare tool calls in the response.


How are you serving them? This is most often caused by an incorrect template. (the thing that tells the inference engine how to parse the think/tool parts of the answer)

For vllm the official recipe [1] (for another model in the same family) is this:

  --reasoning-parser qwen3 \
  --enable-auto-tool-choice \
  --tool-call-parser lfm2
[1] - https://recipes.vllm.ai/LiquidAI/LFM2.5-8B-A1B


It does? I wonder, what it does to the gut biome then


It's already bad for your cardiovascular system. It's food as a mouth wash but don't swallow it, especially at the recommended doses of 3+ mouthwash per day with 3-5g each time


The recent study on it gave the subjects a 300mL "standardized, artificially sweetened beverage containing 30g of xylitol diluted in water."

Recommended amount is 5–10g per day (5 pieces of 100% xylitol gum, 36% is already effective), taken after meals.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: