Hacker News: carterschonwald

New comment by carterschonwald in "Reducing Doom Loops with Final Token Preference Optimization"

carterschonwald — Mon, 13 Jul 2026 19:05:42 +0000

ive had doom loops on release day with opus 4.6. quantization aint the culprit ;)

New comment by carterschonwald in "Precursor"

carterschonwald — Mon, 13 Jul 2026 15:35:00 +0000

even before the llm era sites would flag me as a bot for opening 15 links to read later. its fucking infuriating now

New comment by carterschonwald in "Reducing Doom Loops with Final Token Preference Optimization"

carterschonwald — Tue, 07 Jul 2026 17:51:20 +0000

this is pretty cool. i think part of the root cause is current rlhf post training design around confidence and optics rather than cooperative transparent honesty. though its kinda an expensive hypothesis to dig into as a private individual

New comment by carterschonwald in "The Log is the Agent"

carterschonwald — Sun, 05 Jul 2026 10:44:27 +0000

This paper points at an idea, but its really only legible if you have a more developed version of the idea already. I really should write more

New comment by carterschonwald in "The bottleneck might be the air in the room"

carterschonwald — Sat, 04 Jul 2026 10:10:49 +0000

i literally had a co2 sensor for my engineering team last fall cause the space was so poorly ventilated. just measuring it continuously radically changed how everyone approached using the space packing wise and ventilation. smelled better too :p

New comment by carterschonwald in "US Supreme Court rules geofence warrants require constitutional protections"

carterschonwald — Mon, 29 Jun 2026 16:33:17 +0000

good. Of course the precise language of the ruling matters, but good.

New comment by carterschonwald in "Qualcomm to Acquire Modular"

carterschonwald — Thu, 25 Jun 2026 16:15:05 +0000

so the most notoriously patent oriented tech firm is buying this up. lol ;)

good for the founders. also explains why my resume got dropped on the floor as a desk reject :p

New comment by carterschonwald in "Prompt Injection as Role Confusion"

carterschonwald — Mon, 22 Jun 2026 20:50:43 +0000

.... i thought this was more widely known, granted i did write up a pretty wacky doc explaining way more fun experiments than these, and i have a fix that even prevents role collapse in my harness on github

New comment by carterschonwald in "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"

carterschonwald — Mon, 22 Jun 2026 18:41:37 +0000

https://github.com/cartazio/oh-punkin-pi/blob/main/scripts/b...

use that to install assuming you have whatevers needed to set up.

ive a much fancier next gen thing i hope to make available as a sass mid to late summer, but if you have any feedback or questions on mine do reach out

New comment by carterschonwald in "Big Tech is borrowing like never before"

carterschonwald — Sat, 20 Jun 2026 16:26:38 +0000

“sit with” is way over used lately

New comment by carterschonwald in "I restarted a 10 year old Xeon 174 times to delete 12 flags and gain 4 tps"

carterschonwald — Sat, 20 Jun 2026 14:59:32 +0000

“worth sitting with ” is in the way to overused scale :(

New comment by carterschonwald in "Is AI ruining our skills? Early results are in – and they're not good"

carterschonwald — Fri, 19 Jun 2026 20:31:45 +0000

whats your fave/best one shot code gen?

mine def has to be the initial impl of my python to my personal fave compiler ir. i got claude chat to write it in two sentences. 4 turns helping it rmeemver where it put shit becsuse of transcript issues

New comment by carterschonwald in "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"

carterschonwald — Fri, 19 Jun 2026 18:44:41 +0000

my heavily modified test bed for of oh my pi fixes this

New comment by carterschonwald in "DeepSeek Introduces Vision"

carterschonwald — Thu, 18 Jun 2026 15:59:03 +0000

gemini models are also fantastic at understanding non spoken sounds

New comment by carterschonwald in "GLM-5.2 is the new leading open weights model on Artificial Analysis"

carterschonwald — Wed, 17 Jun 2026 19:57:49 +0000

this helps so much. i do it too. with some of the newer frontier models its unclear if you can even turn it off in the first party chat apps. havent compared api semantics yet.

New comment by carterschonwald in "Claude Fable 5"

carterschonwald — Tue, 16 Jun 2026 02:01:36 +0000

its part of making sure the model actually engages in emotive communication, if i'm inventing insults i've never even thought about, i'm furious :)

saying i'm "furious" has lower entropy that incredibly implausible abuse. In some first party harnesses it just results in doom loops, but thats usually because the COT is hidden after the immediate turn in those setups. COT persistence helps with a lotta stuff

New comment by carterschonwald in "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"

carterschonwald — Mon, 15 Jun 2026 23:37:55 +0000

thats a harness issue not a model issue. eg i have my own reasoninf harness that forced persisted cot

New comment by carterschonwald in "Don't trust large context windows"

carterschonwald — Sun, 14 Jun 2026 13:55:41 +0000

context window size isnt quite the issue though, its that the attention mass kinda spreads out too much and everything kinda converges to a sortah global average region full of what we know to be slop! theres some really cool ways at the harness or model layer to mitigate this. just isnt really prioritized by the labs often.

New comment by carterschonwald in "AI OSS tool repo goes archived over night after raising $7.3M Seed"

carterschonwald — Sat, 13 Jun 2026 18:27:06 +0000

i had the very strange experience last week of a recruiter listing your org as an example client, and when i looked stuff up i saw the current state.

New comment by carterschonwald in "Claude Fable 5"

carterschonwald — Tue, 09 Jun 2026 19:48:23 +0000

i actually think this is too tame. it really has to be stuff youd mever say to a real person.