HNHacker News
TopNewBestAskShowJobs

mdgld

36 karma · joined June 10, 2026

Future child psychiatrist, amateur data scientist. Design and photos on the side.

Contact for inquiries: mdgld@berkeley.edu

submissionscomments
mdgld··on Opus 5 is currently #1 on Artificial Analysis Intelligence Leaderboard
I would love for Gemini to be competitive but even 3.6 flash doesn’t match sol, or opus 5, or k3
mdgld··on Interview with Mitchell Hashimoto about Ghostty and Zig
What features from iterm do you miss? I’ve found ghostty to be more performant and aesthetically pleasing and haven’t missed anything
mdgld··on GPT-5.6
Same, I think you both have great points. Idk how you can debug effectively (the model itself) without reasoning traces
mdgld··on GPT-5.6
Why waste time say lot word when few word do trick?
mdgld··on GPT-5.6
I always expose reasoning traces. How else can you seriously debug?
mdgld··on Opinionated and easy Pi.dev configuration
That’s why Mario said “YOLO” is enabled by default from what I remember—something like, I want you to think about what you’re doing.
mdgld··on Opinionated and Easy Pi.dev Configuration
That’s why I like things like oh-my-pi and lazypi, nothing’s stopping you from modifying it after you install; I find that when I’m working on my own sometimes I brick myself into a corner and it’s nice to see someone else’s starting point for reference.
mdgld··on Opinionated and easy Pi.dev configuration
I’ve tried quite a few, from oh-my-openagent’s gargantuan hooks+plugin system to minimal add-i like wozcode, and I’ve been most impressed by that (wozcode) and oh-my-pi. I get that pi is supposed to be roll-your-own but I think OMP’s config is a great starting point; pi removed all the fluff and OMP added just enough orchestration and tooling back in to take things to the next level. Wozcode is great if you like CC because it doesn’t mess with the fundamentals, it just adds things like batching for efficiency—CC’s system prompt alone is absurd.

Caveat: I’m a data scientist/researcher, not a professional SWE, so take my experiences with a grain of salt. I’m just a python monkey most of the time

mdgld··on Hy3
It depends on model size I think, but yeah, from my understanding at ~30B and below Q6 or even Q4 will get you 95%+ of the way there
mdgld··on Decoding the obfuscated bash script on a Uniqlo t-shirt
I wasn’t sure if you meant a Go solver or Go the language. Would be fun if someone wrote a Go program in Go
mdgld··on GPT-5.6 Sol Ultra will be in Codex
since when did HN have downvotes?
mdgld··on Google Books (or similar) all book scans – $200k bounty (2025)
Please say more about accessing HN from Iran, I’m very curious.
mdgld··on GPT-5.5 Codex reasoning-token clustering may be leading to degraded performance
Cool resource and perfect way to track this, thanks for sharing
mdgld··on GPT-5.5 Codex reasoning-token clustering may be leading to degraded performance
I’ve always loved em dashes…very sad they’re a hallmark of LLM slop now. That and trios in arguments. Maybe I’m part LLM?
← PreviousPage 2 of 2