sosuke

Joined 19 years ago

Karma
4,686
Submissions
1,438
Member since
Oct 5
[ my public key: https://keybase.io/barretts; my proof: https://keybase.io/barretts/sigs/dq8TQPscgQmc-a5-6cUgmmGCqM-nV4-HcS7J2EC_iN8 ]

[email protected]

"We software developers, don't do things that are easy. We do things because we thought they would be easy." - akerro (https://news.ycombinator.com/item?id=12479640)

https://sosuke.com

View on Hacker News

Recent submissions

Comment · 20 days ago
In a spirit of trying to do what my imagination wants to be possible I set off to use or recreate Strata but to load Poolside Laguna S 2.1 Q3 on my system.

My machine has an RTX 4060 Ti 16 GB on PCI 3.0, Threadripper 1950X with AVX2, and 128 GB quad-channel DDR4-2133

I stopped after hitting a successful 84-minute autonomous run in OMP. Just 4 bugs in the result found by a frontier-ish model. Hit 770.83 tokens/sec on prefill and 15.85 tokens/sec at 8k context and 11 tokens/sec over the full run at 128k context.

The link is to the main findings and attempts at improving performance. In the end the Laguna architecture didn't lend itself to the same kind of dissection as Qwen3.8 and I've taken it as far as I'm willing to go. It's all dropped into the performance and research folders.

I love competition in the open weight world and Strata has been such a boon to my own local development I was greedy for more! Hope this helps someone, or you find it interesting at least.

Comment · 23 days ago
Funny the sandbox is a real security benefit but it just turned into an extra confirmation for me. I'd rather have it than not.
Comment · 27 days ago
Now that I'm at ~$350 a year on Backblaze I need to look at this again. Or the LTO-6 tapes.
Comment · 1 month ago
I like the fast releases. 6.1 coming so fast after 6.0 means they found some improvement solid enough for a new rollout.

The only time I remember the newest model making an obvious regression was when the first rolled out MoE. Super speed update but each request had less intelligence at hand. We’re way past that now

Comment · 3 months ago
I had no idea Thinking Machine was a brand! I just thought they were "thinking machine super computers" another way of saying "artificial intelligence super computers" or "machine learning" (dunno if ML was around then :shrug:)
Comment · 4 months ago
Just an FYI there are many efforts already available to clean and or provide a clean Firefox.
Comment · 5 months ago
I leaned into the narrative style for this post. The content comes from my concrete talks and work but I don't mention any specifics. Not sure if I like the result but it was still interesting. The topic was light and easy enough I don't feel it suffered for it.
Comment · 5 months ago
I'll bite. I have always been told copyright is inherit. Does it cost money to file a copyright? Do I need to do it for each blog post? For each gist? I'll totally setup some scripts to make it happen if it what actually needs doing to have the copyright I expected.

Edit: remember not to down vote ideas you disagree with. I think it was only down vote things that lower the discourse

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton