SamInTheShell

6 ms·

1.
by SamInTheShell 4d ago
Just going to defer to an IBM thing here: https://www.youtube.com/watch?v=l-QPwk_f4eE And just say: LLMs only amplify the the knowledge you have, even the best models o use like Fable still suffer from promoting false narra
2.
by SamInTheShell 7d ago
If this is anyone’s experience with Claude for a button today, I feel like Jobs would say you’re holding it wrong.
3.
by SamInTheShell 7d ago
We already have practices to deal with AI. It’s called user space. Properly air gap the AI, stop creating routes to open internet, don’t run network wires into the faraday cage. Why the heck someone would risk having an open bridge to the s
4.
by SamInTheShell 12d ago
@q4 is definitely smarter than sonnet from what I’ve seen so far. It’s even caught problems in code made by fable, when using it as a code reviewer.
5.
by SamInTheShell 12d ago
Read the license again.
6.
by SamInTheShell 13d ago
Just wanted to share, I had a SaaS AI drop me a script to bench ninfer against llama-cpp and it is impressive. The place that it's doing better at than llama-cpp seems to really be late in the context window. Initial results boiled dow
7.
by SamInTheShell 16d ago
Thanks for sharing, I'm definitely trying it out after I get through my project milestones for the 5090. The README claims 700tok/s for Qwen 3.8 27b, that would be amazing, I'm only expecting an increase from my ~50tok/s
8.
by SamInTheShell 16d ago
I should have mentioned in my prior comment, I didn't have any issues in Pi either.
9.
by SamInTheShell 16d ago
My use-case is only coding, every model sucks at writing good literature and there is no way around that (have had people try to debate me on this, but it's a taste thing, I have extensive English writing skills from my school years).
10.
by SamInTheShell 16d ago
From what I’ve been seeing, the Mac studios do look like they have potential. I was looking to drop $10k-$15k on one until recently. After comparing a Radeon 7900 XTX vs Ryzen Halos 128GB vs M1 MacBook Pro 64Gb, I landed on just getting an
11.
by SamInTheShell 17d ago
Poor software choice for usecase. `sudo pacman -S podman` didn't come with these problems out of the box and assumed rootless by default.
12.
by SamInTheShell 18d ago
Junk. Just another company that doesn't understand why I pick up a phone or use a support chat or send an email. Before I even started in corporate, I stopped checking email. Now when I call in to get support for things, everytime I ge
13.
by SamInTheShell 26d ago
This is pretty good to know today. I’m currently using cosine similarity in some of my code that could have just been dot products.
14.
by SamInTheShell 1mo ago
Just use a harness that discards all except the most recent thoughts. This model is probably the first real small model that does well on long horizon tasks.
15.
by SamInTheShell 1mo ago
I did. Distributed too and supports erasure coding. The reason is because nothing quite had the architecture or features I wanted in a deployment pattern that matched my use case. All major databases that are validated are complicated to op
16.
by SamInTheShell 1mo ago
It's worth running, even quantized. I liked Meta Muse Glimmer's outputs, but qwen3.8-27b@q4_k_s kinda seems way better. Haiku/Sonnet kinda pairing in workflows?
17.
by SamInTheShell 1mo ago
The proxy caches and package solution for Go has had its own drama. My big contention with this project is that if I wanted to use node (or npm; or vite) in any capacity, I would have chosen node for that. I don’t want to use those tools in
18.
by SamInTheShell 2mo ago
I'm kinda in the camp of wanting nothing to do with node if I'm building in Go. We have our own stdlibs for serving.
19.
by SamInTheShell 2mo ago
You go to person's profile. View their submissions. Use find via CTRL+F to seek out their "Show HN" submissions (or in this case, it's the top of the list because it's my last submission https://news.ycom
20.
by SamInTheShell 2mo ago
Done. You have 2/3rds of Discord or Slack solved as just an insignificant fraction of my Private Cloud Project I just shared. Not interested in video or voice right now, it's easy to solve though. Also you'd have to solve for
21.
Show HN: My Fable 5 Project Was a Multiraft Database and Blob Store(meetthecluster.github.io)
1 points by SamInTheShell 2mo ago|0 comments
22.
by SamInTheShell 2mo ago
Sure, I'll make a Show HN post this weekend.
23.
by SamInTheShell 2mo ago
At least code isn't a moat anymore. Have a weekend and want your own Discord? Doable.
24.
by SamInTheShell 2mo ago
How much does it actually matter? I use to bother with these settings, but never really seems to fix the problems I commonly see. The issues I commonly see are small models getting stuck in a repeating loop, or they just fail at follow up c
25.
by SamInTheShell 2mo ago
Build a distributed system using raft groups. You will see Opus fail.
26.
by SamInTheShell 3mo ago
Opus can make harnesses fine. I barely have to instruct it.
27.
by SamInTheShell 3mo ago
This is probably the first small model I got through some simple web game tests without having to reset the context. It tends to opt to overwrite an entire file instead of doing edits... which editing is where most of these small models fal
28.
by SamInTheShell 3mo ago
Tool calling at Q8 for this model is kinda impressive. It's doing patch edits like a champ. It gets stuck looping and doing the wrong things though as the context window grows.
29.
by SamInTheShell 3mo ago
Related/Dupe: https://news.ycombinator.com/item?id=48494697 Also, I had Claude Opus read through some of the mailing list threads and it made this script to check a few things: https://gist.github.com/S
30.
by SamInTheShell 3mo ago
Lazy check script by Claude Opus: https://gist.github.com/SamInTheShell/c3609743d742f002f1e8c7... It seems my stubborn resistance to using yay saved my bacon here. Good luck everybody else.

More ›