hedgehog

5 ms·

31.
by hedgehog 1mo ago
There's something funky with the renderer, it looks like perspective is wrong. All vibe coded?
32.
by hedgehog 1mo ago
If you've managed people, these are all familiar problems. I found you need much more than a functional specification, you also need motivation, background, related work, ideas tried, etc., because those help disambiguate the right pat
33.
by hedgehog 1mo ago
It only really makes sense for problems that are complex and require iterations that don't themselves require much review. E.g. if you want find, PoC, and patch bugs, the output can be reviewed without reading all the traces. Or if you
34.
by hedgehog 1mo ago
See my sibling comment, you can probably robo-code yourself some tooling to alleviate a lot of that in a few hours but if you want help shoot me an e-mail. I'm interested in seeing other people's workflows.
35.
by hedgehog 1mo ago
Similar to hgoel the model is managing the project's arc and writing + debugging code, but the underlying work is pretty compute intensive and all LLM output that is part of the final product is generated by local LLMs. Claude Code bui
36.
by hedgehog 1mo ago
Pretty much goal + task + dependency infrastructure to help avoid drift during long runs, especially across compaction boundaries. I have spent a lot of time doing automation with models at various strengths including some of the early open
37.
by hedgehog 1mo ago
This roughly lines up with my personal experience that in March a combination of stronger models and better tooling on my end let me start running jobs unattended 24/7 (using Anthropic sub and my own hardware). Their $8000/day per
38.
by hedgehog 1mo ago
Check sampling parameters and chat template, make sure you have adequate context window, turn reasoning effort down. It should be able to one shot a small app without intervention.
39.
by hedgehog 1mo ago
This is fantastic. It would be interesting to have Claude Code export an engineering guide for doing similar ports, including descriptions of the tooling it built to do it. I've done some reversing from binary but never with results th
40.
by hedgehog 1mo ago
Maybe after enough auditing it'll make sense.
41.
by hedgehog 1mo ago
Sounds like I have some reading to do.
42.
by hedgehog 1mo ago
Oh, I can read the output, but that Haiku agent is a good trick. Where I want something less dense I just ask for "plain language" and characterize the reading audience and that term seems to trigger very readable output.
43.
by hedgehog 1mo ago
I don't know, I just pulled up the status for an active session and here's what it said: One thing I found before dispatching, and filed as Q0579. The halt told you C6 was all that was left in the unit. That was true of the
44.
by hedgehog 1mo ago
On the one hand this kind of harness work is pretty interesting, on the other the blog post is light on details and sounds like existing behavior of any high end model in Claude Code, Pi, or one of the litany of other projects that people s
45.
by hedgehog 1mo ago
Does anyone here have experience building custom apps for the Kobo? I'm considering getting an e-ink device as a way to shift document reviews off of my regular LCD screens and away from my desk.
46.
by hedgehog 1mo ago
You can kludge it together by having the agent keep a log of what it does with any notes and lessons about friction / efficiency, and then on some interval review that + session history for items worth promoting into a topic-based memo
47.
by hedgehog 1mo ago
Yes, those engineering discussions must have been interesting. By that point people had a fair amount of experience in Earth orbit but between the relative newness of space and the relative newness of computers there must have been signific
48.
by hedgehog 1mo ago
On the topic of ferromagnetic computing, was there ever any serious investigation of Parametron type machines for at least parts of the flight control or other critical systems where interruptions would have been very serious?
49.
by hedgehog 1mo ago
I had I think a G400 and G450 in my desktop at work at different points around 2001 and my recollection is they had a reputation for high output quality (DAC and such).
50.
by hedgehog 2mo ago
It's me, it's the reams of sessions I share back with a five star rating that are just Claude Code talking to itself about debugging its own generated code in jargon that has slowly diverged from anything a human would understand.
51.
by hedgehog 2mo ago
I think most people that use LLM coding tools enough independently derive most of the stuff in the article. For example my approach to garbage collection is to sample paths within the project and chunks of file content, then assemble contex
52.
by hedgehog 2mo ago
I know a little bit about this problem space from previous work (we were working on performance-portable deep learning back around 2016). The infrastructure has improved but as far as I can tell not many teams have really "squeezed the
53.
by hedgehog 2mo ago
I just took a minute to look at your eider repo, very cool. It looks like most of the code outside the kernels and immediately surrounding plumbing would work well on AMD APUs, and probably also on Apple and newer Intel.
54.
by hedgehog 2mo ago
Ok, at 50k context its about 126 prefill, 13 generation.
55.
by hedgehog 2mo ago
It should be faster, Q6 on Ryzen 395 using Vulkan llama.cpp is about 22 tokens/s with no MTP, and I'd expect the Spark to be 20% faster or something in that neighborhood.
56.
by hedgehog 2mo ago
The PR branch does seem to work, I'm planning to move almost all of my Qwen using workload over to it tonight.
57.
by hedgehog 2mo ago
The active param count is so small I'm not sure how much advantage MTP will have, the current llama.cpp does load the ngram embeddings but I haven't verified it uses them. I expect to redeploy all this stuff every few days as the
58.
by hedgehog 2mo ago
Separating knowledge from reasoning so you only pay for what you use is a big rationale for MoE, the big problem being MoE training has historically been hard to get right. In a dense model every single token you're paying a cost to de
59.
by hedgehog 2mo ago
In my early testing it's way better both quality and speed on Strix Halo (posted recipe in sibling comment).
60.
by hedgehog 2mo ago
In initial testing on Ryzen 395 / Strix Halo it's about 22 tokens/s generation and the output quality is impressive. Better and faster than 3.8 27B, and enough better to justify moving away from 3.6 35B even though 35B is sti

More ›