Viewing profile — jedbrooke
jedbrooke
HN member- Joined
- Wed, Aug 07, 2024, 4:39 AM UTC
- HN karma
- 329
- Public activity
- 100 items
- HN profile
- View on Hacker News ↗
About jedbrooke
Recent public activity
-
comment
Comment #49219235
This is a nice little archive of all these different installers. I’d almost forgotten about all this madness since I’ve been using homebrew to install stuff for over a decade
-
comment
Comment #49214217
I’m surprised I haven’t seen anyone mention video models yet. I don’t know how many fps 17000 tok/s translates to exactly but it’s gotta be a lot. Might make real time AI video pos…
-
comment
Comment #49198975
I wonder, if indeed there is astroturfing going on, if instead of a foreign actor it’s competing AI companies. Eg OpenAI astroturfing opposition to when Anthropic wants to build a …
-
comment
Comment #49157411
I mean, if you had fast enough internet connection, you could just stream it over https
-
comment
Comment #49157362
nonexistent functions being referenced is pretty bad, I haven’t had that issue since like opus 4. Shows either complete lack of oversight and/or using a bargain basement model (so …
-
comment
Comment #49156798
This is one of my pet peeves too. Like when someone says “ChatGPT says: …” I just want to stop them and say “Look, you are a human with a functioning mind and thoughts and I’m sure…
-
comment
Comment #49156561
the bonsai 27B 1bit quant version of Qwen3.6 27B is even more nuts, model fits in 4GB, and with 100k of content model+kv cache fits in 8GB. I’ve been running it locally on my mac m…
-
comment
Comment #49150498
from the readme: > the filtering is optimised for framework 12's hinge stiffness. never in a million years would I expect to have to optimize software for a given hinge stiffness, …
-
comment
Comment #49150395
it would be interesting to see results from works that don't have notable existing film/tv adaptations or illustrations (though I suppose there will always be fan art) to see if it…
-
comment
Comment #49141109
UCLA is often used as a stand in for many colleges in movie sets (notably Harvard, as one of the buildings in UCLA has the Harvard crest on it specifically for this purpose) due to…
-
comment
Comment #49141083
I feel like a “AI answers can be surprisingly good, especially if you ask the right questions” can be applied to almost anything you can do with LLMs. Like the threads a few days a…
-
comment
Comment #49131462
> A personal music player for Apple Music with an embedded agent loop, so I can say things like "I'm going to work in the wood shop, play woodworking-appropriate music and keep the…
-
comment
Comment #48940393
audio CDs use a different wavelength for the laser (infra red) while PS game disks are BluRay (blue laser). Of course, 99.9% of other BluRay players will also support DVD/CD, so ye…
-
comment
Comment #48940040
OCRing handwritten dates and printed timestamps seems useful, but how is “AI” supposed to know when a photo is taken just by looking at it? At least the scan date is a real piece o…
-
comment
Comment #48913008
from what I understand prismml isn’t doing a quant like normal models where you take a model trained at fp16 and then chop off some bits to reduce vram, but rather they’re training…
-
comment
Comment #48862444
I find they still get blocked occasionally :(
-
comment
Comment #48848714
I find sonnet and haiku will do that, but I rarely get that with opus. When it does happen it’s a good cue to start a new session. That’s another benefit of the style guide is it’s…
-
comment
Comment #48842011
> Some small part of me has started to dread reading LLM output because I know what I’m going to find. False assumptions and hallucinations. Emphatic, staccato fragments. Excessive…
-
comment
Comment #48790753
already been tried: https://github.com/samrolken/nokode Maybe LLMs will eventually get to the place where it replaces even the OS kernel
-
comment
Comment #48779989
oh I had thought it was opt-in, but it wasn’t. Yeah ignoring licenses is significantly less cool
-
comment
Comment #48769346
> Supplies are limited, and the first 1,000 eligible submissions will receive one. Limit one per person. Availability may be limited by country or region. This seems quite limited …
-
comment
Comment #48695887
reminds me of the x264 “placebo” encoder setting https://trac.ffmpeg.org/wiki/Encode/H.264#FAQ
-
comment
Comment #48633920
Dolphin only supports Wii games, not Wii U. This is part of the reason the Wii U flopped is that people thought it was just a peripheral for the original Wii and not a separate new…
-
comment
Comment #48370819
looks like the macOS one is Tahoe only. I’ve been putting of upgrading to tahoe but this might be enough to tempt me
-
comment
Comment #48365426
I’d been thinking about if something like this would be possible for https://chatjimmy.ai/ . The underlying model is only llama 3 8B but I’m curious what coding harnesses would be …