xscott

7 ms·

1.
by xscott 6d ago
I had Muse Glimmer (from Meta / Facebook) quoting OpenAI's safety guidelines to me, and I had Poolside's Laguna (a smaller US company) with thinking traces about obeying Chinese law. Both of those are local models, and I didn
2.
by xscott 8d ago
> A common issue is that it's rarely mentioned on which dataset KL-divergence is computed. It seems the most common dataset is wikitext Thank you for calling this out. Using Wikipedia snippets for these is a terrible choice. I did
3.
by xscott 19d ago
Maybe I'm reading too much between the lines, but I suspect the reason is to rub his nose in the duplicity or naivety depending on how generous you're feeling. Publishing the model would be a confession that he was wrong. AI poli
4.
by xscott 21d ago
There's a link you can click to see the personal background of the respondents. It's tough to know what "researcher in a field not listed" means, but it's possible that over 60% are not even physicists: 30.8%
5.
by xscott 25d ago
This seems very cool, but I'm not sure I understand exactly what it's doing. Are they making a new speculative drafter for Qwen 3.8 27B? Maybe they're optimizing the MLX code for the decoder itself? Thank you in advance.
6.
by xscott 27d ago
Your definition is not useful enough. There are people missing any or all of those, and any or all of those can be approximated by a machine.
7.
by xscott 27d ago
I've thought about turning this upside down. To any person who is sure they know what has and doesn't have consciousness: If I say I don't have it, can you prove me wrong? As far as I'm concerned, it's a word with
8.
by xscott 27d ago
I'm not claiming to have any expertise in this area, but I've got a list of things I try to apply when working with LLMs. Possibly relevant here is, "don't tell the model what NOT to do, show it what TO do". I thi
9.
by xscott 28d ago
People over-quantize things, muck with the temperature and other settings based on superstitions or results from models they think are similar. There's lots of ways to make 3.8 27B dumber.
10.
by xscott 29d ago
Not that my opinion matters much, but I like Deno. I never tried Bun.
11.
by xscott 29d ago
Same here - my questions were sophomore level. I think it's notable that when I edited my question to say it was about Gemma 4, it answered without blocking. A cynic like myself would interpret that as evidence they don't care a
12.
by xscott 29d ago
What's the distinction between "protecting their turf" and "competitive reasons"? I see them as the same, but I could be missing something. That's an interesting thought on the current "safety blocking&qu
13.
by xscott 1mo ago
I've gotten flagged for asking questions about tokens and tensors. That makes me believe it's not about safety, it's about protecting their turf. I cancelled my subscription - same fear about getting flagged too much leadin
14.
by xscott 1mo ago
All good, but that seems unrelated to what I said, and I'm not sure why you replied to me. Consider this though: Regulating AI models in the US benefits the data centers, not individuals. That's where regulated models run. Regu
15.
by xscott 1mo ago
I think you might misunderstand. The regulation isn't about what OpenAI and Anthropic can do. It's about what you, a citizen, can do.
16.
by xscott 1mo ago
Yeah, I think I'm seeing the same thing. I don't have all the answers, I just think it'd be a mistake to throw the baby out with the bath water on this model. It seems significantly better than the other dense models near t
17.
by xscott 1mo ago
That's definitely not an argument I was making.
18.
by xscott 1mo ago
Any argument about regulation in the US which doesn't mention that China and other countries aren't bound by that regulation should be heavily questioned. Exactly who are you stopping from doing what you don't like?!? Law ab
19.
by xscott 1mo ago
I don't really understand the argument you're making, but just to add a data point: DeepSeek V4 Flash 0731 is 167 gigabytes from the developer and as a GGUF with no additional quantization. It limps along on my 192GB M2 Mac from
20.
by xscott 1mo ago
It won't satisfy the people who just want to drop a model into their existing toolset and run, but I think there are a lot of ways to deal with this overthinking problem. For instance, it's a step backward, but I put {"reason
21.
by xscott 1mo ago
> [...] we evaluated the behavior of various Claude models in a setting with contradictory objectives. > We consistently saw a multiagent turf war... In fact, they sabotaged others with increasingly aggressive, self-replicating malwar
22.
by xscott 1mo ago
One possibility is that you have Claude write something that would not get you in trouble and concatenate that with something you wrote by hand that would. Your signature from the safe material is now associated with your content on the dan
23.
by xscott 1mo ago
2^N I think, but who's counting.
24.
by xscott 1mo ago
So many possibilities for how you could glue it all together. However, when I send Gemma 4 12B in llama.cpp an image with no accompanying text, it assumes I want a description and gives me one. I just tried with an audio file, and it trans
25.
by xscott 1mo ago
So much potential for that channel. He's got a nice range of tests and a no nonsense presentation style. However, watching tests of heavily quantized models that weren't designed for it (non-QAT) is frustrating. There's no
26.
by xscott 1mo ago
You're very right about KL divergence. I spent a couple days playing with the Gemma 4 models. That's 10 separate models (varying weights, MoE, QAT or not, etc...) with identical tokenizers. I treated 31B at BF16 as the gold sta
27.
by xscott 1mo ago
Probably not what you're after, but I've considered having a separate small mm-model act as a seeing-eye dog for the bigger more capable one.
28.
by xscott 1mo ago
You're going to punish Bolivia, Uzbekistan, Vietnam, Sri Lanka, and Congo?!? I don't think that will help: https://www.worldometers.info/co2-emissions/co2-emissions-by... Maybe you'll suggest changing t
29.
by xscott 1mo ago
How much more than 6 percent? Support your claim with reputable sources. If your arguments aren't believable, I'm buying extra steak this evening.
30.
by xscott 1mo ago
Actually focusing on CO2, same site: https://ourworldindata.org/ghg-emissions-by-sector More than 70% of CO2 from energy, less than 6% from ALL livestock.

More ›