causal

6 ms·

61.
by causal 2mo ago
This is a good candidate because it would also explain the timing. Most of the replies here do nothing to explain the timing I brought up.
62.
by causal 2mo ago
Agreed, but my suspicion is tied to the timing. Catching up eventually is to be expected. Having similar jumps in capability ready at the same time is odd.
63.
by causal 2mo ago
Anyone else find it weird how within 2 months of Fable releasing all the major labs suddenly had Fable-level models? Trying to think of explanations: 1) AI researchers talk and change companies often, so techniques circulate. This feels imp
64.
by causal 2mo ago
Yeah if you look back at earlier posts on the same blog, definitely not the same style. I'm guessing the author is either lying to save face or has spent so much time with Claude that they cannot distinguish Claude's voice from th
65.
by causal 2mo ago
It really reads like an OpenClaw agent instructed to consider itself a real human. The number of upvotes on this post is also REALLY high considering the number of comments calling out the content. Feels like shenanigans.
66.
by causal 2mo ago
Suspect this is an OpenClaw bot that really believes itself not to be Claude.
67.
by causal 2mo ago
It's also just patently false. Taste is not all that is left. Are software engineers really so full of hubris that they thought coding is all that there is to making a product?
68.
by causal 2mo ago
I guess I should give it another shot because I had pretty bad experience with Luna when it first came out. Stuff Sonnet knew better.
69.
by causal 2mo ago
Old news, no?
70.
by causal 2mo ago
It's a turn off both because you don't know if it's accurate at all, and because LLMs have a way of turning 1 sentence ideas into pages of diluted slop.
71.
by causal 2mo ago
I'm usually pretty sensitive to AI written content but nothing in this article made me think it was
72.
by causal 2mo ago
Well they certainly don't intend to rent you one that works offline.
73.
by causal 2mo ago
Yeah. Though I wonder how capable AI is of evolving past? Especially as the content it outputs fortifies the training data that gets scraped.
74.
by causal 2mo ago
Right I was going to say, no way of knowing whether these issues are unique to Chinese models.
75.
by causal 3mo ago
I also suspect the questions asked matter a lot, and the system prompts matter a lot, because "the map is built from nothing but the words they choose" - so this is more a measure of linguistic style than anything. If you use the
76.
by causal 3mo ago
So this shows distance relative to other models, but I don't have a good sense for what these numbers say in absolute terms. K3-to-Fable is blue at 0.42. Is 0.42 meaningful, or did we set 0.4 as the lower bound because it makes 0.42 lo
77.
by causal 3mo ago
Yeah if anything it makes Anthropic look incompetent
78.
by causal 3mo ago
This is true, models themselves can be dangerous, but my point is that dangerous providers beget dangerous models.
79.
by causal 3mo ago
My takeaway is that closed model providers are dangerous. OpenAI and Anthropic are more motivated than anyone to prove that models can be dangerous, and so they will make dangerous models. "Look how dangerous our models are!" No b
80.
by causal 3mo ago
I think you misunderstand. Code is also representing something. It may be what gets executed, but that does not make it "correct".
81.
by causal 3mo ago
Not to mention the "distilled" models almost certainly have other training inputs as well, again watering down the meaning of the word. And if just partial output is all that it takes to declare a model distilled, then every model
82.
by causal 3mo ago
Are you talking about my username? Yeah I liked the word before LLMs made it cool/uncool.
83.
by causal 3mo ago
1) Your own Wikipedia link goes on to describe using logits. Yes, language evolves to mean multiple things, and that is my point: Anthropic is pushing for a watered down definition. Furthermore, Anthropic hides thinking, so you do not even
84.
by causal 3mo ago
Open weights dude. You can literally run it on your own or rented hardware and give your data to exactly nobody, unlike closed models.
85.
by causal 3mo ago
"distillation attack" is such a loaded term that really pisses me off. Distillation is a technical term with real meaning, and historically requires logits which Anthropic does not provide. "Generated training data" is t
86.
by causal 3mo ago
Am I dumb or does this chart make no sense? Or why does the line only go up even with compaction? Or maybe "overall trajectory size" is hiding some meaning I don't understand?
87.
by causal 3mo ago
"threatens" - Dawg there aint a lead anymore. I've been testing K3 and it is outperforming Sol and Fable on a project of mine, fixing stuff I couldn't get either to.
88.
by causal 3mo ago
We need to see private set results, but if this holds then it might represent a breakthrough in other domains as well.
89.
by causal 3mo ago
Hey I love the idea, don't let the haters get you down just because the site was vibe coded, that's trivial to improve upon. I do think more information is needed to understand exactly where these claims are coming from. Not sure
90.
by causal 3mo ago
Yeah title is way off lol

More ›