E-Reverance

6 ms·

31.
by E-Reverance 16d ago
Are y'all using any sort of self-distillation similar to https://self-evo.github.io/ to sharpen representations?
32.
Reward Guided Speculative Decoding(arxiv.org)
1 points by E-Reverance 18d ago|0 comments
33.
Renderformer V2 (Full neural rendering)(renderformer.github.io)
3 points by E-Reverance 19d ago|0 comments
34.
DIY Gene Gun [video](youtube.com)
1 points by E-Reverance 19d ago|0 comments
35.
by E-Reverance 20d ago
Neither do LLMs https://arxiv.org/pdf/2607.03502
36.
by E-Reverance 20d ago
I definitely belong to the latter camp. After LLMs I view everything humans do very systematically and whenever said thing still feels fuzzy I just treat it as having a noise/smoothing term
37.
Why not to use the Gaussian kernel(arxiv.org)
2 points by E-Reverance 20d ago|0 comments
38.
Prefix Sliding for efficient test-time scaling(arxiv.org)
2 points by E-Reverance 22d ago|0 comments
39.
by E-Reverance 22d ago
Apologies for making the reply chain so long but I think a video like this somewhat proves how a lot of aesthetic preferences can be *ultra* sensitive to small visual details : https://youtu.be/twcMra_67-w?t=88 The video is
40.
by E-Reverance 22d ago
I strictly meant using the embeddings for training a reward model, not the generator. By good for generation I just meant the reward model might find more visual cues for aesthetic preference and avoid some of the spurious semantic correlat
41.
by E-Reverance 22d ago
> Like Dino-V3 or something of that ilk? Yes but for generation LingBot seems uniquely compelling https://technology.robbyant.com/lingbot-vision because it has a very strong spatial prior
42.
by E-Reverance 22d ago
Regarding the LAION aesthetic predictor footnote, I don't see why a modern model and nonlinear classifier won't do a good a job. Is there a fundamental technical problem with the idea?
43.
Generating Interpretable Networks Using Hypernetworks(arxiv.org)
1 points by E-Reverance 22d ago|0 comments
44.
by E-Reverance 27d ago
No one mentioning the possibility of it being StepFun?
45.
The Spectral Neuron(arxiv.org)
4 points by E-Reverance 28d ago|0 comments
46.
by E-Reverance 29d ago
I don't see it, did he delete it?
47.
Universality of Gradient Descent Neural Network Training(arxiv.org)
39 points by E-Reverance 29d ago|2 comments
48.
by E-Reverance 1mo ago
I'm confused, other people finding it useful is evidence of it being useful for them . How is that merely jumping on a bandwagon? As someone who doesn't like coding, all the one off scripts and ideations its permitted are aweso
49.
by E-Reverance 1mo ago
>botlickers > Linus and Greg K-H suddenly deciding they like it is not evidence. You don't seem particularly interested in changing your mind
50.
Mobius: Foundation Model with Decoupled Knowledge and Reasoning(arxiv.org)
2 points by E-Reverance 1mo ago|0 comments
51.
Idea for Expository AI
1 points by E-Reverance 1mo ago|0 comments
52.
by E-Reverance 1mo ago
Would it? AGI was always about the efficiency gains of versatility , not strictly all forms of efficiency
53.
by E-Reverance 1mo ago
I think the fundamental rebuttal to this is merely that, we’re still so early , the $1.5 trillion number is irrelevant for developing technologies in the same way two women can’t birth a baby in 4.5 months
54.
by E-Reverance 1mo ago
(March 2026)
55.
by E-Reverance 1mo ago
where I heard of this from : https://youtu.be/4ZB2PNUYR1Y
56.
Simplifying and Refactoring Introductory Calculus (2018)(arxiv.org)
145 points by E-Reverance 1mo ago|82 comments
57.
by E-Reverance 1mo ago
Also this part was quite funny https://youtu.be/G35ybr0PZhM?t=751
58.
3B1B Virtual Career Fair(3blue1brown.com)
2 points by E-Reverance 1mo ago|1 comments
59.
by E-Reverance 1mo ago
I get what you generally mean but it’s worth emphasizing that a coding agent probably helped with setting this up
60.
Geometric Fluid Mechanics (SGP'26 Course)(youtube.com)
1 points by E-Reverance 1mo ago|0 comments

More ›