E-Reverance

8 ms·

91.
99% on ARC-AGI 3 (public eval, with harness)(xcancel.com)
5 points by E-Reverance 2mo ago|0 comments
92.
by E-Reverance 3mo ago
https://www.youtube.com/watch?v=LXqZMKXSPJo
93.
Audio Transport: A Generalized Portamento via Optimal Transport(arxiv.org)
2 points by E-Reverance 3mo ago|1 comments
94.
Video Generation Models Are General-Purpose Vision Learners(genception.github.io)
2 points by E-Reverance 3mo ago|0 comments
95.
Universal Learning of Nonlinear Dynamics(arxiv.org)
2 points by E-Reverance 3mo ago|0 comments
96.
Learning-to-Optimize via Deep Unfolded Flows(mit-realm.github.io)
3 points by E-Reverance 3mo ago|0 comments
97.
A Theory of Contrastive Learning with Natural Images(arxiv.org)
4 points by E-Reverance 3mo ago|0 comments
98.
Lingbot Vision(technology.robbyant.com)
2 points by E-Reverance 3mo ago|0 comments
99.
Zero-Flow Encoders(arxiv.org)
1 points by E-Reverance 3mo ago|0 comments
100.
Dispersion loss counteracts embedding condensation in small language models(chenliu-1996.github.io)
42 points by E-Reverance 3mo ago|8 comments
101.
Meshtryoshka: Differentiable Mesh Rendering for Unbounded Scenes(danielxu9393.github.io)
2 points by E-Reverance 3mo ago|0 comments
102.
Simplified Sparse Attention via Gist Tokens(arxiv.org)
4 points by E-Reverance 3mo ago|0 comments
103.
Continuity-Enhancing Degree Elevation and Splits(graphics.cs.utah.edu)
2 points by E-Reverance 3mo ago|0 comments
104.
Forward Self Models(jagilley.github.io)
2 points by E-Reverance 3mo ago|0 comments
105.
Tapered Language Models(arxiv.org)
3 points by E-Reverance 3mo ago|0 comments
106.
Everything is logarithms(alexkritchevsky.com)
316 points by E-Reverance 3mo ago|83 comments
107.
by E-Reverance 3mo ago
I don't think the issue is that they used ai behind the scenes, but there is an implicit proof of work from forcing it beyond the style you'd expect. I for one roll my eyes whenever I see that specific kind of rounded corner, fros
108.
Explaining Attention with Program Synthesis(arxiv.org)
2 points by E-Reverance 3mo ago|0 comments
109.
by E-Reverance 4mo ago
"At the same time, China is also the world's leading producer of electric cars..." Kind of interesting for a professionally branded company to use "..." like that
110.
by E-Reverance 4mo ago
I think its worth emphasizing that his argument isn't completely against generative ai, but rather its environment. Although I don't see why it would be impossible for something like an LLM to learn some sort of self-play within i
111.
by E-Reverance 4mo ago
I don't completely disagree but its worth noting how new a lot of the empirical evidence in favour of LLMs are, so its not impossible to be a tad ignorant of the present
112.
by E-Reverance 4mo ago
No > It uses 384 routed experts (top-8) with hybrid attention (full-attention + sliding-window 128 at 6:1 ratio) over 70 layers (1 dense + 69 MoE) https://recipes.vllm.ai/XiaomiMiMo/MiMo-V2.5-Pro
113.
Sparsely gated tiny linear experts(arxiv.org)
3 points by E-Reverance 4mo ago|0 comments
114.
Training-Free Single-Image Diffusion Models(haojunqiu.github.io)
6 points by E-Reverance 4mo ago|0 comments
115.
Simple input–output dependencies explain neuronal activity(nature.com)
4 points by E-Reverance 4mo ago|0 comments
116.
Self Teaching Autoencoder(the-puzzler.github.io)
2 points by E-Reverance 4mo ago|0 comments
117.
by E-Reverance 4mo ago
> P.S.: I like discussing such topics. If anyone knows a forum or discord with like-minded people, please let me know :) Unironically twitter (and only use the "Following" tab as opposed to the "For You") Make an acco
118.
by E-Reverance 4mo ago
> But I am a bit reasurred that at least my job won't be fully replaced with AI :) I honestly can't comment with certainty that training from videos alone and whatever tokenization scheme they're using will ever get perfec
119.
Distance Marching for Generative Modeling(arxiv.org)
1 points by E-Reverance 4mo ago|0 comments
120.
A Theory of Generalization in Deep Learning(arxiv.org)
4 points by E-Reverance 5mo ago|0 comments

More ›