E-Reverance

4 ms·

31.
by E-Reverance 18d ago
whoops lol
32.
Astra replicates Adobe Lightroom(xcancel.com)
3 points by E-Reverance 18d ago|12 comments
33.
The Foundations of Modern AI: Generalization, Data Selection, and Epiplexity(youtube.com)
2 points by E-Reverance 19d ago|0 comments
34.
by E-Reverance 21d ago
Smart fridges are a pathological virus, yes /jk
35.
Diffusion as a Training Curriculum for Timestep-Free Iterative Reasoning(arxiv.org)
3 points by E-Reverance 22d ago|0 comments
36.
Unlocking Lossless Speedups in LLMs via Discrete Diffusion (5000 Tk/S)(s-sahoo.com)
3 points by E-Reverance 23d ago|0 comments
37.
Matching Features, Not Tokens: Energy-Based Fine-Tuning of Language Models(energy-based-fine-tuning.github.io)
1 points by E-Reverance 23d ago|0 comments
38.
by E-Reverance 23d ago
At this point the primary axes for improvement seem to only/mostly be speed and personalized reward models. We seemingly have the general of notion "learning" and "intelligence" functionally complete
39.
Language Models Can Control Their Own Attention(arxiv.org)
2 points by E-Reverance 24d ago|0 comments
40.
Inverse Rendering for Modeling with Line Primitives (Good for Fur)(kenji-tojo.github.io)
1 points by E-Reverance 24d ago|0 comments
41.
by E-Reverance 25d ago
Are y'all using any sort of self-distillation similar to https://self-evo.github.io/ to sharpen representations?
42.
Reward Guided Speculative Decoding(arxiv.org)
1 points by E-Reverance 27d ago|0 comments
43.
Renderformer V2 (Full neural rendering)(renderformer.github.io)
3 points by E-Reverance 28d ago|0 comments
44.
DIY Gene Gun [video](youtube.com)
1 points by E-Reverance 28d ago|0 comments
45.
by E-Reverance 29d ago
Neither do LLMs https://arxiv.org/pdf/2607.03502
46.
by E-Reverance 29d 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
47.
Why not to use the Gaussian kernel(arxiv.org)
2 points by E-Reverance 29d ago|0 comments
48.
Prefix Sliding for efficient test-time scaling(arxiv.org)
2 points by E-Reverance 1mo ago|0 comments
49.
by E-Reverance 1mo 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
50.
by E-Reverance 1mo 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
51.
by E-Reverance 1mo 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
52.
by E-Reverance 1mo 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?
53.
Generating Interpretable Networks Using Hypernetworks(arxiv.org)
1 points by E-Reverance 1mo ago|0 comments
54.
by E-Reverance 1mo ago
No one mentioning the possibility of it being StepFun?
55.
The Spectral Neuron(arxiv.org)
4 points by E-Reverance 1mo ago|0 comments
56.
by E-Reverance 1mo ago
I don't see it, did he delete it?
57.
Universality of Gradient Descent Neural Network Training(arxiv.org)
39 points by E-Reverance 1mo ago|2 comments
58.
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
59.
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
60.
Mobius: Foundation Model with Decoupled Knowledge and Reasoning(arxiv.org)
2 points by E-Reverance 1mo ago|0 comments

More ›