Owen L
@diogenesdude.bsky.social
jax posting
https://lockwo.github.io/
https://lockwo.github.io/
That’s actually believable lol, imagine a ML specific LLM then with some RAG for the papers and they just share a collective interface over slack/discord to the LLM and authors and that’s the conference
January 29, 2026 at 6:14 PM
That’s actually believable lol, imagine a ML specific LLM then with some RAG for the papers and they just share a collective interface over slack/discord to the LLM and authors and that’s the conference
That does seem the be the growing consensus www.iatp.org/climates-eme...
Climate's Emergency Brake: The imperative for a livestock transition to slow global warming
In this brief, we outline why action on methane is so urgently needed and where there are risks of a sudden shift in the policy or consumer landscape. We need to plan and fund the transition now — the...
www.iatp.org
January 29, 2026 at 6:09 PM
That does seem the be the growing consensus www.iatp.org/climates-eme...
I choose to believe he lived for 300 years and just published all of them in the last 83.
January 27, 2026 at 11:38 PM
I choose to believe he lived for 300 years and just published all of them in the last 83.
The secret third option of humans changing their writing patterns to sound like an LLM 👀 arxiv.org/abs/2409.017...
Empirical evidence of Large Language Model's influence on human spoken communication
From the invention of writing and the printing press, to television and social media, human history is punctuated by major innovations in communication technology, which fundamentally altered how idea...
arxiv.org
January 27, 2026 at 11:07 PM
The secret third option of humans changing their writing patterns to sound like an LLM 👀 arxiv.org/abs/2409.017...
@joss-openjournals.bsky.social (forgot it also had an account)
January 22, 2026 at 5:30 AM
@joss-openjournals.bsky.social (forgot it also had an account)
In fact JOSS (joss.theoj.org ) has a tool that automatically does this for every submission (ie github.com/openjournals...). It’s not perfect, but it’s very useful.
January 22, 2026 at 5:28 AM
In fact JOSS (joss.theoj.org ) has a tool that automatically does this for every submission (ie github.com/openjournals...). It’s not perfect, but it’s very useful.
Probably depends a lot of on the variance in quality that the generation distribution has, but I bet if you use a single image from each class it could be a decent estimate
January 8, 2026 at 4:54 PM
Probably depends a lot of on the variance in quality that the generation distribution has, but I bet if you use a single image from each class it could be a decent estimate
I've always wondered about the "variational" part of the VQ-VAE. Like it kind of just seems like an autoencoder with some codebook regularization (perhaps why it works well for latent diffusion, since continuous latent VAEs use tiny betas for KL during training).
January 7, 2026 at 6:26 AM
I've always wondered about the "variational" part of the VQ-VAE. Like it kind of just seems like an autoencoder with some codebook regularization (perhaps why it works well for latent diffusion, since continuous latent VAEs use tiny betas for KL during training).