OWL
@wayfarerlabs
Followers
395
Following
18
Media
8
Statuses
56
We're hiring. https://t.co/94dmsJ50SU
Providence, RI
Joined May 2025
Turbocharge speed & quality gains in Diffusion World Models! 🚨.︀︀.︀︀- 8x8 AE w/ depth latents → 4x fewer tokens, 4x FPS boost.︀︀- 4x4 flow+depth AE in progress → next-level consistency.︀︀- DMD distillation: 16→2 steps = 8x faster sampling.︀︀- Custom RoPE fix → 20x
3
9
59
We're skipping the blog post this week since all of our team is at ICML. We're excited to share what we've been working on next week :).
0
0
4
RT @lcastricato: Anyone at ICML wanna see the future of world models? We're walking around with a laptop running our world model at 500 FPS….
0
5
0
We're hiring! . Join our discord:
discord.com
Check out the OWL community on Discord - hang out with 375 other members and enjoy free voice and text chat.
0
0
1
Original paper authors: @zhang_yunzhi Carson Murtuza-Lanier @zizhang_li @du_yilun @jiajunwu_cs. Paper:
arxiv.org
Modern neural models capture rich priors and have complementary knowledge over shared data domains, e.g., images and videos. Integrating diverse knowledge from multiple sources -- including visual...
1
0
1
🌟Got multiple expert models and want them to steer your image/video generation? . We’ve re-implemented the Product of Experts for Visual Generation paper on a toy example, and broken it down step by step in our new blog post! . Includes: .- Github repo: Annealed Importance
2
8
18
RT @lcastricato: Gonna be at ICML Monday if anyone is in town lmk.
lu.ma
Wanna learn more about open science diffusion world models? Come to our lunch @ ICML, meet our research scientists, and learn how you can contribute to our…
0
1
0
RT @lcastricato: What!? A Tuesday release from Wayfarer Labs?!. Heresy! Good thing the post is so good :).
0
1
0
🧵3/3. We're hiring: Link to discord:
discord.com
Check out the OWL community on Discord - hang out with 375 other members and enjoy free voice and text chat.
0
0
3
RT @lcastricato: Another week well spent. No blog post today, we're all too busy making the future of local world models.
0
2
0
We're delaying our release this week a few days because most of our team is busy at an offsite today :) excited to share what we're working on.
0
0
3
Take a look at Generation vs. Reconstruction: Striking A Balance 2/3.
1
0
2
Last week we tried pushing visual fidelity to the limit, with just 4x4 latents. Key notes:.🔹 Decoupled reconstruction and generation.🔹 ControlNet photorealistic supervision.🔹 Autoencoder + GAN.🔹 100+ fps at 360p 🚨.See below our technical blog 1/3 🧵.
2
2
17
FYI, we're rebranding this week. Stay tuned to see our new coat of paint :).
1
0
8