Field Notes by Linum
Research, releases, and real-world lessons building generative video.
Introducing Linum v2
2B parameter, Apache 2.0 licensed text-to-video models (360p, 720p)
Read more →Latest
From our partners · Jun 22, 2026
How Linum built a frontier video model on Crusoe ↗︎
Pyramid-JiT: Low-Res Drafts, High-Res Images
Accelerating convergence in text-to-image with multiresolution prediction
20 min readRoad to Linum v3· Issue 03
Training Text-to-Image Models 3.6× Faster
Beating latent diffusion models with pixel-space encoder-decoders
40 min readRoad to Linum v3· Issue 02
Data Filtering for Generative Video Pre-training
Getting video models to learn better, faster
22 min readRoad to Linum v3· Issue 01
Building a $8/Month GPU-Cluster Monitor
Maintaining our sanity, training a foundation model over the holidays
6 min read
Better Reconstruction ≠ Better Generation
Learnings from 4 months of Image-Video VAE experiments
20 min read
Get Field Notes in your inbox