A public study garden on diffusion & video generation, agentic systems, and the systems that serve them.https://latents-and-logs.vercel.app/https://latents-and-logs.vercel.app/posts/uno-one-change-to-flux-and-a-data-flywheel/https://latents-and-logs.vercel.app/posts/uno-one-change-to-flux-and-a-data-flywheel/Reading notes on UNO: how a minimal positional change, synthetic training pairs, and a less-to-more curriculum turn FLUX into a multi-subject customization model.Sun, 26 Jul 2026 00:00:00 GMThttps://latents-and-logs.vercel.app/posts/dit-control-is-decisions-about-tokens/https://latents-and-logs.vercel.app/posts/dit-control-is-decisions-about-tokens/A working skeleton: what a diffusion transformer actually is, why every control method decomposes into a handful of token-level decisions, and a starter table mapping papers onto those axes.Sat, 25 Jul 2026 00:00:00 GMThttps://latents-and-logs.vercel.app/posts/harness-not-agent-reading-notes/https://latents-and-logs.vercel.app/posts/harness-not-agent-reading-notes/What a harness actually is, how I would pick harness components for my own research loop, and the failure patterns I recognize from my own training runs.Sat, 25 Jul 2026 00:00:00 GMThttps://latents-and-logs.vercel.app/posts/a-garden-not-a-blog/https://latents-and-logs.vercel.app/posts/a-garden-not-a-blog/Why this site publishes unfinished notes on purpose — the three tracks, the status system, and the ADHD-proof pipeline behind it.Fri, 24 Jul 2026 00:00:00 GMThttps://latents-and-logs.vercel.app/posts/ominicontrol2-reading-notes/https://latents-and-logs.vercel.app/posts/ominicontrol2-reading-notes/Reading notes on efficient conditioning for diffusion transformers — where the multi-condition cost comes from, how the paper measures efficiency, and the gap it leaves in control fidelity evaluation.Sat, 20 Jun 2026 00:00:00 GMT