As video diffusion models scale, sequence lengths get uncomfortably large. A practical answer is context parallelism, and within that family, Ulysses is the canonical approach. The core idea is simple: full sequence, sharded heads.
Need help?
Contact usAs video diffusion models scale, sequence lengths get uncomfortably large. A practical answer is context parallelism, and within that family, Ulysses is the canonical approach. The core idea is simple: full sequence, sharded heads.