A new technique called In-Distribution Forcing targets drifting in autoregressive video diffusion models, where colors and textures shift and motion dynamics decay during long-horizon video generation. Existing approaches rely primarily on KV conditioning, which selects or modifies cached keys and values, according to the research. The method is applied at test time to extend generation beyond the short horizons where modern autoregressive video diffusion models currently excel.
Modern autoregressive (AR) video diffusion models excel at short-horizon video generation, yet generating long videos remains challenging due to drifting, where colors and textures shift, and motion dynamics decay. Existing works primarily rely on KV conditioning, which selects or modifies cached ke
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.