diffusers-internal-dev/abot-world-interactive

1---2title: ABot World Interactive3emoji: 🕹️4colorFrom: yellow5colorTo: green6sdk: gradio7sdk_version: 6.25.08python_version: '3.12'9app_file: app.py10pinned: false11short_description: Drive a live world model with Modular Diffusers12---13 14Interactive [ABot-World](https://huggingface.co/YiYiXu/ABot-World-0-5B-LF-Diffusers) (a real-time,15action-conditioned world model, Wan2.2-TI2V-5B) running natively on Modular Diffusers. The engine is one16`pipe.stream(action_source=...)` loop over the `ABotWorldStreamingBlocks` preset17([huggingface/diffusers#14159](https://github.com/huggingface/diffusers/pull/14159)): the pipeline polls the18held keys once per generated block and yields every block's decoded frames.19 20The serving shell — `gradio.Server` start/stop endpoints, a WebSocket that streams binary JPEG frames with21steady pacing, per-session control queues that cross the ZeroGPU fork, and the front-end — is reused from22[acvlab/abot-world-interactive](https://huggingface.co/spaces/acvlab/abot-world-interactive); only the23~40-line GPU rollout body is ours.24