Search papers, labs, and topics across Lattice.
ABot-World-0 introduces an innovative action-conditioned video world model that enables real-time, long-horizon interactions on a single desktop GPU by leveraging a diverse multi-source data infrastructure. The system employs a novel training approach, including teacher forcing and LongForcing, to effectively align a causal student model with a long-horizon teacher, thereby addressing challenges like distribution shift and autoregressive drift. Experimental results show that ABot-World-0 achieves competitive performance in controllability and coherent world evolution, streaming high-quality video at 16 FPS with minimal latency and memory usage.
Achieving real-time, long-horizon interactive world rollouts on a single desktop GPU could revolutionize how we develop and deploy AI-driven simulations and games.
We present ABot-World-0, an action-conditioned video world model for real-time, long-horizon closed-loop interaction, supported by a multi-source data infrastructure spanning AAA games, simulation engines, and internet videos to learn controllable world dynamics. WorldExplorer performs agent-driven collection guided by training feedback, while a unified pipeline applies 14 deterministic quality checks, VLM-based assessment, and synchronized action and text annotation. We progressively distill a bidirectional action-conditioned teacher into a causal student through teacher forcing and ODE distillation, and introduce LongForcing to align long student self-rollouts with an extended-horizon teacher, mitigating accumulated distribution shift and autoregressive drift. Raw keyboard actions provide a unified control interface for scene roaming and third-person character interaction, while reference-character memory provides persistent appearance cues for identity consistency during third-person rollouts. For deployment, we co-design a streaming inference stack with a lightweight VAE decoder, efficient attention, memory-aware scheduling, and low-bit DiT inference. Across optimized low-bit configurations, ABot-World-0 streams 720P video at up to 16 FPS on a single NVIDIA RTX 5090 desktop GPU, with 1.2s action-to-first-frame latency and approximately 19GiB peak VRAM. Experiments on WorldRoamBench and extended interactive rollouts demonstrate competitive controllability and coherent long-horizon world evolution.
