NVIDIA's WorldTrace: Overcoming Memory Limits in Video World Models

nvidia · hf · 2026-08-10

NVIDIA's new research addresses visual persistence in interactive video world models. The study reveals that when rollouts extend beyond the training horizon, models struggle to retrieve visual information from the KV cache due to temporal Rotary Positional Embedding (RoPE) offsets, and naive compression corrupts memory.

To solve this, they propose WorldTrace, a training-free memory framework. It keeps compressed memory addressable by assigning distinct virtual positions to summary slots, featuring two approaches:

They also introduced LoopBench, a benchmark evaluating a compressed cache's ability to reconstruct visited scenes after long detours. WorldTrace improved temporal consistency by 15.5% and episodic recall by 19.5%.

Original post →

More from Research

Research channel →