Chain-of-Frames reasoning
A reasoning approach that involves analyzing a sequence of frames in a video to derive conclusions.
Category: Concepts
pattern-based inference
The process of identifying patterns in data to make predictions or draw conclusions.
physical dynamics
The study of how physical systems evolve over time, including motion and interaction of objects.
spatial cognition
The ability to understand and remember the spatial relationships between objects.
zero-shot reasoning
The ability of a model to make predictions or inferences without having seen any examples of the task during training.
structured problem-solving
A reasoning approach that involves breaking down complex problems into manageable parts to find solutions.
VANS-Data-100K
A dedicated dataset created for training and evaluating models on the Video-Next-Event Prediction task.
V-ReasonBench
A benchmark suite for assessing video reasoning abilities in generative video models.
Video Diffusion Model
A generative model that creates videos by diffusing information over time, often conditioned on textual or visual inputs.
Joint-GRPO
A method that orchestrates the collaboration between a Vision-Language Model and a Video Diffusion Model to optimize their outputs based on a shared reward.