4 papers
How Wrong Can a Good Predictor Be? Diverging Updates with Vanishing Predictive KL
Qifu Wen, Shuaijun Liu, Zihan Zhou +2
Accurate posterior prediction need not require accurate approximation of Bayesian updates. We prove that an unbounded gap between the update maps can coexist with vanishing predict…
SR-JEPA: Learning Predictive Latent State in 3D Scenes
Zihan Zhou, Qifu Wen, Xi Zeng
Joint-embedding predictive architectures learn by predicting latent representations of missing observations, yet many masked JEPAs are evaluated primarily through the encoders they…
PhyLatent: Learning Dynamics-Relevant Representations for JEPA World Models
Xi Zeng, Haojie Ren, Ziying Song +2
We propose PhyLatent, a dynamics-relevant training objective for Joint-Embedding Predictive Architecture (JEPA) world models. Our key observation is that preventing global latent c…
GradES: Significantly Faster Training in Transformers with Gradient-Based Early Stopping
Qifu Wen, Xi Zeng, Zihan Zhou +4
Early stopping monitors global validation loss and halts all parameter updates simultaneously, which is computationally costly for large transformers due to the extended time requi…