r3m-50 / README.md
surajnair's picture
Create README.md
51a1ff9

This model contains the pre-trained ResNet50 R3M model from the paper "R3M: A Universal Visual Representation for Robot Manipulation" (Nair et al.) The model is trained on the Ego4D dataset using time-contrastive learning, video-language alignment, and sparsity objectives. It is used for efficient downstream robotic learning.