Robot Learning on CortexBench
74.7Adroit ScoreR3M
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| R3MModel=RN-502025.12 | 74.7 | 72.4 | 94.1 | 67.4 | 77.2 | |
| DINOv3 + SpatialBoostVision Encoder=ViT-7B/16, SpatialBoost augmentation=true2026.03 | 71.8 | 80.4 | 92 | 79 | 80.8 | |
| PixioModel=ViT-H2025.12 | 70.7 | 77.5 | 92.8 | 72.8 | 78.4 | |
| DINOv2 + SpatialBoostVision Encoder=ViT-g/14, SpatialBoost augmentation=true2026.03 | 68.1 | 75 | 88.5 | 71.4 | 75.8 | |
| SigLIPv2 + SpatialBoostVision Encoder=ViT-g/16, SpatialBoost augmentation=true2026.03 | 66.5 | 73.5 | 89.1 | 73.9 | 75.8 | |
| DINOv2Model=ViT-L2025.12 | 64 | 70.4 | 90.9 | 66.9 | 73.1 | |
| DINOv3Vision Encoder=ViT-7B/16, SpatialBoost augmentation=false2026.03 | 63.9 | 70.8 | 83.8 | 72.8 | 72.8 | |
| DINOv3Model=ViT-H+2025.12 | 63.3 | 78.5 | 89.3 | 70 | 75.3 | |
| OpenCLIP + SpatialBoostVision Encoder=ViT-G/14, SpatialBoost augmentation=true2026.03 | 61.1 | 61 | 87 | 72.9 | 70.5 | |
| ToBoBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 60.4 | 73.5 | 87.8 | 66.5 | — | |
| TheiaModel=ViT-B2025.12 | 60 | 79.9 | 87.7 | 65.8 | 73.4 | |
| SigLIPv2Vision Encoder=ViT-g/16, SpatialBoost augmentation=false2026.03 | 56.5 | 69.4 | 84.7 | 68.3 | 69.7 | |
| DINOv2Vision Encoder=ViT-g/14, SpatialBoost augmentation=false2026.03 | 55.4 | 67.9 | 82.4 | 66.8 | 68.1 | |
| OpenCLIPVision Encoder=ViT-G/14, SpatialBoost augmentation=false2026.03 | 52.6 | 58.5 | 83 | 67.7 | 65.5 | |
| VC1Model=ViT-L2025.12 | 52 | 81.3 | 88.3 | 67.5 | 72.3 | |
| CropMAEBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 50 | 46.4 | 82.4 | 46.3 | — | |
| DINOBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 45.6 | 50.9 | 82.4 | 64.2 | — | |
| RSPBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 45.6 | 61.6 | 84.5 | 66.2 | — | |
| SiamMAEBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 44 | 56 | 81.1 | 52.1 | — | |
| SimCLRBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 40.4 | 39.7 | 78.4 | 63.3 | — | |
| MoCo v3Backbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 39.6 | 43.7 | 65.4 | 53.3 | — | |
| MAEBackbone=ViT-S/16, Pre-trained=Kinetics-400, Evaluation Protocol=Imitation Learning2025.07 | 39.6 | 43.7 | 65.4 | 53.3 | — |