Dr2Net: Dynamic Reversible Dual-Residual Networks for Memory-Efficient Finetuning
Open paper
Benchmark
Model
Rank
Results
action-recognition-in-videos-on-something-something-v2
Dr2Net
–
Top-1 Accuracy: 68.6
object-detection-on-coco-val2017
Dr2Net
–
box AP: 54.7
semantic-segmentation-on-s3dis-area5
Dr2Net
–
mIoU: 68.13
Rank counts only results with a code link.