Views
No views yet
mobilevitv2_125timm/mobilevitv2_125.cvnets_in1kmodel.tflitefloat32 tensor in NCHW layout, shape [1, 3, 256, 256][1, 1000]CompiledModel.RESHAPE, TRANSPOSE, and related window/attention operations. The model is published as CPU-ready while GPU support is being improved.1@article{Mehta2022SeparableSF,
2 title={Separable Self-attention for Mobile Vision Transformers},
3 author={Sachin Mehta and Mohammad Rastegari},
4 journal={ArXiv},
5 year={2022},
6 volume={abs/2206.02680}
7}