nni
nni copied to clipboard
Unsupported Layer nn.TransformerEncoderLayer
Describe the bug:
Environment:
- NNI version: 3.0
- Training service (local|remote|pai|aml|etc): local
- Python version: 3.10
- PyTorch version: 1.13
- Cpu or cuda version: Cuda/Cpu
Reproduce the problem
- How to reproduce: It seems like the class
TransformerEncoderLayer
is not currently supported by NNI. I am attempting to prune the network RepNet . Can someone help adding the implementation for that?