# RKNN OP Support
Base on RKNN Toolkit Version 1.7.1
## Caffe OPs supported by RKNN
There are two caffe protocols RKNN Toolkit uses, one based on the officially modified protocol of berkeley, and one based on the protocol containing the LSTM layer.
The protocol based on the official revision of berkeley comes from [berkeley caffe](https://github.com/BVLC/caffe/tree/master/src/caffe/proto "Berkeley Caffe"), commit hash is 21d0608. On this basis RKNN Toolkit have added some OPs.
The protocol containing the LSTM layer refers to [warpctc caffe](https://github.com/xmfbit/warpctc-caffe/tree/master/src/caffe/proto "warpctc caffe"), commit hash is bd6181b.
Based on these protocols, the list of Caffe OPs supported by RKNN is as follows:
| **Operators** |
|---|
| absval |
| axpy |
| batchnorm |
| bn |
| concat |
| convolution |
| convolutiondepthwise |
| deconvolution |
| dropout |
This file has been truncated. show original