To compile a sentiment RNN model for an Alveo™ U50 data center accelerator card with batch=4 engine, follow these steps:
- Activate the vai_c_rnn conda environment in the docker image:
- For Pytorch:
conda activate vitis-ai-rnn-pytorch
- For Tensorflow:
conda activate vitis-ai-rnn-tensorflow
- For Pytorch:
- Compile
Assuming the quantized models for sentiment are prepared in the XMODEL directory, run the compilation command and get results. The output XMODEL is present in the output directory.
Figure 1. Compile XMODEL