mirror of
https://github.com/saymrwulf/onnxruntime.git
synced 2026-05-15 20:50:42 +00:00
1.2 KiB
1.2 KiB
TransformerModel example
This example was adapted from Pytorch's Sequence-to-Sequence Modeling with nn.Transformer and TorchText tutorial
Requirements
- PyTorch 1.6+
- TorchText 0.6+
- ONNX Runtime 1.5+
Running PyTorch version
python pt_model.py
Running ONNX Runtime version
python ort_model.py
Optional arguments
| Argument | Description | Default |
|---|---|---|
| --batch-size | input batch size for training | 20 |
| --test-batch-size | input batch size for testing | 20 |
| --epochs | number of epochs to train | 2 |
| --lr | learning rate | 0.001 |
| --no-cuda | disables CUDA training | False |
| --seed | random seed | 1 |
| --log-interval | how many batches to wait before logging training status | 200 |