README.md

Graphcore

This example demonstrates how to train a network on the MNIST dataset using PopTorch.

Prepare the environment.

Install the Poplar SDK following the instructions in the Getting Started guide for your IPU system. Make sure to run the enable.sh scripts for Poplar and PopART and activate a Python virtualenv with PopTorch installed.

Then install the package requirements:
```
pip install -r requirements.txt
```
Run the program. Note that the PopTorch Python API only supports Python 3. Data will be automatically downloaded using torch vision utils.
```
python3 mnist_poptorch.py
```

The program has a few command-line options:

-h Show usage information.

--batch-size Sets the batch size for training.

--batches-per-step Number on mini-batches to perform on the device before returning to the host.

--test-batch-size Sets the batch size for inference.

--epochs Number of epoch to train for.

--lr Learning rate of the optimizer.