Inference with PyTorch

Here are the steps to use Galileo for Inference with a text classification model.
  1. 1.
    Initialize the Galileo project - Use the dq.init function to initialize the Galileo project. The function takes three arguments: the project type, the project name, and the run name.
  2. 2.
    Log the inference data - Use the dq.log_dataset function to log the inference data. The function takes two arguments: the inference data and the split name.
  3. 3.
    Set the class labels - Use the dq.set_labels_for_run function to set the class labels for the run. The function takes one argument: the names of the classes.
  4. 4.
    Set the split to inference with dq.set_split
  5. 5.
    Monitor the model - Use the watch function from the Galileo's PyTorch integration to monitor the model. The function takes two arguments: the model and the dataloader.
  6. 6.
    Run the model on the data using the data loader
  7. 7.
    Finish the run - Use the dq.finish function to finish the run.
In the example notebook you can see how it is called:
Once the training is complete in the console you select the inference dataset as your subset in the top right corner.
Example Notebook: PyTorch Inference​