Closed lzw950905 closed 3 years ago
The model.fit()
function uses generate_corruptions_for_fit internally and generates eta
corruptions per triple.
https://docs.ampligraph.org/en/1.3.2/examples.html
Here is an example to train the model on your dataset
When I use the tutorial about 'train and evaluate the model', we use model = mymodel(....), and one of parameter is eta(the number of corruption produced from a positive triple). And I notice we could generate corruption from batch(generate_corruptions_for_fit()), so I want to know how could I use this function at the whole training process?