NVIDIA-Merlin / models

Merlin Models is a collection of deep learning recommender system model reference implementations
https://nvidia-merlin.github.io/models/main/index.html
Apache License 2.0
262 stars 50 forks source link

Adding evaluate- and test-step + introduce custom Trainer #1176

Closed marcromeyn closed 1 year ago

marcromeyn commented 1 year ago

Goals :soccer:

This PR introduces a custom Trainer object that reduces boilerplate. With this, a model can be trained as follows:

model = mm.Model(
   TabularInputBlock(schema),
   MLPBlock([32, 16]),
   BinaryOutput(schema.select_by_tag(Tags.TARGET).first),
)
trainer = mm.Trainer(max_epochs=1)
trainer.fit(model, dataset, batch_size=16)
github-actions[bot] commented 1 year ago

Documentation preview

https://nvidia-merlin.github.io/models/review/pr-1176