Questions about metrics

[handsomelys]

In the given source code, acc@k is used to evaluate the model on the valid set, and hit@k is used to evaluate the model on the test set. Are there any similarities and differences between these two metrics?

[intfloat]

Acc@k and Hit@k are the same metrics with different names. For both metrics, a prediction is considered correct if it is in the top-k ranked list.

The main difference in the implementation is that:

Acc@k is computed within a batch, since computing embeddings for all the entities is time-consuming.

Hit@k is computed over all candidate entities, this is the metric reported in the paper.

[handsomelys]

thanks