on the training_nli.py of NLI training example

[lukemao]

Hi,

Thanks for providing an example of training SBERT on NLI examples.

I have a question on the example code provided training_nli.py

At line: train_loss = losses.SoftmaxLoss(model=model, sentence_embedding_dimension=model.get_sentence_embedding_dimension(), ....

Does this line automatically prepare the input to the softmax operator.

When I saw 'input' I mean the concatenated vector:

(u, v, |u-v|)

Lu.

[nreimers]

Yes

[lukemao]

Thanks Nils !