JRLin

Contrastive loss weights

1

According to the paper, the contrastive loss weight is set to 1e-5 * num_train_steps. However, in the given config, the 'loss_weights: [10,1]', where 1 is the contrastive loss weight. May...

Hi, thank you for open source the datasets! Just wondering if the trained model and the model scripts will be open-sourced in the future? Thank you!