search

QIFEIDKN / STAGATE

Adaptive Graph Attention Auto-encoder for Spatial Domain Identification of Spatial Transcriptomics
MIT License
36 stars 19 forks source link

I think this part in model.py is wrong. #18

Open SanoKyohei opened 1 year ago

SanoKyohei commented 1 year ago

[Original] for layer in range(self.n_layers): weight_decay_loss = 0 weight_decay_loss += tf.multiply(tf.nn.l2_loss(self.W[layer]), self.weight_decay, name='weight_loss') [Right] weight_decay_loss = 0 for layer in range(self.n_layers): weight_decay_loss += tf.multiply(tf.nn.l2_loss(self.W[layer]), self.weight_decay, name='weight_loss')

QIFEIDKN commented 1 year ago

Thank you very much for the correction! We fixed this error in the new code. Due to the relatively shallow network structure, this error did not have a big impact on the results.