Error when multiplying prediction logits by output mask tensor

AndreasMadsen / python-textualheatmap

Create interactive textual heat maps for Jupiter notebooks

MIT License

196 stars 14 forks source link

Error when multiplying prediction logits by output mask tensor #6

Open dr-smgad opened 3 years ago

dr-smgad commented 3 years ago

I'm using TF 2.4 and this line predict_mask_correct_token = tf.reduce_sum(predict * output_mask_tensor) gives error TypeError: Cannot convert 'logits' to EagerTensor of dtype float

Any ideas?

wenlinyao commented 3 years ago

Have the same problem, anyone can help us?