CyberZHG / keras-transformer

Transformer implemented in Keras

https://pypi.org/project/keras-transformer/

CyberZHG/keras-transformer Issues

Code example of a transformer model based on character level input sequence and word level output sequence
Closed a year ago1
Installation problem on MAC m1
Closed 2 years ago1
what the required version of tensorflow and keras for this project?
Closed 2 years ago
ValueError: An operation has `None` for gradient. Please make sure that all of your ops have a gradient defined (i.e. are differentiable). Common ops without gradient: K.argmax, K.round, K.eval.
Closed 2 years ago1
Easiest way to output attention scores
Closed 3 years ago1
Question concerning default settings
Closed 3 years ago1
How can I adjust model to disable teacher forcing?
Closed 3 years ago4
Slow prediction
Closed 4 years ago1
why get the same output with different input?
Closed 4 years ago1
The input of function "_get_max_suffix_repeat_times" is not correct.
Closed 4 years ago
doesnt work when test set contains more than two sentences
Closed 4 years ago2
Hierarchy of Encoders Not Working
Closed 4 years ago1
How do you use masking with the transformer?
Closed 4 years ago2
How to handle multiple inputs
Closed 4 years ago1
Some question about the decode method
Closed 4 years ago1
module 'keras.layers' has no attribute 'Wrapper'
Closed 4 years ago1
Questions about data input and output
Closed 4 years ago1
Do I need to specify <start> and <end> of sequences
Closed 5 years ago1
why get same output with different input?
Closed 5 years ago1
Using get_encoders in my model, output is all zeros
Closed 5 years ago3
How to train a transformer model with fit_generator?
Closed 5 years ago2
star-transformer
Closed 5 years ago2
get metrics accuracy in compile
Closed 5 years ago1
experiments on Real world examples
Closed 5 years ago1
Error in get_model() : ValueError: too many values to unpack (expected 2)
Closed 5 years ago8
beam search
Closed 5 years ago2
**Describe the Bug**
Closed 5 years ago1
Possible bug: Encoder fed every step
Closed 5 years ago1
Is there only Multi-Head Attention?
Closed 5 years ago5
Use it as a layer
Closed 5 years ago4
ValueError: Unknown layer: EmbeddingRet
Closed 5 years ago1
Tensorflow dependency remove
Closed 5 years ago2
after stacked decoder layer，I think it should be concatenate with a linear and softmax layer.
Closed 5 years ago
Default value max_len=None in keras_transformer.decode is dangerous
Closed 5 years ago1
unexpected keyword argument 'activation'
Closed 5 years ago1
Invalid head number 3 with the given input dim 128
Closed 6 years ago1