katakana icon indicating copy to clipboard operation
katakana copied to clipboard

verified publisher icon wanasit
→

Metadata

Training machine to write Katakana using Sequence-to-Sequence technique

Results 2 katakana issues
Sort by recently updated
recently updated
newest added

About the method of passing encoder states to decoder

Based on `lstm_seq2seq.py` example seems there is a better way to pass encoder states to decoder: https://github.com/keras-team/keras/blob/7a39b6c62d43c25472b2c2476bd2a8983ae4f682/examples/lstm_seq2seq.py#L129-L132 https://github.com/keras-team/keras/blob/7a39b6c62d43c25472b2c2476bd2a8983ae4f682/examples/lstm_seq2seq.py#L139-L141 compared to https://github.com/wanasit/katakana/blob/72d2f91066aa22531f242ba67ff9092ca5f2394b/katakana/model.py#L62-L65

mrgloom avatar mrgloom

If you need advise how to train your model

3 comment

Ping me

philipperemy avatar philipperemy

← Metadata

62
Stars
17
Forks
Watchers

Owner

verified publisher icon wanasit

Metadata

Training machine to write Katakana using Sequence-to-Sequence technique

Back