luozhouyang
luozhouyang
I install OpenNMT-tf using pip, and train this model in a **docker container** based on **tensorflow/tensorflow:2.0.0-gpu-py3** image.
`tf.data` API方便且强大，个人很喜欢。
`tf.keras.callbacks.ModelCheckpoint` has an arugument `save_best_only`, you can use this callback to save the best model. [ModelCheckpoint](https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/ModelCheckpoint)
按batch切词即可，比如一次性切32/64个句子，batch size可以根据机器性能调整。
Update: This upstream will be reset when kong pods are redeployed. But when I update the upstream, I found that this upstream is persisted to my postgres database. So why...
This code is not for tensorflow 2.x. tensorflow 2.x has many break changes.
you need these file: * train.src, a file contains questions. * train.tgt, a file contains responses. * dev.src, a file contains questions. * dev.tgt, a file contains responses. * test.src,...
The logout is clear, your vocab file contains empty line. You need to make sure: 1. NO empty line(s) in the vocab files 2. NO repeated words in the vocab...
No. If you do not provide the pretrained embeddings, it will create an trainable variable, and initialize it by some algorithm. When you train the model on your data, this...