search

sherjilozair / char-rnn-tensorflow

Multi-layer Recurrent Neural Networks (LSTM, RNN) for character-level language models in Python using Tensorflow
MIT License
2.64k stars 959 forks source link

the tensorboard summary is slow #112

Open liyuanhangsc opened 6 years ago

liyuanhangsc commented 6 years ago

in my gpu server, remove the summary tensor make training 10x times faster. Please consider add a switch to close tensorboard summary.

john-parton commented 6 years ago

I'm pretty sure the culprit is https://github.com/sherjilozair/char-rnn-tensorflow/blob/401ebfd639eb32586a9bf9312e3b442abc47ebed/model.py#L85-L86

(Histogram summaries involve transferring a lot of data from the GPU to the CPU.)

You might try commenting them out. The graph and train_loss summaries are usually more useful, anyways.

Mahawi3211 commented 5 years ago

How can I see the accuracy rate in tensorboard?