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 960 forks source link

when training, model can be initialized from previously saved model #14

Closed wohnjayne closed 8 years ago

wohnjayne commented 8 years ago

implements the --init-from as in char-rnn.

note that this implements model initialization only - optimization specific params (such as the previous learning rate/epoch/batch number) are not restored

sherjilozair commented 8 years ago

@wohnjayne Cool! This is really useful. I'll make sure I play with this later today. Thanks for your contribution.

hunkim commented 8 years ago

This is cool. Is it merged?

bver commented 8 years ago

+1 for this proposal.

hunkim commented 8 years ago

+1

sherjilozair commented 8 years ago

I can merge this if the merge conflicts are fixed.

wohnjayne commented 8 years ago

Sorry, I wasn't aware of the upstream changes. Should be ready to merge now

sherjilozair commented 8 years ago

Thanks!