search

tensorflow / skflow

Simplified interface for TensorFlow (mimicking Scikit Learn) for Deep Learning
Apache License 2.0
3.18k stars 439 forks source link

Example of language model #64

Closed ilblackdragon closed 8 years ago

ilblackdragon commented 8 years ago

Add an example of language model (RNN). For example character level on sheikspear book (similar to https://github.com/sherjilozair/char-rnn-tensorflow).

ilblackdragon commented 8 years ago

Added a simple character language model in #74

evanmiltenburg commented 8 years ago

Could you please add an RNN/LSTM example for labeling tasks as well? (E.g. sentiment analysis or NER)

ilblackdragon commented 8 years ago

Hi Emiel, the sentiment analysis can be done from text_classification examples (sentiment is just a two or five class document classification). NER is more like language model structure. We should add some concrete examples though with datasets though over time. On Feb 14, 2016 7:36 AM, "Emiel van Miltenburg" notifications@github.com wrote:

Any chance of an RNN/LSTM example for labeling tasks as well? (E.g. sentiment analysis http://www.cs.cornell.edu/%7Eoirsoy/files/emnlp14drnt.pdf or NER http://www.aclweb.org/anthology/W03-0426)

— Reply to this email directly or view it on GitHub https://github.com/tensorflow/skflow/issues/64#issuecomment-183906297.

evanmiltenburg commented 8 years ago

Cool, thanks!

unography commented 8 years ago

Hi, not a related comment, but how do I go about doing multi label image classification using skflow? Do I do something similar to the multioutput code?

ilblackdragon commented 8 years ago

@eldor4do For multi label classification - you can just pass a number of active classes to the model - e.g. y to be a matrix [examples, n_classes]. You just need to decide on your loss in the function, regular Estimators may not work very well as this point.

ilblackdragon commented 8 years ago

Closing this issue as language model is in repository in limited capacity.