lmjohns3 / theanets

Neural network toolkit for Python

http://theanets.rtfd.org

lmjohns3/theanets Issues

Proposing a PR to fix a few small typos
Updated 3 years ago
xor example not working
Updated 6 years ago
Problem: Cannot pass theano.shared to .itertrain()
Updated 7 years ago
Error in Quick Start: Recurrent Models
Updated 7 years ago2
Preserve recurrent hidden state across training examples
Updated 8 years ago2
Problem with reproducibility for release 0.6.2
Closed 8 years ago2
add feed_forward-like method for computing loss value
Updated 8 years ago1
create a model for jointly optimizing multiple models
Closed 9 years ago4
Add 3D convolution layers and corresponding models
Updated 8 years ago
New loss function: ROC AUC score
Updated 8 years ago1
Problem saving a network with compound activations
Closed 8 years ago3
feature: use file-like objects for net saving and loading
Closed 8 years ago5
autoencoder stop training
Updated 8 years ago3
use shape information in layers where possible
Closed 8 years ago
UnsupervisedPretrainer broken
Closed 8 years ago1
Is there any way to get feedback during training?
Closed 8 years ago1
The current clockwork RNN implementation
Closed 8 years ago7
for clockwork RNN
Closed 8 years ago
After finishing training, get the number of iterations
Closed 8 years ago1
Hidden_l1 problem
Updated 8 years ago1
add "bayesian rnn" implementation
Updated 8 years ago1
add
Closed 8 years ago
memory leaks when calling predict() and load()
Updated 9 years ago
[DOCS] Update MNIST example to use 2D Convolutions
Closed 9 years ago1
Unable to load a theanets.Regressor from file using the load
Closed 9 years ago2
'SGD' object has no attribute 'iterate'
Closed 9 years ago2
RNN not converging
Closed 9 years ago6
unexpected keyword argument 'rng' in graph.py
Closed 9 years ago7
Error in build_graph (The updates parameter must be an OrderedDict/dict)
Closed 9 years ago3
feature: automatically switch b/w GPU and CPU for performance
Closed 9 years ago5
theanets package on PyPI is broken or obsolete
Closed 9 years ago1
hidden dropout w/ 'predict' not performing as expected
Closed 9 years ago1
new feature request: Allow theanets.recurrent.batches(...) to accept a pandas dataframe
Updated 9 years ago
does theanets support gradient of subtensor?
Closed 9 years ago1
Any plans for supporting multi dimensional RNN?
Closed 9 years ago1
Classification with two classes
Closed 9 years ago1
Bug in MUT1
Closed 9 years ago1
add model-wide rng parameter
Closed 9 years ago
Model-rng only for Regressor
Closed 9 years ago1
RNN not able to predict using existing state
Closed 9 years ago1
python and ipython crashes when running examples on Win 8 x64
Closed 9 years ago6
Layerwise GPU memory use
Updated 9 years ago1
CPU/GPU saving/conversion -- what actually needs to be serialized?
Closed 9 years ago3
Perilous pretraining, poorly named outputs causing key errors? 'pretrain' issue.
Closed 9 years ago4
Can't deserialize Regressor networks: AttributeError: 'Regressor' object has no attribute '_functions' as of 33775e7c96
Closed 9 years ago2
Using theanets with ipython notebook
Closed 9 years ago7
Looks like save_progress and save_every are not working in v0.6.2
Closed 9 years ago2
Problems continuing to train Experiment loaded from file.
Closed 9 years ago3
specifying layers -- output layer
Closed 9 years ago1
Why do I get this error message? IndexError: index 4 is out of bounds for axis 1 with size 4
Closed 9 years ago3