SMART-Lab smartlearner issues

SMART-Lab / smartlearner

SMART-Learner is a machine learning library built with researchers in mind.

BSD 3-Clause "New" or "Revised" License

10 stars 4 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Make sure we take a copy of the parameters and not a reference when stashing parameters during earlystopping

#76 MarcCote closed 8 years ago
0
Now saving the logger task. Also adds CallbackView

#75 MarcCote closed 8 years ago
0
Use right dtype when loading numpy arry from a json.

#74 MarcCote closed 8 years ago
0
Temporary bugfix; LossView fetches wrong updates

#73 ppoulin91 closed 8 years ago
2
LossView.__init__ should call loss.losses before loss.updates

#72 ppoulin91 opened 8 years ago
1
Using multiple losses/datasets on the same model grows the update dictionary

#71 ppoulin91 opened 8 years ago
0
Made getstate and setstate optional for Loss

#70 MarcCote closed 8 years ago
1
Direction/gradient clipping

#69 MarcCote closed 8 years ago
5
Added RMSProp and Adadelta optimizers

#68 MarcCote closed 8 years ago
1
Fix ClassificationError and added unit tests for all losses.

#67 MarcCote closed 8 years ago
1
Sampling is very slow

#66 ASalvail closed 8 years ago
3
Added a GradientNoise direction modifier

#65 ppoulin91 closed 8 years ago
5
Make AveragePerEpoch indexable

#64 MarcCote closed 8 years ago
0
Made sure BinaryCrossEntropy return the losses of the batch.

#63 MarcCote closed 8 years ago
1
Added save and load functions to the framework.

#62 MarcCote closed 8 years ago
4
When stopping criteria raise TrainingExit remaining tasks might be skipped

#61 MarcCote opened 8 years ago
0
Use a fixed path for miniconda

#60 MarcCote closed 9 years ago
1
Adds a functionality to link a dataset to another.

#59 ASalvail closed 9 years ago
3
Derive dataset

#58 ASalvail closed 9 years ago
0
Preprocess interface + implementations

#57 ASalvail closed 8 years ago
3
Modified init task of early stopping

#56 ASalvail closed 9 years ago
0
Two changes that make life easier.

#55 ASalvail closed 9 years ago
1
Experiments, Benchmark and Hyperoptimizer

#54 ASalvail opened 9 years ago
1
Added a task to save results to file

#53 ASalvail closed 9 years ago
0
Make `name` parameter of `sharedX` mandatory

#52 MarcCote opened 9 years ago
0
WIP: Adam optimizer

#51 ASalvail closed 9 years ago
1
Change MonitorVariable uses in test_optimizer

#50 ASalvail closed 9 years ago
2
Logger

#49 ASalvail closed 9 years ago
9
Add Adam optimizer

#48 MarcCote closed 9 years ago
1
Add a task that prints the average training loss at each epoch.

#47 MarcCote closed 9 years ago
4
Move the graph compilation to its own private method.

#46 MarcCote closed 9 years ago
2
`Loss` subclasses should compute the loss of every example in the batch

#45 MarcCote closed 9 years ago
3
Add a meta trainer

#44 ASalvail opened 9 years ago
0
Cross validation utilities

#43 ASalvail closed 9 years ago
2
Add option telling Theano do not transfer a dataset to the GPU

#42 MarcCote closed 9 years ago
0
In EarlyStopping restore model in method finished

#41 MarcCote closed 9 years ago
0
Removed duplicate code for batch scheduler.

#40 MarcCote closed 9 years ago
0
Added a finished print to the console to tell which epoch and cost was best

#39 ASalvail closed 9 years ago
0
Moved and renamed the module `minibatch.py` to `batch_schedulers.py`

#38 ASalvail closed 9 years ago
0
Moved all tests in another folder.

#37 MarcCote closed 9 years ago
2
Move tests_stopping_criteria.py to folder tests/tasks/

#36 MarcCote closed 9 years ago
1
Fixes #33, added unit tests for dataset.py

#35 MarcCote closed 9 years ago
1
Refactor how the updates are handled

#34 ASalvail closed 9 years ago
2
Move tests at the root of the library.

#33 MarcCote closed 9 years ago
0
The type of inputs and target symbolic variables should be guessed from the data.

#32 havaeimo closed 9 years ago
1
Refactored optimizer to easily access the gradients and the directions

#31 MarcCote closed 9 years ago
3
Use symbolic variables coming from the dataset

#30 MarcCote closed 9 years ago
0
Added EarlyStopping

#29 MarcCote closed 9 years ago
1
Install automatically *all* submodules contained in the library

#28 MarcCote closed 9 years ago
1
Add travis support

#27 MarcCote closed 9 years ago
1