Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add files via upload #1

Open
wants to merge 104 commits into
base: master
Choose a base branch
from
Open

Add files via upload #1

wants to merge 104 commits into from

Conversation

xinqianpku
Copy link
Collaborator

No description provided.

xinqianpku and others added 30 commits April 19, 2019 15:37
@xinqianpku
Add files via upload
2d642bf
@xinqianpku
Add files via upload
00c772a
@matthewkennedy5
Removed unnecessary files and added the training data.
1f14359
@matthewkennedy5
Merge branch 'planB' of https://github.com/matthewkennedy5/RRNN into …
e009b32 
…planB
@matthewkennedy5
Removed unnecessary files for real
ad7dcef
@xinqianpku
Update tree_methods.py
2e93f15
@matthewkennedy5
The hyperparameters are now loaded from an external JSON parameter fi…
9dc21fc 
…le instead of coded into the .py file. This makes it easier to keep track of many runs.
@matthewkennedy5
Merge branch 'planB' of https://github.com/matthewkennedy5/RRNN into …
4322ebb 
…planB
@matthewkennedy5
Added code to perform validation on external saved model checkpoints.
082a9f8
@matthewkennedy5
Fixed inputs to RRNNforGRU constructor
81bb855
@matthewkennedy5
Added a Dataset subclass to represent the enwik8 data. This way we ca…
f3d6e5a 
…n use Pytorch DataLoaders.
@matthewkennedy5
Modified to include the new dataloaders from standard_data.py.
bbcd279
@xinqianpku
Update GRU.py
28354d2
@xinqianpku
Update tree_methods.py
307ad5f
@xinqianpku
Update trainer.py
3aa1347
@xinqianpku
Update GRU.py
442830b
@xinqianpku
Update trainer_temp.py
d01fc95
@xinqianpku
Update tree_methods.py
8b6f2ed
@xinqianpku
Update trainer.py
3790de0
@xinqianpku
Removed all codes about Nodes etc.
dd7469e
@xinqianpku
Update tree_methods.py
4d60af8
@xinqianpku
Update trainer_temp.py
c19fe5b
@xinqianpku
Update trainer.py
5d58de8
@matthewkennedy5
Allowing the EnWik8Clean dataset to have a variable number of samples…
d845405 
… instead of the default 10,000.
@matthewkennedy5
Updated the training code to work with Xin's new vectorized implement…
fbe7344 
…ation.
@matthewkennedy5
Turned off pretrained_weights by default since they cause exploding g…
9be28b4 
…radients.
@matthewkennedy5
Removed unused params.
ed5ad91
@xinqianpku
Updated margin function
9f962fd
@xinqianpku
Update trainer_temp.py
1428304
@xinqianpku
Support margin loss
4592e56
matthewkennedy5 and others added 30 commits April 29, 2019 16:56
@matthewkennedy5
First steps at a script to convert Keras GRU weights to Pytorch tensors.
2776ad7
@matthewkennedy5
Made the scoring hidden size range more realistic in the random searc…
abad3f4 
…h (too small previously).
@matthewkennedy5
Forgot to actually add the data to the zero-padded tensor so it was j…
c402572 
…ust outputting all zeros for the SST dataset.
@matthewkennedy5
Updated display code
692629a
@xinqianpku
Update GRU.py
e5c0400
@xinqianpku
Update tree_methods.py
7a38957
@xinqianpku
Update trainer.py
ef7954d
@matthewkennedy5
Made the model save after training.
3c45b05
@matthewkennedy5
Wrote code to convert the Keras saved weights to Pytorch format.
df1752b
@matthewkennedy5
Increased the number of validation samples to 10000 and the number of…
75bf6a6 
… test to 20,000 for increased precision.
@matthewkennedy5
Divides loss2 and loss4 by the batch size for consistent comparison w…
7d88bf9 
…ith datasets of different sequence lengths.
@matthewkennedy5
Changed the gru_model parameters to the trained baseline parameters f…
93e3565 
…rom keras_gru.
@matthewkennedy5
Added code to save Adam's configuration during training for smooth wa…
f10a86d 
…rm-starting.
@matthewkennedy5
Added functionality for the PTB dataset for the keras baseline GRU.
f95c5f6
@matthewkennedy5
Enabled normalization of PTB and SST based on training mean and stand…
6453f3b 
…ard deviation.
@matthewkennedy5
Serialized normalization statistics so the program starts training mo…
663a048 
…re quickly.
@matthewkennedy5
Found better ground truth GRU parameters for the wiki dataset.
2b7143d
@matthewkennedy5
Rough PTB random search parameters.
0924bdc
@matthewkennedy5
GRU baseline PTB fine random search range.
2e9a01f
@matthewkennedy5
Now works for the SST dataset, but somehow I broke it for the other t…
276ba81 
…wo datasets. Not sure what the issue is since the code should be the same.
@matthewkennedy5
Ranges for course SST random hyperparameter baseline search.
85c420e
@matthewkennedy5
SST fine random search ranges.
349d07a
@matthewkennedy5
Re-add SST dataset
a92efa2 
Somehow a prior commit where I added the sst dataset got reverted, so I included those changes manually here again. Hopefully no other issues remain with the accidental revert.
@matthewkennedy5
Random hyperparameter search for the RRNN.
624aabc 
The ranges listed in lines 428-440 are the hyperparameter ranges used for the initial rough wiki random search.
@matthewkennedy5
RRNN PTB rough hyperparameter ranges.
da64a1e
@matthewkennedy5
Fixed a CUDA/CPU tensor type bug in the normalization.
f7a6447
@matthewkennedy5
RRNN SST rough param ranges.
fb4a6e3
@matthewkennedy5
RRNN Wiki fine param ranges.
518fd2e
@matthewkennedy5
RRNN PTB fine hp ranges.
f7e51a9
@matthewkennedy5
Added code to evaluate the test set.
e9fe074
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
2 participants
@xinqianpku @matthewkennedy5