Skip to content

PyTorch implementation of ICML 2017 paper, SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization

License

Notifications You must be signed in to change notification settings

kuc2477/pytorch-splitnet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SplitNet PyTorch Implementation

PyTorch implementation of SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization, ICML 2017

model-architecture

Results

Learned Convolution Filters of g2-b1-w3 (left) and g3-b1-w3 (right).

Learned Group Indicators of g2-b2-p (left) and g3-b2-p (right).

Installation

$ git clone https://github.com:kuc2477/dl-papers && cd dl-papers
$ pip install -r requirements.txt

CLI

Implementation CLI is provided by main.py.

Usage

$ ./main.py --help
$ usage: SplitNet torch implementation [-h] [--dataset {cifar10,mnist,cifar100}]
                                 [--total-block-number TOTAL_BLOCK_NUMBER]
                                 [--widen-factor WIDEN_FACTOR]
                                 [--baseline-strides BASELINE_STRIDES [BASELINE_STRIDES ...]]
                                 [--baseline-channels BASELINE_CHANNELS [BASELINE_CHANNELS ...]]
                                 [--split-sizes SPLIT_SIZES [SPLIT_SIZES ...]]
                                 [--gamma1 GAMMA1] [--gamma2 GAMMA2]
                                 [--gamma3 GAMMA3]
                                 [--weight-decay WEIGHT_DECAY]
                                 [--dropout-prob DROPOUT_PROB] [--lr LR]
                                 [--lr-decay LR_DECAY]
                                 [--lr-decay-epochs LR_DECAY_EPOCHS [LR_DECAY_EPOCHS ...]]
                                 [--epochs EPOCHS] [--batch-size BATCH_SIZE]
                                 [--test-size TEST_SIZE]
                                 [--eval-log-interval EVAL_LOG_INTERVAL]
                                 [--loss-log-interval LOSS_LOG_INTERVAL]
                                 [--weight-log-interval WEIGHT_LOG_INTERVAL]
                                 [--checkpoint-interval CHECKPOINT_INTERVAL]
                                 [--model-dir MODEL_DIR]
                                 [--resume-best | --resume-latest] [--best]
                                 [--no-gpus] (--test | --train)
 

Train

$ python -m visom.server &
$ ./main.py --train [--resume-latest|--resume-best]

Test

$ ./main.py --test

Note

  • Block diagonal weight utilizing test phase is not implemented yet.

Reference

Author

Ha Junsoo / @kuc2477 / MIT License

About

PyTorch implementation of ICML 2017 paper, SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages