MemN2N PyTorch Implementation
PyTorch implementation of End-To-End Memory Networks, NIPS 2015
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: End-to-End Memory Network PyTorch Implementation [-h]
[--vocabulary-size VOCABULARY_SIZE]
[--embedding-size EMBEDDING_SIZE]
[--sentence-size SENTENCE_SIZE]
[--memory-size MEMORY_SIZE]
[--hops HOPS]
[--weight-tying-scheme {adjacent,layerwise,None}]
[--babi-dataset-name BABI_DATASET_NAME]
[--babi-tasks BABI_TASKS [BABI_TASKS ...]]
[--epochs EPOCHS]
[--test-size TEST_SIZE]
[--batch-size BATCH_SIZE]
[--weight-decay WEIGHT_DECAY]
[--grad-clip-norm GRAD_CLIP_NORM]
[--lr LR]
[--lr-decay LR_DECAY]
[--lr-decay-epochs LR_DECAY_EPOCHS [LR_DECAY_EPOCHS ...]]
[--checkpoint-interval CHECKPOINT_INTERVAL]
[--eval-log-interval EVAL_LOG_INTERVAL]
[--loss-log-interval LOSS_LOG_INTERVAL]
[--gradient-log-interval GRADIENT_LOG_INTERVAL]
[--model-dir MODEL_DIR]
[--dataset-dir DATASET_DIR]
[--resume-best | --resume-latest]
[--best] [--no-gpus]
(--train | --test)
Train
$ python -m visom.server &
$ ./main.py --train [--resume-latest | --resume-best]
Test
$ ./main.py --test
Reference
Author
Ha Junsoo / @kuc2477 / MIT License