Skip to content

Commit f18a3a8

Browse files
Create README.md
1 parent cb51c43 commit f18a3a8

File tree

1 file changed

+34
-0
lines changed

1 file changed

+34
-0
lines changed

README.md

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
# GA_Reader
2+
Implementation of the paper [Gated-Attention Readers for Text Comprehension](https://arxiv.org/pdf/1606.01549.pdf) in PyTorch.
3+
4+
### Dataset
5+
This code only supports training and validation on the CNN and Daily Mail cloze style dataset.
6+
7+
Data can be dowloaded [here](https://drive.google.com/drive/folders/0B7aCzQIaRTDUZS1EWlRKMmt3OXM?usp=sharing)
8+
9+
### Training
10+
Include all training data, validation data and testing data in the same folder, first training data followed by validation and testing data.
11+
12+
For training and testing the model run
13+
```
14+
python train.py
15+
```
16+
17+
Optional Arguments
18+
```
19+
--epochs number of epochs
20+
--iterations iterations per epoch
21+
--lr learning rate for Adam optimizer
22+
--char_size size for character embeddings
23+
--embed_size size for glove embeddings
24+
--char_hidden_size hidden size for character GRU
25+
--hidden_size hidden size for document and query GRU
26+
--use_char whether to use character embeddings or not
27+
--use_features whether to use qe-comm features or not
28+
--batch_size batch size for training
29+
--gru_layers number of GRU layers for document and query
30+
--train_file file having all the data (training+validation+testing)
31+
--training_size number of training examples
32+
--dev_size number og validation examples
33+
--test_size number of testing examples
34+
```

0 commit comments

Comments
 (0)