wassname/Castor
1
0
Fork 0
mirror of https://github.com/wassname/Castor.git synced 2026-09-09 11:13:20 +08:00
Code Issues Packages Projects Releases Wiki Activity

GPU Support for SM Model (#26)

Browse Source 
Add code to using GPU for the SM Model (#25). To use the GPU if one is available, add the --cuda optional parameter when calling main.py.
This commit is contained in:
Michael Tu
2017-06-02 08:47:13 -04:00
committed by Jimmy Lin
parent 92789cb9f5
commit 43dd6fdb1f
6 changed files with 40 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.gitignore
+1
View File
@@ -1,3 +1,4 @@
.DS_Store
.idea/
*idfsim
*.swp
sm_cnn/README.md
+8 -6
View File
@@ -1,4 +1,4 @@
## SM model
## SM model
#### References:
1. Aliaksei _S_everyn and Alessandro _M_oschitti. 2015. Learning to Rank Short Text Pairs with Convolutional Deep Neural Networks. In Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '15). ACM, New York, NY, USA, 373-382. DOI: http://dx.doi.org/10.1145/2766462.2767738
@@ -20,9 +20,9 @@ Please install the requirements. See [Castor/README.md](../README.md) for pytorc
```
mkdir castorini
cd castorini
git clone https://github.com/castorini/data.git
git clone https://github.com/castorini/models.git
git clone https://github.com/castorini/Castor.git
git clone https://github.com/castorini/data.git
git clone https://github.com/castorini/models.git
git clone https://github.com/castorini/Castor.git
```
This should generate:
@@ -67,11 +67,13 @@ cd ..
To train the S&M model on TrecQA
```
python main.py ../../models/sm_model/sm_model.TrecQA.TRAIN-ALL.2017-04-02.castor --paper-ext-features
python main.py ../../models/sm_model/sm_model.TrecQA.TRAIN-ALL.2017-04-02.castor --paper-ext-feats
```
**To use the GPU, add `--cuda`.**
The final model will be saved to ```../../models/sm_model/sm_model.TrecQA.TRAIN-ALL.2017-04-02.castor```
_NOTE:_ On first run, the program will create a memory-mapped cache for word e mbeddings (943MB) in ``data/word2vec``.
_NOTE:_ On first run, the program will create a memory-mapped cache for word e mbeddings (943MB) in ``data/word2vec``.
The cache allows for faster loading of data in future runs.
Run ```python main.py -h``` for more default options.
sm_cnn/main.py
+5 -5
View File
@@ -91,7 +91,7 @@ if __name__ == "__main__":
ap.add_argument('--paper-ext-feats-stem', action="store_true", \
help="external features as per the paper")
# system arguments
# TODO: add arguments for CUDA
ap.add_argument('--cuda', action='store_true', help='use CUDA if available')
ap.add_argument('--num_threads', help="the number of simultaneous processes to run", \
type=int, default=4)
# training arguments
@@ -136,10 +136,10 @@ if __name__ == "__main__":
vocab_size, vec_dim = utils.load_embedding_dimensions(cache_file)
# instantiate model
net = QAModel(vec_dim, args.filter_width, args.num_conv_filters, args.no_ext_feats)
net = QAModel(vec_dim, args.filter_width, args.num_conv_filters, args.no_ext_feats, cuda=args.cuda)
# initialize the trainer
trainer = Trainer(net, args.eta, args.mom, args.no_loss_reg, vec_dim)
trainer = Trainer(net, args.eta, args.mom, args.no_loss_reg, vec_dim, args.cuda)
logger.info("Loading input data...")
# load input data
trainer.load_input_data(args.dataset_folder, cache_file, train_set, dev_set, test_set)
@@ -156,7 +156,7 @@ if __name__ == "__main__":
ext_feats_for_splits = \
set_external_features_as_per_paper_and_stem(trainer, args.index_for_corpusIDF)
if not args.skip_training:
best_map = 0.0
best_model = 0
@@ -192,7 +192,7 @@ if __name__ == "__main__":
logger.info('Best dev MAP in training phase = {:.4f}'.format(best_map))
trained_model = QAModel.load(args.model_outfile)
evaluator = Trainer(trained_model, args.eta, args.mom, args.no_loss_reg, vec_dim)
evaluator = Trainer(trained_model, args.eta, args.mom, args.no_loss_reg, vec_dim, args.cuda)
for split in [test_set, dev_set]:
evaluator.load_input_data(args.dataset_folder, cache_file, None, None, split)
sm_cnn/model.py
+6 -1
View File
@@ -27,7 +27,7 @@ class QAModel(nn.Module):
def __init__(self, input_n_dim, filter_width, \
conv_filters=100, no_ext_feats=False, ext_feats_size=4, n_classes=2):
conv_filters=100, no_ext_feats=False, ext_feats_size=4, n_classes=2, cuda=False):
super(QAModel, self).__init__()
self.no_ext_feats = no_ext_feats
@@ -53,6 +53,11 @@ class QAModel(nn.Module):
self.hidden = nn.Linear(n_hidden, n_classes)
self.logsoftmax = nn.LogSoftmax()
if cuda and torch.cuda.is_available():
self.conv_q, self.conv_a = self.conv_q.cuda(), self.conv_a.cuda()
self.combined_feature_vector = self.combined_feature_vector.cuda()
self.combined_features_activation = self.combined_features_activation.cuda()
self.dropout, self.hidden, self.logsoftmax = self.dropout.cuda(), self.hidden.cuda(), self.logsoftmax.cuda()
def forward(self, question, answer, ext_feats):
sm_cnn/train.py
+14 -6
View File
@@ -22,11 +22,12 @@ logger.addHandler(ch)
class Trainer(object):
def __init__(self, model, eta, mom, no_loss_reg, vec_dim):
def __init__(self, model, eta, mom, no_loss_reg, vec_dim, cuda=False):
# set the random seeds for every instance of trainer.
# needed to ensure reproduction of random word vectors for out of vocab terms
torch.manual_seed(1234)
np.random.seed(1234)
self.cuda = cuda
self.unk_term = np.random.uniform(-0.25, 0.25, vec_dim)
self.reg = 1e-5
@@ -56,7 +57,7 @@ class Trainer(object):
utils.load_cached_embeddings(word_vectors_cache_file, vocab, self.embeddings,
[] if "train" in set_folder else self.unk_term)
def regularize_loss(self, loss):
@@ -141,7 +142,7 @@ class Trainer(object):
labels[batch_start:batch_end],
ext_feats[batch_start:batch_end],
word_vectors, vec_dim
)
)
xq, xa, x_ext_feats = batch_inputs[0]
y = batch_labels[0]
@@ -189,7 +190,7 @@ class Trainer(object):
labels[batch_start:batch_end],
ext_feats[batch_start:batch_end],
word_vectors, vec_dim
)
)
xq, xa, x_ext_feats = batch_inputs[0]
@@ -200,7 +201,7 @@ class Trainer(object):
# logger.debug('batch_loss {}, batch_correct {}'.format(batch_loss, batch_correct))
train_loss += batch_loss
# train_correct += batch_correct
if debug_single_batch:
if debug_single_batch:
break
# logger.info('train_correct {}'.format(train_correct))
@@ -225,6 +226,8 @@ class Trainer(object):
input_tensor = torch.zeros(1, vec_dim, len(terms))
input_tensor[0] = torch.transpose(word_embeddings, 0, 1)
if self.cuda and torch.cuda.is_available():
input_tensor = input_tensor.cuda()
return input_tensor
@@ -240,12 +243,17 @@ class Trainer(object):
# - would zero endings work for other smaller sentences?
y = torch.LongTensor(batch_size).type(torch.LongTensor)
if self.cuda and torch.cuda.is_available():
y = y.cuda()
tensorized_inputs = []
for i in range(len(batch_ques)):
xq = Variable(self.make_input_matrix(batch_ques[i], word_vectors, vec_dim))
xs = Variable(self.make_input_matrix(batch_sents[i], word_vectors, vec_dim))
ext_feats = Variable(torch.FloatTensor(batch_ext_feats[i]))
ext_feats = torch.FloatTensor(batch_ext_feats[i])
if self.cuda and torch.cuda.is_available():
ext_feats = ext_feats.cuda()
ext_feats = Variable(ext_feats)
ext_feats = torch.unsqueeze(ext_feats, 0)
y[i] = batch_labels[i]
tensorized_inputs.append((xq, xs, ext_feats))
sm_cnn/utils.py
+6 -5
View File
@@ -1,11 +1,12 @@
# file input output
import os
import sys
import re
import string
import numpy as np
import torch
from gensim.models.keyedvectors import KeyedVectors
from nltk.corpus import stopwords
from nltk.stem import PorterStemmer
import numpy as np
# logging setup
import logging
@@ -155,7 +156,7 @@ def read_in_dataset(dataset_folder, set_folder, stop_punct=False, dash_split=Fal
for term in sentence.split():
vocab_set.add(term)
vocab = list(vocab_set)
return [questions, sentences, labels, max(len_q_list), max(len_s_list), vocab]