Skip to content

Commit

Permalink
trainable subword-unit embeddings (data)
Browse files Browse the repository at this point in the history
Lenz Furrer committed Jun 20, 2018
1 parent 4820de0 commit bd736be
Showing 2 changed files with 42 additions and 41 deletions.
4 changes: 2 additions & 2 deletions config
View file
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
[DEFAULT]
rootpath = /home/lenz/disease-normalization
timestamp = 20180620-102655
timestamp = 20180620-103131
workers = 0

[general]
@@ -26,7 +26,7 @@ embedding_voc = 10000
vectorizer_cache = True
tokenizer = whitespace
embedding_fn = ${rootpath}/data/embeddings/wvec_50_haodi-li-et-al.bin
trainable = False
trainable = True

[emb_sub]
sample_size = ${emb:sample_size}
79 changes: 40 additions & 39 deletions log
View file
Original file line number Diff line number Diff line change
@@ -1,41 +1,42 @@
2018-06-20 10:26:56,560 - The cuda backend is deprecated and will be removed in the next release (v0.10). Please switch to the gpuarray backend. You can get more information about how to switch at this URL:
2018-06-20 10:31:32,824 - The cuda backend is deprecated and will be removed in the next release (v0.10). Please switch to the gpuarray backend. You can get more information about how to switch at this URL:
https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29

2018-06-20 10:26:59,651 - 'pattern' package not found; tag filters are not available for English
2018-06-20 10:26:59,659 - loading terminology...
2018-06-20 10:26:59,972 - loading pretrained embeddings...
2018-06-20 10:26:59,973 - loading projection weights from /home/lenz/disease-normalization/data/embeddings/bpe_vectors_10000_50_w2v.txt
2018-06-20 10:27:00,835 - loaded (10257, 50) matrix from /home/lenz/disease-normalization/data/embeddings/bpe_vectors_10000_50_w2v.txt
2018-06-20 10:27:00,846 - loading vectorizer...
2018-06-20 10:27:00,957 - loading candidate generator...
2018-06-20 10:27:16,208 - preprocessing validation data...
2018-06-20 10:27:16,209 - loading corpus...
2018-06-20 10:27:16,224 - generating candidates with 0 workers...
2018-06-20 10:27:20,309 - generated 5671 pair-wise samples (11585 with duplicates)
2018-06-20 10:27:20,311 - compiling model architecture...
2018-06-20 10:27:21,473 - preprocessing training data...
2018-06-20 10:27:21,473 - loading corpus...
2018-06-20 10:27:21,667 - generating candidates with 0 workers...
2018-06-20 10:27:42,750 - generated 26308 pair-wise samples (71125 with duplicates)
2018-06-20 10:27:42,762 - training CNN...
2018-06-20 10:28:00,728 - Ranking accuracy: 0.593393
2018-06-20 10:28:07,438 - Ranking accuracy: 0.6277
2018-06-20 10:28:13,852 - Ranking accuracy: 0.631512
2018-06-20 10:28:20,425 - Ranking accuracy: 0.635324
2018-06-20 10:28:27,022 - Ranking accuracy: 0.655654
2018-06-20 10:28:33,199 - Ranking accuracy: 0.674714
2018-06-20 10:28:39,943 - Ranking accuracy: 0.682338
2018-06-20 10:28:46,831 - Ranking accuracy: 0.710292
2018-06-20 10:28:53,180 - Ranking accuracy: 0.722999
2018-06-20 10:28:59,295 - Ranking accuracy: 0.736976
2018-06-20 10:29:06,193 - Ranking accuracy: 0.739517
2018-06-20 10:29:12,656 - Ranking accuracy: 0.740788
2018-06-20 10:29:19,170 - Ranking accuracy: 0.749682
2018-06-20 10:29:25,554 - Ranking accuracy: 0.749682
2018-06-20 10:29:31,515 - Ranking accuracy: 0.753494
2018-06-20 10:29:38,366 - Ranking accuracy: 0.752224
2018-06-20 10:29:45,021 - Ranking accuracy: 0.756036
2018-06-20 10:29:51,503 - Ranking accuracy: 0.753494
2018-06-20 10:29:57,791 - Ranking accuracy: 0.754765
2018-06-20 10:29:57,791 - Epoch 00019: early stopping
2018-06-20 10:29:57,792 - done training.
2018-06-20 10:31:36,084 - 'pattern' package not found; tag filters are not available for English
2018-06-20 10:31:36,092 - loading terminology...
2018-06-20 10:31:36,403 - loading pretrained embeddings...
2018-06-20 10:31:36,403 - loading projection weights from /home/lenz/disease-normalization/data/embeddings/bpe_vectors_10000_50_w2v.txt
2018-06-20 10:31:37,271 - loaded (10257, 50) matrix from /home/lenz/disease-normalization/data/embeddings/bpe_vectors_10000_50_w2v.txt
2018-06-20 10:31:37,284 - loading vectorizer...
2018-06-20 10:31:37,416 - loading candidate generator...
2018-06-20 10:31:51,496 - preprocessing validation data...
2018-06-20 10:31:51,496 - loading corpus...
2018-06-20 10:31:51,505 - generating candidates with 0 workers...
2018-06-20 10:31:55,387 - generated 5671 pair-wise samples (11585 with duplicates)
2018-06-20 10:31:55,389 - compiling model architecture...
2018-06-20 10:31:56,393 - preprocessing training data...
2018-06-20 10:31:56,393 - loading corpus...
2018-06-20 10:31:56,587 - generating candidates with 0 workers...
2018-06-20 10:32:14,387 - generated 26308 pair-wise samples (71125 with duplicates)
2018-06-20 10:32:14,409 - training CNN...
2018-06-20 10:32:37,062 - Ranking accuracy: 0.491741
2018-06-20 10:32:46,149 - Ranking accuracy: 0.550191
2018-06-20 10:32:55,435 - Ranking accuracy: 0.645489
2018-06-20 10:33:04,643 - Ranking accuracy: 0.664549
2018-06-20 10:33:13,936 - Ranking accuracy: 0.669632
2018-06-20 10:33:22,963 - Ranking accuracy: 0.673443
2018-06-20 10:33:32,155 - Ranking accuracy: 0.691233
2018-06-20 10:33:41,079 - Ranking accuracy: 0.696315
2018-06-20 10:33:50,323 - Ranking accuracy: 0.715375
2018-06-20 10:33:59,614 - Ranking accuracy: 0.729352
2018-06-20 10:34:08,834 - Ranking accuracy: 0.733164
2018-06-20 10:34:18,368 - Ranking accuracy: 0.747141
2018-06-20 10:34:27,433 - Ranking accuracy: 0.747141
2018-06-20 10:34:36,540 - Ranking accuracy: 0.750953
2018-06-20 10:34:45,699 - Ranking accuracy: 0.750953
2018-06-20 10:34:54,978 - Ranking accuracy: 0.752224
2018-06-20 10:35:04,327 - Ranking accuracy: 0.750953
2018-06-20 10:35:13,851 - Ranking accuracy: 0.756036
2018-06-20 10:35:23,100 - Ranking accuracy: 0.756036
2018-06-20 10:35:32,328 - Ranking accuracy: 0.756036
2018-06-20 10:35:32,329 - Epoch 00020: early stopping
2018-06-20 10:35:32,330 - done training.

0 comments on commit bd736be

Please sign in to comment.