if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('output_folder', help="folder to write results to")
parser.add_argument(
'--turian_embeddings',
default='/home/dfried/data/embeddings-scaled.EMBEDDING_SIZE=50.txt.gz')
parser.add_argument('--semeval_root',
help="folder containing semeval data",
default="/home/dfried/code/semeval")
args = parser.parse_args()
turian_embeddings = read_turian_embeddings(args.turian_embeddings)
analogy_fn, choose_best = make_analogy_fns(turian_embeddings)
def semeval_path(suffix):
return os.path.join(args.semeval_root, suffix)
sets_by_folder = {
# semeval_path('Training'): "10a 1a 2c 2h 3a 3c 4c 5d 5i 7a".split(),
semeval_path('Testing'):
"1b 1c 1d 1e 2a 2b 2d 2e 2f 2g 2i 2j 3b 3d 3e 3f 3g 3h 4a 4b 4d 4e 4f 4g 4h 5a 5b 5c 5e 5f 5g 5h 6a 6b 6c 6d 6e 6f 6g 6h 7b 7c 7d 7e 7f 7g 7h 8a 8b 8c 8d 8e 8f 8g 8h 9a 9b 9c 9d 9e 9f 9g 9h 9i 10b 10c 10d 10e 10f"
.split()
}
def nlm_scaled_path(suffix):
return os.path.join(args.output_folder, 'ModelScaled-%s.txt' % suffix)
def turker_scaled_path(suffix):
scores = [cosine_similarity(reference_analogy, other) for other in other_analogies]
return list(reversed(sorted(zip(scores, other_pairs))))
return analogy_fn, choose_best
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('output_folder', help="folder to write results to")
parser.add_argument('--turian_embeddings', default='/home/dfried/data/embeddings-scaled.EMBEDDING_SIZE=50.txt.gz')
parser.add_argument('--semeval_root', help="folder containing semeval data", default="/home/dfried/code/semeval")
args = parser.parse_args()
turian_embeddings = read_turian_embeddings(args.turian_embeddings)
analogy_fn, choose_best = make_analogy_fns(turian_embeddings)
def semeval_path(suffix):
return os.path.join(args.semeval_root, suffix)
sets_by_folder = {
# semeval_path('Training'): "10a 1a 2c 2h 3a 3c 4c 5d 5i 7a".split(),
semeval_path('Testing'): "1b 1c 1d 1e 2a 2b 2d 2e 2f 2g 2i 2j 3b 3d 3e 3f 3g 3h 4a 4b 4d 4e 4f 4g 4h 5a 5b 5c 5e 5f 5g 5h 6a 6b 6c 6d 6e 6f 6g 6h 7b 7c 7d 7e 7f 7g 7h 8a 8b 8c 8d 8e 8f 8g 8h 9a 9b 9c 9d 9e 9f 9g 9h 9i 10b 10c 10d 10e 10f".split()
}
def nlm_scaled_path(suffix):
return os.path.join(args.output_folder, 'ModelScaled-%s.txt' % suffix)
def turker_scaled_path(suffix):
return os.path.join(args.output_folder, 'TurkerScaled-%s.txt' % suffix)
