posCount += 1
if entry.parent_id == entry.pred_parent_id and entry.pred_relation == entry.relation:
lasCount += 1
if entry.parent_id == entry.pred_parent_id:
uasCount += 1
count += 1
print("---\nLAS accuracy:\t%.2f" % (lasCount * 100 / count))
print("UAS accuracy:\t%.2f" % (uasCount * 100 / count))
print("POS accuracy:\t%.2f" % (posCount * 100 / count))
print("POS&LAS:\t%.2f" % (poslasCount * 100 / count))
score = poslasCount * 100 / count
if score >= highestScore:
parser.Save(
os.path.join(args.output,
os.path.basename(args.model)))
highestScore = score
print("POS&LAS of the previous saved model: %.2f" %
(highestScore))
else:
print('Extracting vocabulary')
words, w2i, c2i, pos, rels = utils.vocab(args.conll_train)
with open(os.path.join(args.output, args.params),
'wb') as paramsfp:
pickle.dump((words, w2i, c2i, pos, rels, args),
paramsfp,
protocol=2)
parser.Train(args.conll_train)
devpath = os.path.join(args.output, 'dev_epoch_' + str(epoch+1) + '.conll')
lysfastparse.bcovington.utils_bcovington.write_conll(devpath, parser.Predict(args.conll_dev))
lysfastparse.utils.dump_lookup_extra_into_conll(devpath, lookup_conll_data)
lysfastparse.utils.transform_to_single_root(devpath)
print 'Executing conll17_eval'
os.system('python '+config[YAML_CONLL17_EVAL]+' '+args.conll_dev + ' '+devpath+ ' > ' + devpath + '.txt ')
with codecs.open(devpath+".txt") as f_devpath:
content = f_devpath.readlines()
las_lines = [l for l in content if
l.startswith("LAS F1 Score")]
if len(las_lines) != 1:
warnings.warn("Cannot determine LAS F1 Score from file")
else:
las = float(las_lines[0].split(":")[1])
log_results_file.write('\t'.join([args.output.rsplit("/",1)[1],str(epoch),"\n".join(content)]))
print 'Finished predicting dev'
#Only saves the best performing model
if las >= better_las:
parser.Save(os.path.join(args.output, args.model))
better_las = las
log_results_file.close()
