log_path.flush()
# sys.exit()
p = sub.Popen(cmd.split(), stderr=sub.PIPE, stdout=sub.PIPE)
while p.poll() == None:
out = p.stdout.read(1)
log_path.write(out.decode('utf-8'))
log_path.flush()
log_path.close()
if __name__ == '__main__':
p = argparse.ArgumentParser()
p.add_argument('-c', '--conf', help='Configuration file.')
args = p.parse_args()
c = ConfigFile(args.conf)
log_f = sys.stdout
if c.get('log_path'):
log_f = open(c.get('log_path'), 'w', encoding='utf-8')
train(c['train_path'], c['test_path'], c['constraint_path'], c['model_path'], log_f)
write_and_eval(c['test_path'], c['model_path'], c['output_path'])
pt = StanfordPOSTagger(tagger_model, tagger_jar)
return pt.tag(string)
if __name__ == '__main__':
p = OptionParser()
p.add_option('-c', '--conf', help='configuration file')
opts, args = p.parse_args(sys.argv)
errors = require_opt(opts.conf, "You must specify a configuration file with -c or --conf", True)
if errors:
p.print_help()
sys.exit()
c = ConfigFile(opts.conf)
# Set up the log path
logpath = c.get('log_path')
log_f = sys.stdout
if logpath:
logdir = os.makedirs(os.path.dirname(logpath), exist_ok=True)
log_f = open(c.get('log_path'), 'w', encoding='utf-8')
# Now do the testing and training
train_postagger(c['train_file'],
c['model'],
c['delimeter'])
test_postagger(c['test_file'],
c['model'],
c['out_file'],
for name, featlist in combos:
# RESET
for other_title, other_feat in other_feats:
c[other_feat] = False
# Add combo features
for feat in featlist:
c[feat] = True
train, test = test_feature(**c)
performance[name] = (train, test)
for feat in performance:
print('%s,%s,%s' % (feat, performance[feat][0], performance[feat][1]))
if __name__ == '__main__':
c = ConfigFile('/Users/rgeorgi/Dropbox/code/eclipse/dissertation/conf/classification/feature_ablation.conf')
if c.get('posdict'):
c['posdict'] = pickle.load(open(c.get('posdict'), 'rb'))
m = MalletMaxentTrainer()
c['maxent_model'] = m
ablation(**c)
#c['feat_align'] = True
#print(test_feature(**c))
