# Step 2 - train perceptron for however many rounds on training data
training_data = open("data/engf.train")
sentence = []
poslist = []
true_tags = []
start_time = time.time()
for r in xrange(ROUNDS):
print "round:", r
line_num = 0
for line in training_data:
if line_num != 0 and line_num % 5000 == 0:
print "done", line_num, "/", 219552, time.time() - start_time
line_num += 1
line = line.strip()
if line == '':
percept.train_greedy(sentence, poslist, true_tags)
sentence = []
poslist = []
true_tags = []
else:
line = line.split(' ')
sentence.append(line[0])
poslist.append(line[1])
true_tags.append(line[-1])
training_data.seek(0)
percept.finish_round()
training_data.close()
#percept.finish_training()
# Step 3 - classify
percept.finish_training()
test_data = open("data/eng.testb")
