def train_model_for_debug():
feature_extraction_config = {
'text_to_vector_uni_vocabulary': 'vocabularies/text_to_vector_uni_vocabulary_10.txt',
'text_to_vector_bi_vocabulary': 'vocabularies/text_to_vector_bi_vocabulary.txt',
'tf_idf_vector': False,
'counter_vector': True,
'binary_vector': False,
'best_representing_words_list': 'vocabularies/service_best_words_custom.txt',
'surrounding_words': True,
'polarity_vocabulary': 'vocabularies/polarity_words.txt',
'positive_words_count': True,
'negative_words_count': True,
'polarity_count': True,
'parts_of_speech': True,
'uni_gram': True,
'bi_gram': False,
'not_count': False,
'remove_stop_words': False
}
class_map = {0: 0,
1: 1,
# 2: 2,
3: 3}
data, target, original_text = train_model.prepare_data(src_path=src_path,
data_field=data_field,
target_field=target_field,
class_map=class_map,
balance_classes=False,
randomize=False,
feature_config=feature_extraction_config)
clf = SVC(kernel='linear')
train_model.test_model(clf, data, target, original_text, 'service_quality_mistakes.csv', verbose=True)
def main(model_path, test_data_path):
model = keras.models.load_model(model_path)
x_test_data, y_test_data, classes = get_file_data(test_data_path)
test_model(model, x_test_data, y_test_data)
import os
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID" # see issue #152
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
from train_model import test_model
from keras.models import load_model
import pickle
# Parameters
# to_load = "nietzsche"
to_load = "obama2"
length = 200
outfile = open("generated/{}_long.txt".format(to_load), "w")
model = load_model("saved_models/{}.h5".format(to_load))
char_to_indices = pickle.load(
open("saved_models/{}c2i.p".format(to_load), "rb"))
indices_to_char = pickle.load(
open("saved_models/{}i2c.p".format(to_load), "rb"))
for temperature in [0.2, 0.4, 0.6]:
generated_string = test_model(model=model, char_to_indices=char_to_indices, indices_to_char=indices_to_char, \
seed_string=" ", temperature=0.3, test_length=length)
output = "Temperature: {} Generated string: {}".format(
temperature, generated_string)
print(output)
outfile.write(output + "\n")
outfile.flush()
outfile.close()
from keras.models import load_model
import pickle
# Parameters
origin = "full_pandas_with_import"
model_name = "full_pandas_with_import_usize512_maxlen240_numlayers3_dropout0"
weights = "weights-13-0.94.h5"
length = 10000
keep_chars = 50
model = load_model("saved_models/{}/{}".format(model_name, weights))
char_to_indices = pickle.load(
open("saved_models/{}_c2i.p".format(origin), "rb"))
indices_to_char = pickle.load(
open("saved_models/{}_i2c.p".format(origin), "rb"))
for temperature in [0.1, 0.35, 0.5, 1, 2]:
generated_string = test_model(model=model,
char_to_indices=char_to_indices,
indices_to_char=indices_to_char,
seed_string="def ",
temperature=temperature,
test_length=length,
keep_chars=keep_chars)
with open("generated/{}.txt".format(model_name), "a",
encoding="utf-8") as outfile:
output = "Temperature: {} Generated string: \n{}\n".format(
temperature, generated_string)
print(output)
outfile.write(output + "\n")
text_to_vector_vocabulary = vocabularies.read_vocabulary('text_to_vector_vocabulary.txt')
best_representing_words = vocabularies.read_best_words_list('clean_best_words.txt')
polarity_vocabulary = vocabularies.read_polarity_vocabulary('polarity_words.txt')
feature_extraction_config = {
'text_to_vector_vocabulary': text_to_vector_vocabulary,
'tf_idf_vector': False,
'counter_vector': True,
'binary_vector': False,
'best_representing_words_list': best_representing_words,
'surrounding_words': True,
'polarity_vocabulary': polarity_vocabulary,
'positive_words_count': True,
'negative_words_count': True,
'polarity_count': True,
'parts_of_speech': False
}
data, target = train_model.prepare_data(src_path=data_path,
data_field=data_field,
target_field=target_field,
class_map=class_map,
balance_classes=True,
randomize=False,
feature_config=feature_extraction_config)
clf = SVC(kernel='linear', degree=3)
#clf = tree.DecisionTreeClassifier(max_depth=5)
train_model.test_model(clf, data, target)