def __init__(self, args, label2index_map, input_size, paths, config): self.batch_size = args.batch_size self.epoch_num = args.epoch self.optimier = args.optimizer self.hidden_dim1 = args.hidden_dim1 self.hidden_dim2 = args.hidden_dim2 self.hidden_dim3 = args.hidden_dim3 self.dropout_keep_prob = args.dropout self.beta = args.beta self.lr = args.lr self.clip_grad = args.clip self.optimizer = args.optimizer self.test_data_path = args.test_data self.tag2label = label2index_map self.num_tags = len(label2index_map) self.input_size = input_size self.config = config self.model_path = paths['model_path'] self.summary_path = paths['summary_path'] self.logger = base_util.get_logger(paths['log_path']) self.result_path = paths['result_path']
time.time())) if args.mode == 'train' else args.demo_model output_path = os.path.join(args.result, "dnn_save", tag_timestamp) if not os.path.exists(output_path): os.makedirs(output_path) summary_path = os.path.join(output_path, "summaries") paths['summary_path'] = summary_path if not os.path.exists(summary_path): os.makedirs(summary_path) model_path = os.path.join(output_path, "checkpoints/") if not os.path.exists(model_path): os.makedirs(model_path) ckpt_prefix = os.path.join(model_path, "model") paths['model_path'] = ckpt_prefix result_path = os.path.join(output_path, "results") paths['result_path'] = result_path if not os.path.exists(result_path): os.makedirs(result_path) log_path = os.path.join(result_path, "log.txt") paths['log_path'] = log_path get_logger(log_path).info(str(args)) os.environ['CUDA_VISIBLE_DEVICES'] = '1' _, label_list = load_label2index() print(label_list) # training model train_path = os.path.join(args.train_data, 'train_modified.csv') test_path = os.path.join(args.test_data, 'test_modified.csv') if args.mode == 'train': ids, train_data = read_corpus(train_path, shuffle=False) print("train data: {}".format(len(train_data))) train = train_data[:1000] val = train_data[1000:1050] input_size = len(train.columns) - 1 print('input_size', input_size)
import pandas as pd import numpy as np from sklearn import preprocessing from code.util.base_util import get_logger from code.util.base_util import pickle_load from code.util.base_util import pickle_dump from code.util.base_util import timer import os import pickle LABEL = 'current_service' ID = 'user_id' log = get_logger() category_list = [ 'gender', 'service_type', 'is_mix_service', 'contract_type', 'net_service', 'complaint_level', 'age_group', 'ctims' ] def __get_dir(dir): if not os.getcwd().endswith('code'): return '../' + dir return dir def base_data_prepare(age2group=True, one_hot=True): df_train = pd.read_csv(__get_dir('../origin_data/train.csv')) df_test = pd.read_csv(__get_dir('../origin_data/test.csv'))
import pandas as pd import numpy as np from sklearn.metrics import f1_score from sklearn.model_selection import KFold, StratifiedKFold from code.util.base_util import timer import os from code import base_data_process import tensorflow as tf from code.util import base_util log = base_util.get_logger() ID_COLUMN_NAME = 'user_id' LABEL_COLUMN_NAME = 'current_service' def nn_model(df_train, df_test): pass class FeatureNN(): def __init__(self, x_train, y_train, x_val, y_val, epoch=10, batch_size=30): self.epoch = epoch