Serie D Final Stage stats & predictions

Analisi Post-Match

Dopo ogni giornata di gare, forniremo un'analisi dettagliata delle partite più importanti del giorno:

Analisi Partita: A.S.D. F.C. Juventus vs S.S. Lazio Primavera

L'incontro tra Juventus e Lazio è stato caratterizzato da una fase centrale dominata dalla difesa juventina, che ha saputo contenere gli attacchi rapidi dei giovani biancocelesti.

• Tatticamente parlando,
  

• Juventus:
  

• Hanno adottato una formazione a zona molto stretta,
  

• Promuovendo il pressing alto sul portatore di palla avversario,
  

• Cercando di interrompere immediatamente le catene offensive laziste.
  

• Lazio:
  

• Hanno cercato di sfruttare i corridoi laterali,
  

• Cercando ampiezza attraverso cross dalla fascia,
  

• E tentando inserimenti rapidi dei loro attaccanti esterni.
  

In conclusione,

• Juventus:
  

• Hanno dimostrato una solidità difensiva impressionante,
  

• E hanno saputo sfruttare al meglio le occasioni create.
  

• Lazio:
  

• Hanno mostrato creatività ed entusiasmo,
  

<|file_sep|># -*- coding:utf-8 -*- from __future__ import print_function import os import sys import time import json import logging import argparse import multiprocessing import numpy as np import tensorflow as tf from src.data.dataset import Dataset from src.models.model import Model flags = tf.app.flags FLAGS = flags.FLAGS flags.DEFINE_string('dataset', 'yelp_review_polarity_csv', 'dataset name') flags.DEFINE_string('model_name', 'rnn', 'model name') flags.DEFINE_string('ckpt_dir', '/home/ubuntu/CKPT_DIR/', 'checkpoint directory') flags.DEFINE_string('data_dir', '/home/ubuntu/DATA_DIR/', 'data directory') flags.DEFINE_integer('batch_size', None, 'The number of samples contained in each batch') flags.DEFINE_integer('num_epochs', None, 'The number of epochs to train the model for') flags.DEFINE_integer('hidden_size', None, 'Number of units in the hidden layer of the RNN cell') flags.DEFINE_integer('num_layers', None, 'Number of layers in the RNN cell') flags.DEFINE_integer('max_seq_len', None, 'Maximum sequence length') flags.DEFINE_float('learning_rate', None, 'Learning rate for Adam Optimizer') # TODO add parameter parsing def train(args): """ Train function. :param args: :return: """ dataset = Dataset(args) model = Model(args) # input data pipeline train_dataset = dataset.get_dataset( mode=Dataset.TRAIN, batch_size=args.batch_size) dev_dataset = dataset.get_dataset( mode=Dataset.DEV, batch_size=args.batch_size) test_dataset = dataset.get_dataset( mode=Dataset.TEST, batch_size=args.batch_size) # build model graph train_op = model.build_graph( dataset=train_dataset, mode=Model.TRAIN) eval_op = model.build_graph( dataset=dev_dataset, mode=Model.EVAL) # create session and initialize variables with tf.Session() as sess: sess.run(tf.global_variables_initializer()) sess.run(tf.local_variables_initializer()) saver = tf.train.Saver() for epoch in range(args.num_epochs): # training loop logging.info("Epoch {}/{}".format(epoch + 1, args.num_epochs)) start_time = time.time() for step in range(dataset.num_train_batches): if step % args.log_every == args.log_every -1: loss_value = sess.run(train_op) elapsed_time = time.time() - start_time logging.info("Step {:d} of {:d}, loss={:.5f}, ({:.2f} sec/batch)" .format(step + epoch * dataset.num_train_batches, args.num_epochs * dataset.num_train_batches, loss_value[0], elapsed_time / args.log_every)) start_time = time.time() else: sess.run(train_op) # evaluation loop dev_correct_preds = [] dev_correct_labels = [] for step in range(dataset.num_dev_batches): correct_pred_values = sess.run(eval_op) dev_correct_preds.extend(correct_pred_values[0]) dev_correct_labels.extend(correct_pred_values[1]) dev_accuracy = np.mean(np.equal(dev_correct_preds, dev_correct_labels).astype(float)) logging.info("Dev accuracy={:.5f}".format(dev_accuracy)) if epoch == args.num_epochs -1: # test loop test_correct_preds = [] test_correct_labels = [] for step in range(dataset.num_test_batches): correct_pred_values = sess.run(eval_op) test_correct_preds.extend(correct_pred_values[0]) test_correct_labels.extend(correct_pred_values[1]) test_accuracy = np.mean(np.equal(test_correct_preds, test_correct_labels).astype(float)) logging.info("Test accuracy={:.5f}".format(test_accuracy)) if epoch % args.save_every == args.save_every -1: checkpoint_path = os.path.join(args.checkpoint_dir, "model_epoch_{}".format(epoch)) saver.save(sess=sess, save_path=checkpoint_path) def main(): parser = argparse.ArgumentParser() parser.add_argument('--config_file', type=str, help='Path to the configuration file') parser.add_argument('--log_level', type=str, default='info', choices=['debug', 'info', 'warn', 'error'], help='Logging level') args_parsed = parser.parse_args() logging.basicConfig(level=args_parsed.log_level.upper(), format='%(asctime)s %(levelname)s %(message)s', datefmt='%Y-%m-%d %H:%M:%S') if not os.path.isfile(args_parsed.config_file): logging.error("Configuration file does not exist") sys.exit(1) with open(args_parsed.config_file) as f: config_dict = json.load(f) FLAGS.batch_size = config_dict['batch_size'] FLAGS.num_epochs = config_dict['num_epochs'] FLAGS.hidden_size = config_dict['hidden_size'] FLAGS.num_layers = config_dict['num_layers'] FLAGS.max_seq_len = config_dict['max_seq_len'] FLAGS.learning_rate = config_dict['learning_rate'] # multiprocessing.set_start_method('spawn') # necessary when using CUDA # multiprocessing.set_start_method('forkserver') # necessary when using CUDA # processes = [] # for i in range(multiprocessing.cpu_count()): # process_i_args = dict(config_dict) # process_i_args['gpu_id'] += i # process_i_args['data_dir'] += str(i) # p_i = multiprocessing.Process(target=train, # kwargs={'args': process_i_args}) # processes.append(p_i) # for process_i in processes: # process_i.start() # for process_i in processes: # process_i.join() train(config_dict) if __name__ == '__main__': main() <|repo_name|>kchung97/NLP<|file_sep|>/src/models/model.py from __future__ import absolute_import import tensorflow as tf from src.models.rnn_model import RNNModel class Model(object): TRAIN='train' EVAL='eval' def __init__(self,args): self.args=args if self.args.model_name=='rnn': self.model=RNNModel(self.args)