代码拉取完成,页面将自动刷新
import numpy as np
import logging, logging.config
import torch
from torch.nn.init import xavier_normal_
from torch.nn import Parameter
np.set_printoptions(precision=4)
def set_gpu(gpus):
"""
Sets the GPU to be used for the run
Parameters
----------
gpus: List of GPUs to be used for the run
Returns
-------
"""
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = gpus
def get_logger(name, log_dir, config_dir):
"""
Creates a logger object
Parameters
----------
name: Name of the logger file
log_dir: Directory where logger file needs to be stored
config_dir: Directory from where log_config.json needs to be read
Returns
-------
A logger object which writes to both file and stdout
"""
config_dict = json.load(open(config_dir + 'log_config.json'))
config_dict['handlers']['file_handler']['filename'] = log_dir + name.replace('/', '-')
logging.config.dictConfig(config_dict)
logger = logging.getLogger(name)
std_out_format = '%(asctime)s - [%(levelname)s] - %(message)s'
consoleHandler = logging.StreamHandler(sys.stdout)
consoleHandler.setFormatter(logging.Formatter(std_out_format))
logger.addHandler(consoleHandler)
return logger
def get_combined_results(left_results, right_results):
results = {}
count = float(left_results['count'])
results['left_mr'] = round(left_results['mr'] / count, 5)
results['left_mrr'] = round(left_results['mrr'] / count, 5)
results['right_mr'] = round(right_results['mr'] / count, 5)
results['right_mrr'] = round(right_results['mrr'] / count, 5)
results['mr'] = round((left_results['mr'] + right_results['mr']) / (2 * count), 5)
results['mrr'] = round((left_results['mrr'] + right_results['mrr']) / (2 * count), 5)
for k in range(10):
results['left_hits@{}'.format(k + 1)] = round(left_results['hits@{}'.format(k + 1)] / count, 5)
results['right_hits@{}'.format(k + 1)] = round(right_results['hits@{}'.format(k + 1)] / count, 5)
results['hits@{}'.format(k + 1)] = round(
(left_results['hits@{}'.format(k + 1)] + right_results['hits@{}'.format(k + 1)]) / (2 * count), 5)
return results
def get_param(shape):
param = Parameter(torch.Tensor(*shape));
xavier_normal_(param.data)
return param
def com_mult(a, b):
r1, i1 = a[..., 0], a[..., 1]
r2, i2 = b[..., 0], b[..., 1]
return torch.stack([r1 * r2 - i1 * i2, r1 * i2 + i1 * r2], dim=-1)
def conj(a):
a[..., 1] = -a[..., 1]
return a
def cconv(a, b):
return torch.irfft(com_mult(torch.rfft(a, 1), torch.rfft(b, 1)), 1, signal_sizes=(a.shape[-1],))
def ccorr(a, b):
return torch.irfft(com_mult(conj(torch.rfft(a, 1)), torch.rfft(b, 1)), 1, signal_sizes=(a.shape[-1],))
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。