import torch
import numpy as np
from torch.nn.parallel.data_parallel import DataParallel
from source.utils import get_logger, count_parameters_in_M
from source.losses import *
from source.autoencoders import *
from source.auditors import *
logger = get_logger(__name__)
class _CustomDataParallel(DataParallel):
"""
DataParallel distribute batches across multiple GPUs
https://github.com/pytorch/pytorch/issues/16885
"""
def __init__(self, model):
super(_CustomDataParallel, self).__init__(model)
def __getattr__(self, name):
try:
return super(_CustomDataParallel, self).__getattr__(name)
except AttributeError:
return getattr(self.module, name)
class Model(object):
"""
from source.utils import get_logger
logger = get_logger()
# -----------------------------------------------------------------------------------------------------------
# Function to check if reference & data files are non-empty and within wavelength range
def Ref_Data_is_valid(ref_df,
data_df,
startNM,
stopNM,
range_no
):
"""
:param ref_df: dataFrame of reference diode
:param data_df: dataFrame of measurement data
:param startNM: start wavelength [float or int]
:param stopNM: stop wavelength [float or int]
:param range_no: number of measurment range to plot / reference correct [int]
:return: True or False
"""
if len(ref_df) != 0 and len(data_df) != 0:
if startNM >= data_df['Wavelength'][0] and \
startNM >= ref_df['Wavelength'][0] and \
stopNM <= data_df['Wavelength'][int(len(data_df['Wavelength'])) - 1] and \
stopNM <= ref_df['Wavelength'][int(len(ref_df['Wavelength'])) - 1]:
return True