File size: 7,486 Bytes
eb90a8a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 | """
Module of functions defining program parameters
"""
import sys
import argparse
def makeArgParserCodeML(description, task = "classification"):
"""
Make program arguments parser for machine learning
for analysis of source code problems
Parameters:
- description -- string describing the parser
- task -- task for which dataset is used:
either 'classification' or 'similarity'
Returns: constructed argument parser
"""
parser = argparse.ArgumentParser(description = description)
parser.add_argument("dataset",
type=str, help="dataset directory")
parser.add_argument("--min_solutions", type=int, default=8,
help = "min number of problem solutions")
parser.add_argument("--problems", default=None, type=int,
help = "max number of problems to load")
parser.add_argument("--short_code", default = 4, type=int,
help = "shortest code to load")
parser.add_argument("--long_code", default = None, type=int,
help = "longest code to load")
parser.add_argument("--testpart", default=0, type=float,
help="part of whole dataset for testing")
parser.add_argument("--valpart", default=0.2, type=float,
help="fraction of training dataset for validation")
parser.add_argument("--batch", default=32, type=int,
help="batch size")
parser.add_argument("--seed_ds", type=int, default=101,
help = "seed for making randomized dataset")
parser.add_argument("--seed_model", type=int, default=101,
help = "seed for making randomized model")
parser.add_argument("--dnn", default="basic",
type=str, help="dnn model")
parser.add_argument('--progress', default=1, type=int,
choices=[0, 1, 2],
help="mode of keras training progress bar")
parser.add_argument("--epochs", default=32, type=int,
help="number of train epochs")
parser.add_argument("--patience", default=32, type=int,
help="number of epoch to train without improvements")
parser.add_argument("--dense", type=int, nargs='*',
help="sizes of dense layers")
parser.add_argument("--regul_dense_only", action="store_true",
default=False,
help="regulirize only dense layers")
parser.add_argument("--history", type=str,
default="./train_history.pcl",
help="filename to write history of training")
parser.add_argument("--optimizer", type=str, default="rmsprop",
choices=["sgd", "rmsprop", "adam", "adadelta",
"adagrad", "adamax", "nadam", "ftrl"],
help="training optimizer")
parser.add_argument("--ckpt_dir", default = None,
type=str, help="checkpoint directory")
if task == "similarity":
parser.add_argument("--validation", type=str,
default="same", choices=["same", "different"],
help="validation with same or different " +
"as training problems")
parser.add_argument("--valsize", default=5000, type=int,
help="size of validation dataset")
parser.add_argument("--trainsize", default=20000, type=int,
help="size of training dataset")
parser.add_argument("--similpart", default=0.5, type=float,
help="fraction of all samples, " +
"representing pairs of similar source code")
elif task == "classification":
parser.add_argument("--balanced_split", action="store_true",
default=False,
help="balanced split of problem solutions " +
"for training, validation and testing")
else:
sys.exit(f"Wrong machine learning task {task}. " +
"It should be either 'classification' or 'similarity'")
return parser
def addSeqTokensArgs(parser):
"""
Add arguments defining DNN processing sequences of tokens
Parameters:
- parser -- argument parser as an object of ArgumentParser
to add arguments
Returns: constructed argument parser
"""
parser.add_argument('--seq_len', default=None, type=int,
help='maximum lengths of token sequence')
parser.add_argument("--coding", type=str, default="categorical",
choices=["categorical", "trainable", "one_hot"],
help="dnn input data coding: " +
"either one-hot or categorical or trainable")
parser.add_argument("--embed", default=None, type=int,
help="size of embedding vectors; " +
"number of tokens if it is not defined" )
parser.add_argument("--kernels", required=True,
type=int, nargs='+',
help="width of convolution kernels")
parser.add_argument("--filters", required=True,
type=int, nargs='+',
help="number of convolution filters")
parser.add_argument("--strides", default = None,
type=int, nargs='+',
help="strides of convolutions")
parser.add_argument("--conv_act", type=str,
default= None, choices=["relu", "sigmoid", "tanh"],
help="activation of convolutional layers")
parser.add_argument("--pool", type=str,
default="max", choices=["max", "aver"],
help="operation of pooling layer")
parser.add_argument("--dropout", default=0, type=float,
help="rate of dropout")
return parser
def addRegularizationArgs(parser):
"""
Add arguments defining regularization
Parameters:
- parser -- argument parser as an object of ArgumentParser
to add arguments
Returns: constructed argument parser
"""
parser.add_argument("--l1", default=0, type=float,
help="l1 regularization")
parser.add_argument("--l2", default=0, type=float,
help="l2 regularization")
return parser
def parseArguments(parser):
"""
Parse program arguments
Parameters:
- parser -- argument parser as an object of ArgumentParser
to add arguments
Returns: object with program arguments
"""
args = parser.parse_args()
print("Parameter settings used:")
for k,v in sorted(vars(args).items()):
print("{}: {}".format(k,v))
return args
def checkConvolution(args):
"""
Check definition of convolution by program arguments
Parameters:
- args object with program arguments
"""
if len(args.kernels) != len(args.filters):
print("Error: Numbers of convolution kernels and filters are different")
sys.exit(1)
if args.strides and len(args.strides) != len(args.filters):
print("Error: Numbers of convolution kernels and strides are different")
sys.exit(1)
|