BlueSTARR/NeuralConfig.py at main · bmajoros/BlueSTARR · GitHub

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
#!/usr/bin/env python
#=========================================================================
# This is OPEN SOURCE SOFTWARE governed by the Gnu General Public
# License (GPL) version 3, as described at www.opensource.org.
# Copyright (C)2021 William H. Majoros <bmajoros@alumni.duke.edu>
#=========================================================================
from __future__ import (absolute_import, division, print_function,
   unicode_literals, generators, nested_scopes, with_statement)
from builtins import (bytes, dict, int, list, object, range, str, ascii,
   chr, hex, input, next, oct, open, pow, round, super, filter, map, zip)
# The above imports should allow this program to run in both Python 2 and
# Python 3.  You might need to update your version of module "future".
import sys
import ProgramName
from ConfigFile import ConfigFile

class NeuralConfig:
    def __init__(self,filename):
        config=ConfigFile(filename)

        # Slurm parameters
        #self.NumRestarts=int(config.lookupOrDie("NumRestarts"))

        # Parameters pertaining to the input
        self.RevComp=int(config.lookupOrDie("RevComp"))

        # Data
        self.MaxTrain=config.lookup("MaxTrain")
        if(self.MaxTrain is None or int(self.MaxTrain)<=0):
            self.MaxTrain=sys.maxsize
        else: self.MaxTrain=int(self.MaxTrain)
        self.MaxTest=config.lookup("MaxTest")
        if(self.MaxTest is None or int(self.MaxTest)<=0):
            self.MaxTtest=sys.maxsize
        else: self.MaxTest=int(self.MaxTest)
        self.ShouldTest=config.lookup("ShouldTest")
        if(self.ShouldTest is None): self.ShouldTest=0
        else: self.ShouldTest=int(self.ShouldTest)

        # Convolutional parameters
        self.NumConv=int(config.lookupOrDie("NumConvLayers"))
        self.NumKernels=\
            [int(x) for x in config.lookupListOrDie("NumKernels")]
        self.checkSize("NumKernels",self.NumKernels,self.NumConv)
        self.KernelSizes=\
            [int(x) for x in config.lookupListOrDie("KernelSizes")]
        self.checkSize("KernelSizes",self.KernelSizes,self.NumConv)
        self.ConvResidualSkip=int(config.lookupOrDie("ConvResidualSkip"))
        self.ConvPoolSize=int(config.lookupOrDie("ConvPoolSize"))
        self.DilationFactor=int(config.lookupOrDie("DilationFactor"))
        self.ConvPad=config.lookupOrDie("ConvPad")
        self.ConvDropout=int(config.lookupOrDie("ConvDropout"))

        # Attention parameters
        self.NumAttn=int(config.lookupOrDie("NumAttentionLayers"))
        self.AttnHeads=\
            [int(x) for x in config.lookupListOrDie("AttentionHeads")]
        self.checkSize("AttentionHeads",self.AttnHeads,self.NumAttn)
        self.AttnKeyDim=\
            [int(x) for x in config.lookupListOrDie("AttentionKeyDim")]
        self.checkSize("AttentionKeyDim",self.AttnKeyDim,self.NumAttn)
        self.AttnResidualSkip=int(config.lookupOrDie("AttentionResidualSkip"))

        # Global pool & flatten
        self.GlobalMaxPool=int(config.lookupOrDie("GlobalMaxPool"))
        self.GlobalAvePool=int(config.lookupOrDie("GlobalAvePool"))
        self.Flatten=config.lookup("Flatten")
        if(self.Flatten is None or self.Flatten==""): self.Flatten=0
        else: self.Flatten=int(self.Flatten)

        # Dense parameters
        self.NumDense=int(config.lookupOrDie("NumDense"))
        self.DenseSizes=[int(x) for x in config.lookupListOrDie("DenseSizes")]
        self.checkSize("DenseSizes",self.DenseSizes,self.NumDense)

        # Parameters pertaining to output and loss
        self.Tasks=config.lookupList("Tasks")
        self.TaskWeights=config.lookupList("TaskWeights")
        self.useCustomLoss=int(config.lookupOrDie("UseCustomLoss"))
        self.useCustomLoss=self.useCustomLoss!=0
        if(self.useCustomLoss):
            print("Using negative log likelihood loss function")
        else: print("Using MSE loss function with naive estimator")

        # Training hyperparameters
        self.Verbose=int(config.lookupOrDie("Verbose"))
        self.BatchSize=int(config.lookupOrDie("BatchSize"))
        self.Epochs=int(config.lookupOrDie("Epochs"))
        self.EarlyStop=int(config.lookupOrDie("EarlyStop"))
        self.DropoutRate=float(config.lookupOrDie("DropoutRate"))
        self.LearningRate=float(config.lookupOrDie("LearningRate"))

    def checkSize(self,name,array,size):
        if(len(array)<size):
            raise Exception(name+" wrong size: expecting "+str(size)+
                            ", found "+str(len(array)))

    def dump(self):
        print("GlobalMaxPool=",self.GlobalMaxPool,sep="")
        print("GlobalAvePool=",self.GlobalAvePool,sep="")
        #print("NumRestarts=",self.NumRestarts,sep="")
        print("ConvResidualSkip=",self.ConvResidualSkip,sep="")
        print("ConvPad=",self.ConvPad,sep="")
        print("DilationFactor=",self.DilationFactor,sep="")
        print("ConvPoolSize=",self.ConvPoolSize,sep="")
        print("NumDense=",self.NumDense,sep="")
        print("DenseSizes=",self.DenseSizes,sep="")
        print("NumConv=",self.NumConv,sep="")
        print("KernelSizes=",self.KernelSizes,sep="")
        print("NumKernels=",self.NumKernels,sep="")
        print("NumAttn=",self.NumAttn,sep="")
        print("AttnHeads=",self.AttnHeads,sep="")
        print("AttnKeyDim=",self.AttnKeyDim,sep="")
        print("AttnResidualSkip=",self.AttnResidualSkip,sep="")
        print("Verbose=",self.Verbose,sep="")
        print("RevComp=",self.RevComp,sep="")
        print("BatchSize=",self.BatchSize,sep="")
        print("Epochs=",self.Epochs,sep="")
        print("EarlyStop=",self.EarlyStop,sep="")
        print("DropoutRate=",self.DropoutRate,sep="")
        print("LearningRate=",self.LearningRate,sep="")
        print("ConvDropout=",self.ConvDropout,sep="")

#=========================================================================
# main()
#=========================================================================
#if(len(sys.argv)!=2):
#    exit(ProgramName.get()+" <in.config>\n")
#(infile,)=sys.argv[1:]
#config=NeuralConfig(infile)
#config.dump()