-
Notifications
You must be signed in to change notification settings - Fork 1
/
acc_variation_plot.py
211 lines (171 loc) · 7.08 KB
/
acc_variation_plot.py
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
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
# -*- coding: utf-8 -*-
"""
Created on Tue Apr 5 15:48:26 2022
@author: Varad Srivastava
"""
import pickle
import matplotlib.pyplot as plt
import os.path as op
import numpy as np
from matplotlib.ticker import PercentFormatter
import csv
import pandas as pd
"""SET DIRECTORY TO MODELS TOP LEVEL FOLDER"""
model_save_dir = "F:/STUDY/MS Cognitive Science IITD/Sem 3/Cognitive Science Project I - HSD 621/final_models"
def plot_var(train_task, train_setting, model_name):
print(train_task, train_setting, model_name, sep=" ")
if train_task == "Binary Classifier":
with open(op.join(model_save_dir, train_task, train_setting, model_name,
model_name + '_fullGram_intdiff_acc.pickle'), 'rb') as handle:
data = pickle.load(handle)
elif train_task == "Language Model":
if model_name == "DECAY":
model_name = "DRNN"
with open(op.join(model_save_dir, train_task, train_setting, model_name,
model_name + '_intdiff_acc.pickle'), 'rb') as handle:
data = pickle.load(handle)
print("Loaded Data")
print(data)
#FOR 0 attractor sentences
x = [0,1,2,3,">3"]
#sort the y list according to the i[0]
if train_task == "Binary Classifier":
a = 0
pre_y = [(0,a),(1,a),(2,a),(3,a),(">3",a)]
y = [data[i][0] for i in pre_y]
#y = [data[i][0] for i in data if i[1]==0]
elif train_task == "Language Model":
#pre_y = [i for i in data if i[1]==0]
a = 0
pre_y = [(0,a),(1,a),(2,a),(3,a),(">3",a)]
y = [data[i] for i in pre_y]
plt.plot(x, y, label = "0 Attractor", marker="o")
#FOR 1 attractor sentences
x = [1,2,3,">3"]
if train_task == "Binary Classifier":
a = 1
pre_y = [(1,a),(2,a),(3,a),(">3",a)]
y = [data[i][0] for i in pre_y]
#y = [data[i][0] for i in data if i[1]==1]
elif train_task == "Language Model":
#y = [data[i] for i in data if i[1]==1]
a = 1
pre_y = [(1,a),(2,a),(3,a),(">3",a)]
y = [data[i] for i in pre_y]
plt.plot(x, y, label = "1 Attractors", marker="o")
#FOR 2 attractor sentences
x = [2,3,">3"]
if train_task == "Binary Classifier":
a = 2
pre_y = [(2,a),(3,a),(">3",a)]
y = [data[i][0] for i in pre_y]
#y = [data[i][0] for i in data if i[1]==2]
elif train_task == "Language Model":
#y = [data[i] for i in data if i[1]==2]
a = 2
pre_y = [(2,a),(3,a),(">3",a)]
y = [data[i] for i in pre_y]
plt.plot(x, y, label = "2 Attractors", marker="o")
#FOR 3 attractor sentences
x = [3,">3"]
if train_task == "Binary Classifier":
a =3
pre_y = [(3,a),(">3",a)]
y = [data[i][0] for i in pre_y]
#y = [data[i][0] for i in data if i[1]==3]
elif train_task == "Language Model":
#y = [data[i] for i in data if i[1]==3]
a =3
pre_y = [(3,a),(">3",a)]
y = [data[i] for i in pre_y]
plt.plot(x, y, label = "3 Attractors", marker="o")
#FOR >3 attractor sentences
x = [">3"]
if train_task == "Binary Classifier":
a = ">3"
pre_y = [(">3",a)]
y = [data[i][0] for i in pre_y]
#y = [data[i][0] for i in data if i[1]==">3"]
elif train_task == "Language Model":
#y = [data[i] for i in data if i[1]==">3"]
a = ">3"
pre_y = [(">3",a)]
y = [data[i] for i in pre_y]
plt.plot(x, y, label = ">3 Attractors", marker="o")
plt.ylim(0.60, 1.00)
if train_task == "Language Model":
if model_name == "DRNN":
plt.ylim(0.20,1.00)
else:
plt.ylim(0.60,1.00) #0.40 for decay RNN, for others: 0.60
plt.title(train_task+": "+train_setting+" Setting "+model_name)
plt.xlabel("No. of intervening nouns")
plt.ylabel("Accuracy")
plt.grid(True)
plt.legend()
plt.show()
def deps_from_tsv(infile, limit=None):
res = []
for i, d in enumerate(csv.DictReader(open(infile, encoding='utf-8'), delimiter='\t')):
if limit is not None and i >= limit:
break
res.append({x: int(y) if y.isdigit() else y for x, y in d.items()})
return res
def plot_bydist(train_task, train_setting):
"""
Investigating whether distance plays a role in the errors
by checking variability in distance in error sentences
"""
print(train_task, train_setting, sep=" ")
if train_task == "Language Model":
model_name = "DRNN"
else:
model_name = "DECAY"
err1 = pd.read_csv(op.join(model_save_dir, train_task, train_setting, "LSTM",'incorr.tsv'),sep='\t')
err2 = pd.read_csv(op.join(model_save_dir, train_task, train_setting, "ONLSTM",'incorr.tsv'),sep='\t')
err3 = pd.read_csv(op.join(model_save_dir, train_task, train_setting, "GRU",'incorr.tsv'),sep='\t')
err4 = pd.read_csv(op.join(model_save_dir, train_task, train_setting, model_name,'incorr.tsv'),sep='\t')
for i,data in enumerate([err1, err2, err3, err4]):
dist = data["distance"]
plt.hist(dist, bins=5, weights=np.ones(len(dist)) / len(dist), range=(0,10))
plt.legend(loc='upper right')
plt.xlabel("Distance between verb and noun")
plt.ylabel("% of sentences")
if i==0:
model_name = "LSTM"
elif i==1:
model_name = "ONLSTM"
elif i==2:
model_name = "GRU"
elif i==3:
model_name = "DECAY RNN"
plt.title(train_task+": "+train_setting+" Setting "+model_name+"\n Variation in distance in error sentences")
plt.gca().yaxis.set_major_formatter(PercentFormatter(1))
plt.show()
if __name__ == "__main__":
print("Analysis Program for Accuracy variations over interveners x attractors")
train_taskn = int(input("Input type of train task (1: BC,2: LM): "))
if train_taskn==1:
train_task = "Binary Classifier"
elif train_taskn==2:
train_task = "Language Model"
train_settingn = int(input("Input type of training setting (1: Natural/2: Selective/3: Selective2 /4: Intermediate): "))
if train_settingn==1:
train_setting = "Natural"
elif train_settingn==2:
train_setting = "Selective"
elif train_settingn==3:
train_setting = "Selective 2"
elif train_settingn==4:
train_setting = "Intermediate"
model_namen = int(input("Input type of model (1: LSTM/2: ONLSTM/3: GRU/4: DECAY RNN): "))
if model_namen==1:
model_name = "LSTM"
elif model_namen==2:
model_name = "ONLSTM"
elif model_namen==3:
model_name = "GRU"
elif model_namen==4:
model_name = "DECAY"
plot_var(train_task, train_setting, model_name)
#plot_bydist(train_task, train_setting)