forked from andreamad8/ToDCL
-
Notifications
You must be signed in to change notification settings - Fork 0
/
results.txt
193 lines (165 loc) · 4.86 KB
/
results.txt
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
DST
python scorer.py --model_checkpoint runs_DST/BEST --task_type DST --all --best
Method ACC
--------- ----------------
MULTI 50.04
VANILLA $5.34 $\pm$ 4.4$
L2 $4.95 $\pm$ 4.4$
EWC $5.36 $\pm$ 4.3$
AGEM $5.17 $\pm$ 4.0$
LAML $4.03 $\pm$ 3.9$
REPLAY $39.42 $\pm$ 0.2$
ADAPTERCL $37.9 $\pm$ 0.6$
ADAPTER MEAN: 0.9819053429791685
ADAPTER STD: 0.0014473477557442552
ABLATION
python scorer.py --model_checkpoint runs_DST/MEMABLATION --task_type DST --all --ablation
Method ACC
-------- -----------------------------------------
10 26.63$\pm$1.26
50 39.41$\pm$0.28
100 43.13$\pm$0.31
500 48.22$\pm$0.53
1000 49.97$\pm$0.46
INTENT
python scorer.py --model_checkpoint runs_INTENT/BEST --task_type INTENT --all --best
Method ACC
--------- ----------------
MULTI 87.5
VANILLA $3.27 $\pm$ 0.33$
L2 $3.52 $\pm$ 0.7$
EWC $3.21 $\pm$ 0.37$
AGEM $9.74 $\pm$ 2.64$
LAML $3.73 $\pm$ 1.07$
REPLAY $76.45 $\pm$ 1.55$
ADAPTERCL $85.05 $\pm$ 0.64$
ADAPTER MEAN: 0.9803538743136059
ADAPTER STD: 0.0009368912609978562
python scorer.py --model_checkpoint runs_INTENT/MEMABLATION --task_type INTENT --all --ablation
Method ACC
-------- -----------------------------
10 57.286$\pm$3.80
50 76.446$\pm$1.55
100 81.496$\pm$0.86
500 85.91$\pm$0.55
1000 87.784$\pm$0.16
NLG
python scorer.py --model_checkpoint runs_NLG/BEST --task_type NLG --all --best
Method ACC
--------- ----------------
MULTI 3.42
VANILLA $17.07 \pm 7.9$
AGEM $39.02 \pm 8.83$
EWC $16.37 \pm 6.33$
L2 $15.0 \pm 5.5$
LAML $30.32 \pm 3.13$
ADAPTERCL $16.65 \pm 0.94$
REPLAY $6.64 \pm 0.54$
ADAPTER MEAN: 0.9398433709889806
ADAPTER STD: 0.0010745608118824044
Method ACC
--------- ----------------
MULTI 26.15
VANILLA $11.14 $\pm$ 2.83$
L2 $11.99 $\pm$ 1.41$
EWC $11.19 $\pm$ 2.37$
AGEM $5.51 $\pm$ 2.1$
LAML $5.42 $\pm$ 1.65$
REPLAY $21.11 $\pm$ 0.41$
ADAPTERCL $21.48 $\pm$ 0.19$
ABLATION
python scorer.py --model_checkpoint runs_NLG/MEMABLATION --task_type NLG --all --ablation
Method ACC
-------- ----------------------------------------
10 8.44$\pm$0.97
50 6.63$\pm$0.53
100 5.75$\pm$0.19
500 4.96$\pm$0.10
1000 4.36$\pm$0.24
Method ACC
-------- -----------------------------------------
10 18.86$\pm$0.68
50 21.11$\pm$0.41
100 21.71$\pm$0.25
500 22.86$\pm$0.25
1000 23.85$\pm$0.12
E2E
python scorer.py --model_checkpoint runs_E2E/BEST --task_type E2E --all --best
DST
Method ACC
--------- ----------------
MULTI 48.9
ADAPTERCL 35.06 $\pm$ 0.52
AGEM 6.37 $\pm$ 4.0
EWC 5.22 $\pm$ 4.46
L2 3.81 $\pm$ 3.44
LAML 4.55 $\pm$ 3.48
VANILLA 4.91 $\pm$ 4.46
REPLAY 30.33$\pm$ 1.24
ADAPTER MEAN: 0.9544274481238677
ADAPTER STD: 0.0025974980461427226
INTENT
Method ACC
--------- ----------------
MULTI 95.45
ADAPTERCL $90.46 \pm 0.6$
AGEM $34.04 \pm 6.36$
EWC $3.95 \pm 1.3$
L2 $3.74 \pm 1.4$
LAML $7.49 \pm 6.35$
VANILLA $4.08 \pm 1.37$
REPLAY $81.08 \pm 1.37$
BLEU
Method ACC
--------- ----------------
MULTI 23.61
ADAPTERCL $16.76 \pm 0.34$
AGEM $4.53 \pm 0.64$
EWC $5.06 \pm 0.48$
L2 $5.4 \pm 0.89$
LAML $3.0 \pm 0.93$
VANILLA $6.38 \pm 0.56$
REPLAY $17.4 \pm 0.68$
EER
Method ACC
--------- ----------------
MULTI 12.56
ADAPTERCL $31.78 $\pm$ 1.28$
AGEM $62.09 $\pm$ 6.88$
EWC $58.2 $\pm$ 3.66$
L2 $55.68 $\pm$ 7.09$
LAML $66.11 $\pm$ 6.97$
VANILLA $48.73 $\pm$ 3.81$
REPLAY $17.72 $\pm$ 0.85$
BLEU
Method ACC
-------- -----------------------------------------
10 13.350648648648647 +- 0.7252128193220151
50 17.401729729729727 +- 0.6779564230939323
100 18.47983783783784 +- 0.5479568158380509
500 19.641675675675675 +- 0.21028863950889007
1000 14.922054054054053 +- 0.20365173153569624
EER
Method ACC
-------- ---------------------------------------
10 25.7 +- 2.4918346654623775
50 17.722 +- 0.8515256895713713
100 16.7 +- 1.594979623694297
500 16.904 +- 0.8790358354469968
1000 44.86800000000001 +- 0.8108612705019269
INTENT
Method ACC
-------- ----------------------------------------
10 56.077999999999996 +- 6.929399396773145
50 81.08200000000001 +- 1.3688447684087492
100 87.53999999999999 +- 0.6051776598652667
500 93.68199999999999 +- 0.43960891710701083
1000 95.02 +- 0.1726267650163214
JGA
Method ACC
-------- ----------------------------------------
10 17.98 +- 2.1158733421450346
50 30.330000000000002 +- 1.2437845472588884
100 35.974000000000004 +- 0.7776271600195035
500 44.306 +- 0.2127533783515561
1000 48.908 +- 0.4503509742412028