File size: 9,965 Bytes
9008d50
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
Visible device: cuda
Seed used: 0
Batch size: 64
Epochs: 40
Learning rate: 1e-05
Entropy weight: 0.01
Regularization weight: 0.0
Only use multiwoz like domains: False
We use: 1.0% of the data
Dialogue order used: 0
Vectorizer: Data set used is multiwoz21
We filter state by active domains: True
Vectorizer: Data set used is multiwoz21
Embedding semantic descriptions: True
Embedded descriptions successfully. Size: torch.Size([338, 768])
Data set used for descriptions: multiwoz21
We use Roberta to embed actions.
Loaded model from experiments/seed0/save/supervised.pol.mdl
Start training
Epoch: 0
Average actions: 1.9973957538604736
Average target actions: 2.5520834922790527
Precision: 0.09615384615384616
Recall: 0.07462686567164178
F1: 0.08403361344537816
<<dialog policy>> epoch 0: saved network to mdl
Best Precision: 0.09615384615384616
Best Recall: 0.07462686567164178
Best F1: 0.08403361344537816
Epoch: 1
Precision: 0.09615384615384616
Recall: 0.07462686567164178
F1: 0.08403361344537816
Best Precision: 0.09615384615384616
Best Recall: 0.07462686567164178
Best F1: 0.08403361344537816
Epoch: 2
Average actions: 2.3515625
Average target actions: 2.6197917461395264
Precision: 0.10526315789473684
Recall: 0.08955223880597014
F1: 0.0967741935483871
<<dialog policy>> epoch 2: saved network to mdl
Best Precision: 0.10526315789473684
Best Recall: 0.08955223880597014
Best F1: 0.0967741935483871
Epoch: 3
Precision: 0.10526315789473684
Recall: 0.08955223880597014
F1: 0.0967741935483871
Best Precision: 0.10526315789473684
Best Recall: 0.08955223880597014
Best F1: 0.0967741935483871
Epoch: 4
Average actions: 1.6770832538604736
Average target actions: 2.8567709922790527
Precision: 0.1347517730496454
Recall: 0.0945273631840796
F1: 0.11111111111111112
<<dialog policy>> epoch 4: saved network to mdl
Best Precision: 0.1347517730496454
Best Recall: 0.0945273631840796
Best F1: 0.11111111111111112
Epoch: 5
Precision: 0.1347517730496454
Recall: 0.0945273631840796
F1: 0.11111111111111112
Best Precision: 0.1347517730496454
Best Recall: 0.0945273631840796
Best F1: 0.11111111111111112
Epoch: 6
Average actions: 1.9088542461395264
Average target actions: 2.7213542461395264
Precision: 0.12080536912751678
Recall: 0.08955223880597014
F1: 0.10285714285714286
Best Precision: 0.1347517730496454
Best Recall: 0.0945273631840796
Best F1: 0.11111111111111112
Epoch: 7
Precision: 0.12080536912751678
Recall: 0.08955223880597014
F1: 0.10285714285714286
Best Precision: 0.1347517730496454
Best Recall: 0.0945273631840796
Best F1: 0.11111111111111112
Epoch: 8
Average actions: 2.0572915077209473
Average target actions: 2.8229167461395264
Precision: 0.12903225806451613
Recall: 0.09950248756218906
F1: 0.11235955056179776
<<dialog policy>> epoch 8: saved network to mdl
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 9
Precision: 0.12903225806451613
Recall: 0.09950248756218906
F1: 0.11235955056179776
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 10
Average actions: 2.0911459922790527
Average target actions: 2.6875
Precision: 0.11612903225806452
Recall: 0.08955223880597014
F1: 0.10112359550561797
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 11
Precision: 0.11612903225806452
Recall: 0.08955223880597014
F1: 0.10112359550561797
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 12
Average actions: 2.0833332538604736
Average target actions: 2.5859375
Precision: 0.11976047904191617
Recall: 0.09950248756218906
F1: 0.10869565217391305
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 13
Precision: 0.11976047904191617
Recall: 0.09950248756218906
F1: 0.10869565217391305
Best Precision: 0.1347517730496454
Best Recall: 0.09950248756218906
Best F1: 0.11235955056179776
Epoch: 14
Average actions: 2.1119790077209473
Average target actions: 2.7213542461395264
Precision: 0.16778523489932887
Recall: 0.12437810945273632
F1: 0.14285714285714285
<<dialog policy>> epoch 14: saved network to mdl
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 15
Precision: 0.16778523489932887
Recall: 0.12437810945273632
F1: 0.14285714285714285
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 16
Average actions: 1.7994792461395264
Average target actions: 2.5520834922790527
Precision: 0.10135135135135136
Recall: 0.07462686567164178
F1: 0.08595988538681948
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 17
Precision: 0.10135135135135136
Recall: 0.07462686567164178
F1: 0.08595988538681948
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 18
Average actions: 2.0572915077209473
Average target actions: 2.7552084922790527
Precision: 0.13548387096774195
Recall: 0.1044776119402985
F1: 0.11797752808988765
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 19
Precision: 0.13548387096774195
Recall: 0.1044776119402985
F1: 0.11797752808988765
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 20
Average actions: 1.9661457538604736
Average target actions: 2.7213542461395264
Precision: 0.1118421052631579
Recall: 0.0845771144278607
F1: 0.0963172804532578
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 21
Precision: 0.1118421052631579
Recall: 0.0845771144278607
F1: 0.0963172804532578
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 22
Average actions: 1.9557292461395264
Average target actions: 2.5520834922790527
Precision: 0.07741935483870968
Recall: 0.05970149253731343
F1: 0.06741573033707865
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 23
Precision: 0.07741935483870968
Recall: 0.05970149253731343
F1: 0.06741573033707865
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 24
Average actions: 2.0833334922790527
Average target actions: 2.8229167461395264
Precision: 0.09090909090909091
Recall: 0.06965174129353234
F1: 0.07887323943661972
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 25
Precision: 0.09090909090909091
Recall: 0.06965174129353234
F1: 0.07887323943661972
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 26
Average actions: 1.7135417461395264
Average target actions: 2.6197917461395264
Precision: 0.145985401459854
Recall: 0.09950248756218906
F1: 0.1183431952662722
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 27
Precision: 0.145985401459854
Recall: 0.09950248756218906
F1: 0.1183431952662722
Best Precision: 0.16778523489932887
Best Recall: 0.12437810945273632
Best F1: 0.14285714285714285
Epoch: 28
Average actions: 2.0364584922790527
Average target actions: 2.5520834922790527
Precision: 0.16891891891891891
Recall: 0.12437810945273632
F1: 0.14326647564469916
<<dialog policy>> epoch 28: saved network to mdl
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 29
Precision: 0.16891891891891891
Recall: 0.12437810945273632
F1: 0.14326647564469916
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 30
Average actions: 2.0026040077209473
Average target actions: 2.3828125
Precision: 0.16216216216216217
Recall: 0.11940298507462686
F1: 0.13753581661891118
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 31
Precision: 0.16216216216216217
Recall: 0.11940298507462686
F1: 0.13753581661891118
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 32
Average actions: 1.8046875
Average target actions: 2.6875
Precision: 0.12142857142857143
Recall: 0.0845771144278607
F1: 0.09970674486803519
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 33
Precision: 0.12142857142857143
Recall: 0.0845771144278607
F1: 0.09970674486803519
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 34
Average actions: 1.9348957538604736
Average target actions: 2.6875
Precision: 0.12162162162162163
Recall: 0.08955223880597014
F1: 0.10315186246418337
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 35
Precision: 0.12162162162162163
Recall: 0.08955223880597014
F1: 0.10315186246418337
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 36
Average actions: 2.0989584922790527
Average target actions: 2.484375
Precision: 0.14743589743589744
Recall: 0.11442786069651742
F1: 0.1288515406162465
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 37
Precision: 0.14743589743589744
Recall: 0.11442786069651742
F1: 0.1288515406162465
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 38
Average actions: 2.0260415077209473
Average target actions: 2.5520834922790527
Precision: 0.1456953642384106
Recall: 0.10945273631840796
F1: 0.12499999999999997
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916
Epoch: 39
Precision: 0.1456953642384106
Recall: 0.10945273631840796
F1: 0.12499999999999997
Best Precision: 0.16891891891891891
Best Recall: 0.12437810945273632
Best F1: 0.14326647564469916