robertou2 commited on
Commit
f6790c1
1 Parent(s): 4afa661

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +15 -255
README.md CHANGED
@@ -19,15 +19,15 @@ model-index:
19
  metrics:
20
  - name: Accuracy
21
  type: accuracy
22
- value: 0.80
23
  ---
24
 
25
  # twitter_sexismo-finetuned-exist2021
26
 
27
  This model is a fine-tuned version of [pysentimiento/robertuito-hate-speech](https://huggingface.co/pysentimiento/robertuito-hate-speech) on the EXIST dataset.
28
  It achieves the following results on the evaluation set:
29
- - Loss: 0.12
30
- - Accuracy: 0.80
31
 
32
  ## Model description
33
 
@@ -46,265 +46,25 @@ More information needed
46
  ### Training hyperparameters
47
 
48
  The following hyperparameters were used during training:
49
- - my_learning_rate = 2E-6
50
  - my_adam_epsilon = 1E-8
51
- - my_number_of_epochs = 15
52
  - my_warmup = 3
53
  - my_mini_batch_size = 32
54
  - optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
55
  - lr_scheduler_type: linear
56
- - num_epochs: 30
57
 
58
  ### Training results
59
-
60
- ======== Epoch 1 / 30 ========
61
- Training...
62
- Batch 50 of 143. Elapsed: 0:00:53.
63
- Batch 100 of 143. Elapsed: 0:01:45.
64
-
65
- Average training loss: 0.39
66
- Training epoch took: 0:02:29
67
-
68
- ======== Epoch 2 / 30 ========
69
- Training...
70
- Batch 50 of 143. Elapsed: 0:00:52.
71
- Batch 100 of 143. Elapsed: 0:01:44.
72
-
73
- Average training loss: 0.36
74
- Training epoch took: 0:02:29
75
-
76
- ======== Epoch 3 / 30 ========
77
- Training...
78
- Batch 50 of 143. Elapsed: 0:00:52.
79
- Batch 100 of 143. Elapsed: 0:01:44.
80
-
81
- Average training loss: 0.34
82
- Training epoch took: 0:02:29
83
-
84
- ======== Epoch 4 / 30 ========
85
- Training...
86
- Batch 50 of 143. Elapsed: 0:00:52.
87
- Batch 100 of 143. Elapsed: 0:01:44.
88
-
89
- Average training loss: 0.33
90
- Training epoch took: 0:02:29
91
-
92
- ======== Epoch 5 / 30 ========
93
- Training...
94
- Batch 50 of 143. Elapsed: 0:00:52.
95
- Batch 100 of 143. Elapsed: 0:01:44.
96
-
97
- Average training loss: 0.31
98
- Training epoch took: 0:02:29
99
-
100
- ======== Epoch 6 / 30 ========
101
- Training...
102
- Batch 50 of 143. Elapsed: 0:00:52.
103
- Batch 100 of 143. Elapsed: 0:01:44.
104
-
105
- Average training loss: 0.29
106
- Training epoch took: 0:02:29
107
-
108
- ======== Epoch 7 / 30 ========
109
- Training...
110
- Batch 50 of 143. Elapsed: 0:00:52.
111
- Batch 100 of 143. Elapsed: 0:01:44.
112
-
113
- Average training loss: 0.28
114
- Training epoch took: 0:02:29
115
-
116
- ======== Epoch 8 / 30 ========
117
- Training...
118
- Batch 50 of 143. Elapsed: 0:00:52.
119
- Batch 100 of 143. Elapsed: 0:01:44.
120
-
121
- Average training loss: 0.27
122
- Training epoch took: 0:02:29
123
-
124
- ======== Epoch 9 / 30 ========
125
- Training...
126
- Batch 50 of 143. Elapsed: 0:00:52.
127
- Batch 100 of 143. Elapsed: 0:01:44.
128
-
129
- Average training loss: 0.25
130
- Training epoch took: 0:02:28
131
-
132
- ======== Epoch 10 / 30 ========
133
- Training...
134
- Batch 50 of 143. Elapsed: 0:00:52.
135
- Batch 100 of 143. Elapsed: 0:01:44.
136
-
137
- Average training loss: 0.24
138
- Training epoch took: 0:02:29
139
-
140
- ======== Epoch 11 / 30 ========
141
- Training...
142
- Batch 50 of 143. Elapsed: 0:00:52.
143
- Batch 100 of 143. Elapsed: 0:01:44.
144
-
145
- Average training loss: 0.23
146
- Training epoch took: 0:02:28
147
-
148
- ======== Epoch 12 / 30 ========
149
- Training...
150
- Batch 50 of 143. Elapsed: 0:00:52.
151
- Batch 100 of 143. Elapsed: 0:01:44.
152
-
153
- Average training loss: 0.22
154
- Training epoch took: 0:02:29
155
-
156
- ======== Epoch 13 / 30 ========
157
- Training...
158
- Batch 50 of 143. Elapsed: 0:00:52.
159
- Batch 100 of 143. Elapsed: 0:01:44.
160
-
161
- Average training loss: 0.21
162
- Training epoch took: 0:02:29
163
-
164
- ======== Epoch 14 / 30 ========
165
- Training...
166
- Batch 50 of 143. Elapsed: 0:00:52.
167
- Batch 100 of 143. Elapsed: 0:01:44.
168
-
169
- Average training loss: 0.20
170
- Training epoch took: 0:02:29
171
-
172
- ======== Epoch 15 / 30 ========
173
- Training...
174
- Batch 50 of 143. Elapsed: 0:00:52.
175
- Batch 100 of 143. Elapsed: 0:01:44.
176
-
177
- Average training loss: 0.19
178
- Training epoch took: 0:02:29
179
-
180
- ======== Epoch 16 / 30 ========
181
- Training...
182
- Batch 50 of 143. Elapsed: 0:00:52.
183
- Batch 100 of 143. Elapsed: 0:01:44.
184
-
185
- Average training loss: 0.18
186
- Training epoch took: 0:02:29
187
-
188
- ======== Epoch 17 / 30 ========
189
- Training...
190
- Batch 50 of 143. Elapsed: 0:00:52.
191
- Batch 100 of 143. Elapsed: 0:01:44.
192
-
193
- Average training loss: 0.17
194
- Training epoch took: 0:02:29
195
-
196
- ======== Epoch 18 / 30 ========
197
- Training...
198
- Batch 50 of 143. Elapsed: 0:00:52.
199
- Batch 100 of 143. Elapsed: 0:01:44.
200
-
201
- Average training loss: 0.17
202
- Training epoch took: 0:02:29
203
-
204
- ======== Epoch 19 / 30 ========
205
- Training...
206
- Batch 50 of 143. Elapsed: 0:00:52.
207
- Batch 100 of 143. Elapsed: 0:01:44.
208
-
209
- Average training loss: 0.16
210
- Training epoch took: 0:02:29
211
-
212
- ======== Epoch 20 / 30 ========
213
- Training...
214
- Batch 50 of 143. Elapsed: 0:00:52.
215
- Batch 100 of 143. Elapsed: 0:01:44.
216
-
217
- Average training loss: 0.15
218
- Training epoch took: 0:02:29
219
-
220
- ======== Epoch 21 / 30 ========
221
- Training...
222
- Batch 50 of 143. Elapsed: 0:00:52.
223
- Batch 100 of 143. Elapsed: 0:01:44.
224
-
225
- Average training loss: 0.15
226
- Training epoch took: 0:02:29
227
-
228
- ======== Epoch 22 / 30 ========
229
- Training...
230
- Batch 50 of 143. Elapsed: 0:00:52.
231
- Batch 100 of 143. Elapsed: 0:01:44.
232
-
233
- Average training loss: 0.15
234
- Training epoch took: 0:02:29
235
-
236
- ======== Epoch 23 / 30 ========
237
- Training...
238
- Batch 50 of 143. Elapsed: 0:00:52.
239
- Batch 100 of 143. Elapsed: 0:01:44.
240
-
241
- Average training loss: 0.14
242
- Training epoch took: 0:02:29
243
-
244
- ======== Epoch 24 / 30 ========
245
- Training...
246
- Batch 50 of 143. Elapsed: 0:00:52.
247
- Batch 100 of 143. Elapsed: 0:01:45.
248
-
249
- Average training loss: 0.14
250
- Training epoch took: 0:02:29
251
-
252
- ======== Epoch 25 / 30 ========
253
- Training...
254
- Batch 50 of 143. Elapsed: 0:00:52.
255
- Batch 100 of 143. Elapsed: 0:01:44.
256
-
257
- Average training loss: 0.14
258
- Training epoch took: 0:02:29
259
-
260
- ======== Epoch 26 / 30 ========
261
- Training...
262
- Batch 50 of 143. Elapsed: 0:00:52.
263
- Batch 100 of 143. Elapsed: 0:01:44.
264
-
265
- Average training loss: 0.13
266
- Training epoch took: 0:02:29
267
-
268
- ======== Epoch 27 / 30 ========
269
- Training...
270
- Batch 50 of 143. Elapsed: 0:00:52.
271
- Batch 100 of 143. Elapsed: 0:01:44.
272
-
273
- Average training loss: 0.13
274
- Training epoch took: 0:02:29
275
-
276
- ======== Epoch 28 / 30 ========
277
- Training...
278
- Batch 50 of 143. Elapsed: 0:00:52.
279
- Batch 100 of 143. Elapsed: 0:01:44.
280
-
281
- Average training loss: 0.13
282
- Training epoch took: 0:02:29
283
-
284
- ======== Epoch 29 / 30 ========
285
- Training...
286
- Batch 50 of 143. Elapsed: 0:00:52.
287
- Batch 100 of 143. Elapsed: 0:01:44.
288
-
289
- Average training loss: 0.12
290
- Training epoch took: 0:02:29
291
-
292
- ======== Epoch 30 / 30 ========
293
- Training...
294
- Batch 50 of 143. Elapsed: 0:00:52.
295
- Batch 100 of 143. Elapsed: 0:01:44.
296
-
297
- Average training loss: 0.13
298
- Training epoch took: 0:02:29
299
-
300
- precision recall f1-score support
301
-
302
- 0 0.78 0.82 0.80 551
303
- 1 0.82 0.79 0.81 590
304
-
305
- accuracy 0.80 1141
306
- macro avg 0.80 0.80 0.80 1141
307
- weighted avg 0.80 0.80 0.80 1141
308
 
309
  ### Framework versions
310
 
 
19
  metrics:
20
  - name: Accuracy
21
  type: accuracy
22
+ value: 0.86
23
  ---
24
 
25
  # twitter_sexismo-finetuned-exist2021
26
 
27
  This model is a fine-tuned version of [pysentimiento/robertuito-hate-speech](https://huggingface.co/pysentimiento/robertuito-hate-speech) on the EXIST dataset.
28
  It achieves the following results on the evaluation set:
29
+ - Loss: 0.4
30
+ - Accuracy: 0.86
31
 
32
  ## Model description
33
 
 
46
  ### Training hyperparameters
47
 
48
  The following hyperparameters were used during training:
49
+ - my_learning_rate = 5E-5
50
  - my_adam_epsilon = 1E-8
51
+ - my_number_of_epochs = 8
52
  - my_warmup = 3
53
  - my_mini_batch_size = 32
54
  - optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
55
  - lr_scheduler_type: linear
56
+ - num_epochs: 8
57
 
58
  ### Training results
59
+ Epoch Training Loss Validation Loss Accuracy F1 Precision Recall
60
+ 1 0.398400 0.336709 0.861404 0.855311 0.872897 0.838420
61
+ 2 0.136100 0.575872 0.846491 0.854772 0.794753 0.924596
62
+ 3 0.105600 0.800685 0.848246 0.837863 0.876471 0.802513
63
+ 4 0.066500 0.928388 0.849123 0.856187 0.801252 0.919210
64
+ 5 0.004500 0.990655 0.851754 0.853680 0.824415 0.885099
65
+ 6 0.005500 1.035315 0.852632 0.856164 0.818331 0.897666
66
+ 7 0.000200 1.052970 0.857895 0.859375 0.831933 0.888689
67
+ 8 0.001700 1.048338 0.856140 0.857143 0.832487 0.883303
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
68
 
69
  ### Framework versions
70