Pablogps commited on
Commit
96fc0c7
1 Parent(s): 96e881d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -1
README.md CHANGED
@@ -149,6 +149,7 @@ Our final models were trained on a different number of steps and sequence length
149
  <figure>
150
 
151
  <caption>Table 1. Evaluation made by the Barcelona Supercomputing Center of their models and BERTIN (beta, seq len 128).</caption>
 
152
  | Dataset | Metric | RoBERTa-b | RoBERTa-l | BETO | mBERT | BERTIN |
153
  |-------------|----------|-----------|-----------|--------|--------|--------|
154
  | UD-POS | F1 | **0.9907** | 0.9901 | 0.9900 | 0.9886 | **0.9904** |
@@ -167,6 +168,7 @@ All of our models attained good accuracy values, in the range of 0.65, as can be
167
  <figure>
168
 
169
  <caption>Table 2. Accuracy for the different language models.</caption>
 
170
  | Model | Accuracy |
171
  |----------------------------------------------------|----------|
172
  | bertin-project/bertin-roberta-base-spanish | 0.6547 |
@@ -191,6 +193,7 @@ All models trained with max length 512 and batch size 8, using the CoNLL 2002 da
191
  <figure>
192
 
193
  <caption>Table 3. Results for POS.</caption>
 
194
  | Model | F1 | Accuracy |
195
  |----------------------------------------------------|----------|----------|
196
  | bert-base-multilingual-cased | 0.9629 | 0.9687 |
@@ -212,6 +215,7 @@ All models trained with max length 512 and batch size 8, using the CoNLL 2002 da
212
  <figure>
213
 
214
  <caption>Table 4. Results for NER.</caption>
 
215
  | Model | F1 | Accuracy |
216
  |----------------------------------------------------|----------|----------|
217
  | bert-base-multilingual-cased | 0.8539 | 0.9779 |
@@ -233,6 +237,7 @@ All models trained with max length 512 and batch size 8. These numbers are surpr
233
  <figure>
234
 
235
  <caption>Table 5. Results for PAWS-X.</caption>
 
236
  | Model | Accuracy |
237
  |----------------------------------------------------|----------|
238
  | bert-base-multilingual-cased | 0.5765 |
@@ -253,6 +258,7 @@ All models trained with max length 512 and batch size 8. These numbers are surpr
253
  <figure>
254
 
255
  <caption>Table 6. Results for XNLI with sequence length 256 and batch size 32.</caption>
 
256
  | Model | Accuracy |
257
  |----------------------------------------------------|----------|
258
  | bert-base-multilingual-cased | 0.7852 |
@@ -270,7 +276,6 @@ All models trained with max length 512 and batch size 8. These numbers are surpr
270
  <figure>
271
 
272
  <caption>Table 7. Results for XNLI with sequence length 512 and batch size 16.</caption>
273
- </figure>
274
 
275
  | Model | Accuracy |
276
  |----------------------------------------------------|----------|
@@ -283,6 +288,7 @@ All models trained with max length 512 and batch size 8. These numbers are surpr
283
  | bertin-project/bertin-base-random-exp-512seqlen | 0.7799 |
284
  | bertin-project/bertin-base-gaussian-exp-512seqlen | 0.7843 |
285
 
 
286
 
287
  # Conclusions
288
 
 
149
  <figure>
150
 
151
  <caption>Table 1. Evaluation made by the Barcelona Supercomputing Center of their models and BERTIN (beta, seq len 128).</caption>
152
+
153
  | Dataset | Metric | RoBERTa-b | RoBERTa-l | BETO | mBERT | BERTIN |
154
  |-------------|----------|-----------|-----------|--------|--------|--------|
155
  | UD-POS | F1 | **0.9907** | 0.9901 | 0.9900 | 0.9886 | **0.9904** |
 
168
  <figure>
169
 
170
  <caption>Table 2. Accuracy for the different language models.</caption>
171
+
172
  | Model | Accuracy |
173
  |----------------------------------------------------|----------|
174
  | bertin-project/bertin-roberta-base-spanish | 0.6547 |
 
193
  <figure>
194
 
195
  <caption>Table 3. Results for POS.</caption>
196
+
197
  | Model | F1 | Accuracy |
198
  |----------------------------------------------------|----------|----------|
199
  | bert-base-multilingual-cased | 0.9629 | 0.9687 |
 
215
  <figure>
216
 
217
  <caption>Table 4. Results for NER.</caption>
218
+
219
  | Model | F1 | Accuracy |
220
  |----------------------------------------------------|----------|----------|
221
  | bert-base-multilingual-cased | 0.8539 | 0.9779 |
 
237
  <figure>
238
 
239
  <caption>Table 5. Results for PAWS-X.</caption>
240
+
241
  | Model | Accuracy |
242
  |----------------------------------------------------|----------|
243
  | bert-base-multilingual-cased | 0.5765 |
 
258
  <figure>
259
 
260
  <caption>Table 6. Results for XNLI with sequence length 256 and batch size 32.</caption>
261
+
262
  | Model | Accuracy |
263
  |----------------------------------------------------|----------|
264
  | bert-base-multilingual-cased | 0.7852 |
 
276
  <figure>
277
 
278
  <caption>Table 7. Results for XNLI with sequence length 512 and batch size 16.</caption>
 
279
 
280
  | Model | Accuracy |
281
  |----------------------------------------------------|----------|
 
288
  | bertin-project/bertin-base-random-exp-512seqlen | 0.7799 |
289
  | bertin-project/bertin-base-gaussian-exp-512seqlen | 0.7843 |
290
 
291
+ </figure>
292
 
293
  # Conclusions
294