htdung167 commited on
Commit
fc4485d
1 Parent(s): a02771e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -2
README.md CHANGED
@@ -172,14 +172,19 @@ mask_filler("đúng nhận sai <mask>", top_k=10)
172
 
173
  ## Fine-tune Configuration
174
  We fine-tune `5CD-AI/viso-twhin-bert-large` on 4 downstream tasks with `transformer` library with the following configuration:
 
175
  - seed: 42
176
- - gradient_accumulation_steps: 1
177
  - weight_decay: 0.01
178
  - optimizer: AdamW with betas=(0.9, 0.999) and epsilon=1e-08
 
179
  - training_epochs: 30
180
  - model_max_length: 128
181
  - learning_rate: 1e-5
 
 
 
182
  And different additional configurations for each task:
183
  | Emotion Recognition | Hate Speech Detection | Spam Reviews Detection | Hate Speech Spans Detection |
184
  | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- |
185
- |\- train_batch_size: 64<br>\- lr_scheduler_type: linear | \- train_batch_size: 32<br>\- lr_scheduler_type: linear | \- train_batch_size: 32<br>\- lr_scheduler_type: cosine | \- train_batch_size: 32<br>\- lr_scheduler_type: cosine |
 
172
 
173
  ## Fine-tune Configuration
174
  We fine-tune `5CD-AI/viso-twhin-bert-large` on 4 downstream tasks with `transformer` library with the following configuration:
175
+ - train_batch_size: 16
176
  - seed: 42
177
+ - gradient_accumulation_steps: 4
178
  - weight_decay: 0.01
179
  - optimizer: AdamW with betas=(0.9, 0.999) and epsilon=1e-08
180
+ - lr_scheduler_type: cosine
181
  - training_epochs: 30
182
  - model_max_length: 128
183
  - learning_rate: 1e-5
184
+ - metric_for_best_model: wf1
185
+ - strategy: epoch
186
+
187
  And different additional configurations for each task:
188
  | Emotion Recognition | Hate Speech Detection | Spam Reviews Detection | Hate Speech Spans Detection |
189
  | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- |
190
+ |\- learning_rate: 1e-5| \- learning_rate: 5e-6 | \- learning_rate: 1e-5 | \- learning_rate: 5e-6 |