Commit
•
5bc36ce
1
Parent(s):
dee9e68
Update README.md
Browse files
README.md
CHANGED
|
@@ -197,4 +197,18 @@ The following hyperparameters were used during training:
|
|
| 197 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 198 |
- lr_scheduler_type: linear
|
| 199 |
- lr_scheduler_warmup_steps: 500
|
| 200 |
-
- num_epochs: 4
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 197 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 198 |
- lr_scheduler_type: linear
|
| 199 |
- lr_scheduler_warmup_steps: 500
|
| 200 |
+
- num_epochs: 4
|
| 201 |
+
|
| 202 |
+
### Training results at last epoch:
|
| 203 |
+
|
| 204 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 205 |
+
|:-------------:|:-----:|:-----:|:---------------:|
|
| 206 |
+
| 0.4487 | 4.0 | 20496 | 0.2799 |
|
| 207 |
+
|
| 208 |
+
|
| 209 |
+
### Framework versions
|
| 210 |
+
|
| 211 |
+
- Transformers 4.42.4
|
| 212 |
+
- Pytorch 2.3.1+cu121
|
| 213 |
+
- Datasets 2.21.0
|
| 214 |
+
- Tokenizers 0.19.1
|