End of training
Browse files
README.md
CHANGED
|
@@ -327,8 +327,7 @@ You can finetune this model on your own dataset.
|
|
| 327 |
### Training Hyperparameters
|
| 328 |
#### Non-Default Hyperparameters
|
| 329 |
|
| 330 |
-
- `
|
| 331 |
-
- `num_train_epochs`: 8
|
| 332 |
- `push_to_hub`: True
|
| 333 |
- `batch_sampler`: no_duplicates
|
| 334 |
|
|
@@ -339,7 +338,7 @@ You can finetune this model on your own dataset.
|
|
| 339 |
- `do_predict`: False
|
| 340 |
- `eval_strategy`: no
|
| 341 |
- `prediction_loss_only`: True
|
| 342 |
-
- `per_device_train_batch_size`:
|
| 343 |
- `per_device_eval_batch_size`: 8
|
| 344 |
- `per_gpu_train_batch_size`: None
|
| 345 |
- `per_gpu_eval_batch_size`: None
|
|
@@ -352,7 +351,7 @@ You can finetune this model on your own dataset.
|
|
| 352 |
- `adam_beta2`: 0.999
|
| 353 |
- `adam_epsilon`: 1e-08
|
| 354 |
- `max_grad_norm`: 1.0
|
| 355 |
-
- `num_train_epochs`:
|
| 356 |
- `max_steps`: -1
|
| 357 |
- `lr_scheduler_type`: linear
|
| 358 |
- `lr_scheduler_kwargs`: {}
|
|
@@ -453,9 +452,14 @@ You can finetune this model on your own dataset.
|
|
| 453 |
### Training Logs
|
| 454 |
| Epoch | Step | Training Loss |
|
| 455 |
|:------:|:----:|:-------------:|
|
| 456 |
-
|
|
| 457 |
-
|
|
| 458 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 459 |
|
| 460 |
|
| 461 |
### Framework Versions
|
|
|
|
| 327 |
### Training Hyperparameters
|
| 328 |
#### Non-Default Hyperparameters
|
| 329 |
|
| 330 |
+
- `num_train_epochs`: 10
|
|
|
|
| 331 |
- `push_to_hub`: True
|
| 332 |
- `batch_sampler`: no_duplicates
|
| 333 |
|
|
|
|
| 338 |
- `do_predict`: False
|
| 339 |
- `eval_strategy`: no
|
| 340 |
- `prediction_loss_only`: True
|
| 341 |
+
- `per_device_train_batch_size`: 8
|
| 342 |
- `per_device_eval_batch_size`: 8
|
| 343 |
- `per_gpu_train_batch_size`: None
|
| 344 |
- `per_gpu_eval_batch_size`: None
|
|
|
|
| 351 |
- `adam_beta2`: 0.999
|
| 352 |
- `adam_epsilon`: 1e-08
|
| 353 |
- `max_grad_norm`: 1.0
|
| 354 |
+
- `num_train_epochs`: 10
|
| 355 |
- `max_steps`: -1
|
| 356 |
- `lr_scheduler_type`: linear
|
| 357 |
- `lr_scheduler_kwargs`: {}
|
|
|
|
| 452 |
### Training Logs
|
| 453 |
| Epoch | Step | Training Loss |
|
| 454 |
|:------:|:----:|:-------------:|
|
| 455 |
+
| 1.1416 | 500 | 0.3244 |
|
| 456 |
+
| 2.2831 | 1000 | 0.1048 |
|
| 457 |
+
| 3.4247 | 1500 | 0.0394 |
|
| 458 |
+
| 4.5662 | 2000 | 0.0211 |
|
| 459 |
+
| 5.7078 | 2500 | 0.0145 |
|
| 460 |
+
| 6.8493 | 3000 | 0.0114 |
|
| 461 |
+
| 7.9909 | 3500 | 0.0106 |
|
| 462 |
+
| 9.1324 | 4000 | 0.0092 |
|
| 463 |
|
| 464 |
|
| 465 |
### Framework Versions
|