Spaces:
Sleeping
Sleeping
Commit
•
a032f99
1
Parent(s):
3cfdee5
Traer los argumentos de training
Browse files- models_config.json +56 -0
models_config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"mon": {
|
| 3 |
+
"dataset_path": "data/baby_cry_detection",
|
| 4 |
+
"output_dir": "distilhubert-finetuned-cry-detector",
|
| 5 |
+
"training_args": {
|
| 6 |
+
"num_train_epochs": 6,
|
| 7 |
+
"learning_rate": 0.0001,
|
| 8 |
+
"warmup_ratio": 0.001,
|
| 9 |
+
"output_dir": "distilhubert-finetuned-cry-detector",
|
| 10 |
+
"eval_strategy": "epoch",
|
| 11 |
+
"save_strategy": "epoch",
|
| 12 |
+
"lr_scheduler_type": "cosine",
|
| 13 |
+
"auto_find_batch_size": true,
|
| 14 |
+
"per_device_train_batch_size": 8,
|
| 15 |
+
"per_device_eval_batch_size": 8,
|
| 16 |
+
"gradient_accumulation_steps": 8,
|
| 17 |
+
"gradient_checkpointing": true,
|
| 18 |
+
"load_best_model_at_end": true,
|
| 19 |
+
"greater_is_better": true,
|
| 20 |
+
"metric_for_best_model": "accuracy",
|
| 21 |
+
"optim": "adamw_torch",
|
| 22 |
+
"hub_strategy": "checkpoint",
|
| 23 |
+
"report_to": "tensorboard",
|
| 24 |
+
"full_determinism": true,
|
| 25 |
+
"seed": 123,
|
| 26 |
+
"data_seed":123
|
| 27 |
+
}
|
| 28 |
+
},
|
| 29 |
+
"class": {
|
| 30 |
+
"dataset_path": "data/mixed_data_nuevo",
|
| 31 |
+
"output_dir": "distilhubert-finetuned-mixed-data",
|
| 32 |
+
"training_args": {
|
| 33 |
+
"num_train_epochs": 2,
|
| 34 |
+
"learning_rate": 0.0001,
|
| 35 |
+
"warmup_ratio": 0.001,
|
| 36 |
+
"output_dir": "distilhubert-finetuned-mixed-data",
|
| 37 |
+
"eval_strategy": "epoch",
|
| 38 |
+
"save_strategy": "epoch",
|
| 39 |
+
"lr_scheduler_type": "cosine",
|
| 40 |
+
"auto_find_batch_size": true,
|
| 41 |
+
"per_device_train_batch_size": 8,
|
| 42 |
+
"per_device_eval_batch_size": 8,
|
| 43 |
+
"gradient_accumulation_steps": 8,
|
| 44 |
+
"gradient_checkpointing": true,
|
| 45 |
+
"load_best_model_at_end": true,
|
| 46 |
+
"greater_is_better": true,
|
| 47 |
+
"metric_for_best_model": "accuracy",
|
| 48 |
+
"optim": "adamw_torch",
|
| 49 |
+
"hub_strategy": "checkpoint",
|
| 50 |
+
"report_to": "tensorboard",
|
| 51 |
+
"full_determinism": true,
|
| 52 |
+
"seed": 123,
|
| 53 |
+
"data_seed":123
|
| 54 |
+
}
|
| 55 |
+
}
|
| 56 |
+
}
|