metadata
license: mit
base_model: PORTULAN/albertina-ptbr-base
tags:
- generated_from_trainer
model-index:
- name: e3_lr2e-05
results: []
e3_lr2e-05
This model is a fine-tuned version of PORTULAN/albertina-ptbr-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.9281
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.4061 | 0.1040 | 100 | 1.1920 |
| 1.2553 | 0.2080 | 200 | 1.1209 |
| 1.2102 | 0.3120 | 300 | 1.0971 |
| 1.1773 | 0.4160 | 400 | 1.0738 |
| 1.1432 | 0.5200 | 500 | 1.0481 |
| 1.1302 | 0.6240 | 600 | 1.0320 |
| 1.1153 | 0.7280 | 700 | 1.0243 |
| 1.1057 | 0.8320 | 800 | 1.0107 |
| 1.0976 | 0.9360 | 900 | 1.0002 |
| 1.0889 | 1.0400 | 1000 | 0.9907 |
| 1.0797 | 1.1440 | 1100 | 0.9836 |
| 1.0633 | 1.2480 | 1200 | 0.9788 |
| 1.0582 | 1.3521 | 1300 | 0.9761 |
| 1.0578 | 1.4561 | 1400 | 0.9635 |
| 1.0423 | 1.5601 | 1500 | 0.9601 |
| 1.0411 | 1.6641 | 1600 | 0.9578 |
| 1.0406 | 1.7681 | 1700 | 0.9527 |
| 1.0436 | 1.8721 | 1800 | 0.9520 |
| 1.0363 | 1.9761 | 1900 | 0.9443 |
| 1.0274 | 2.0801 | 2000 | 0.9419 |
| 1.03 | 2.1841 | 2100 | 0.9417 |
| 1.0232 | 2.2881 | 2200 | 0.9392 |
| 1.0237 | 2.3921 | 2300 | 0.9374 |
| 1.0199 | 2.4961 | 2400 | 0.9354 |
| 1.0095 | 2.6001 | 2500 | 0.9399 |
| 1.0145 | 2.7041 | 2600 | 0.9343 |
| 1.0179 | 2.8081 | 2700 | 0.9297 |
| 1.0148 | 2.9121 | 2800 | 0.9328 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1