|
--- |
|
license: other |
|
library_name: peft |
|
tags: |
|
- trl |
|
- sft |
|
- generated_from_trainer |
|
base_model: google/gemma-2b |
|
model-index: |
|
- name: gemma-2b-spanishbillionwords |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# gemma-2b-spanishbillionwords |
|
|
|
This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on [Spanish Billion Words](https://huggingface.co/datasets/jhonparra18/spanish_billion_words_clean). |
|
This is the base Gemma model fine-tuned to perform better on spanish language. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 4.3306 |
|
|
|
## 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: 0.0003 |
|
- train_batch_size: 1 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 2 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 1 |
|
- training_steps: 60 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 5.1254 | 0.0 | 1 | 5.0205 | |
|
| 4.3187 | 0.0 | 2 | 5.0029 | |
|
| 3.8173 | 0.0 | 3 | 4.9801 | |
|
| 5.3879 | 0.0 | 4 | 4.9582 | |
|
| 5.718 | 0.0 | 5 | 4.9343 | |
|
| 5.8628 | 0.0 | 6 | 4.9104 | |
|
| 4.5401 | 0.0 | 7 | 4.8830 | |
|
| 4.4219 | 0.0 | 8 | 4.8539 | |
|
| 5.5169 | 0.0 | 9 | 4.8234 | |
|
| 4.813 | 0.0 | 10 | 4.7878 | |
|
| 4.2111 | 0.0 | 11 | 4.7576 | |
|
| 4.6504 | 0.0 | 12 | 4.7314 | |
|
| 3.7923 | 0.0 | 13 | 4.7116 | |
|
| 3.7773 | 0.0 | 14 | 4.6890 | |
|
| 4.6773 | 0.0 | 15 | 4.6616 | |
|
| 3.0179 | 0.0 | 16 | 4.6329 | |
|
| 3.8922 | 0.0 | 17 | 4.6099 | |
|
| 4.3289 | 0.0 | 18 | 4.5940 | |
|
| 5.0925 | 0.0 | 19 | 4.5822 | |
|
| 4.6499 | 0.0 | 20 | 4.5711 | |
|
| 3.9758 | 0.0 | 21 | 4.5585 | |
|
| 4.593 | 0.0 | 22 | 4.5454 | |
|
| 5.2496 | 0.0 | 23 | 4.5346 | |
|
| 4.2548 | 0.0 | 24 | 4.5217 | |
|
| 3.5209 | 0.0 | 25 | 4.5059 | |
|
| 4.4781 | 0.0 | 26 | 4.4930 | |
|
| 5.4472 | 0.0 | 27 | 4.4834 | |
|
| 4.1987 | 0.0 | 28 | 4.4756 | |
|
| 5.2324 | 0.0 | 29 | 4.4684 | |
|
| 4.8068 | 0.0 | 30 | 4.4593 | |
|
| 3.5455 | 0.0 | 31 | 4.4521 | |
|
| 3.6516 | 0.0 | 32 | 4.4415 | |
|
| 4.1368 | 0.0 | 33 | 4.4289 | |
|
| 6.4659 | 0.0 | 34 | 4.4289 | |
|
| 3.434 | 0.0 | 35 | 4.4173 | |
|
| 3.9518 | 0.0 | 36 | 4.4085 | |
|
| 3.0758 | 0.0 | 37 | 4.4008 | |
|
| 3.6492 | 0.0 | 38 | 4.3930 | |
|
| 4.0352 | 0.0 | 39 | 4.3857 | |
|
| 5.6527 | 0.0 | 40 | 4.3799 | |
|
| 4.233 | 0.0 | 41 | 4.3747 | |
|
| 5.4082 | 0.0 | 42 | 4.3702 | |
|
| 5.1255 | 0.0 | 43 | 4.3661 | |
|
| 4.4567 | 0.0 | 44 | 4.3622 | |
|
| 4.1874 | 0.0 | 45 | 4.3587 | |
|
| 4.3441 | 0.0 | 46 | 4.3555 | |
|
| 4.1636 | 0.0 | 47 | 4.3524 | |
|
| 4.3146 | 0.0 | 48 | 4.3495 | |
|
| 4.6414 | 0.0 | 49 | 4.3473 | |
|
| 4.3666 | 0.0 | 50 | 4.3451 | |
|
| 3.8627 | 0.0 | 51 | 4.3427 | |
|
| 4.5875 | 0.0 | 52 | 4.3406 | |
|
| 6.0364 | 0.0 | 53 | 4.3387 | |
|
| 4.5669 | 0.0 | 54 | 4.3369 | |
|
| 4.5585 | 0.0 | 55 | 4.3353 | |
|
| 2.7858 | 0.0 | 56 | 4.3340 | |
|
| 4.1845 | 0.0 | 57 | 4.3329 | |
|
| 4.4489 | 0.0 | 58 | 4.3319 | |
|
| 5.3263 | 0.0 | 59 | 4.3311 | |
|
| 5.3856 | 0.0 | 60 | 4.3306 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.8.2 |
|
- Transformers 4.38.0 |
|
- Pytorch 2.2.1+cu121 |
|
- Datasets 2.17.0 |
|
- Tokenizers 0.15.2 |
