metadata
tags:
- generated_from_trainer
model-index:
- name: petro-twitter-assistant-30ep-large
results: []
widget:
- text: Nuestro gobierno es
- text: El pacto de la picota es
datasets:
- jhonparra18/petrogustavo-tweets
language:
- es
petro-twitter-assistant-30ep-large
This model is a fine-tuned version of flax-community/gpt-2-spanish on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 3.3050
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: 0.0001
- train_batch_size: 22
- eval_batch_size: 22
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 44
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.4824 | 0.73 | 1000 | 3.3659 |
| 3.2393 | 1.46 | 2000 | 3.2667 |
| 3.113 | 2.19 | 3000 | 3.2174 |
| 3.0121 | 2.92 | 4000 | 3.1767 |
| 2.9076 | 3.65 | 5000 | 3.1619 |
| 2.8358 | 4.38 | 6000 | 3.1665 |
| 2.7708 | 5.11 | 7000 | 3.1874 |
| 2.6846 | 5.84 | 8000 | 3.1695 |
| 2.6092 | 6.57 | 9000 | 3.2114 |
| 2.5487 | 7.3 | 10000 | 3.2517 |
| 2.4942 | 8.04 | 11000 | 3.3050 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.12.0
- Datasets 2.1.0
- Tokenizers 0.12.1