Electricidad (base) fine-tuned on sst2-es-mt for Spanish Sentiment Analysis 👍👎
This model is a fine-tuned version of mrm8488/electricidad-base-discriminator on the sst2-es-mt dataset. A dataset created using Neural Machine Translation on original SST2 (English) dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4377
- Accuracy: 0.9014
- F1: 0.9034
Usage
from transformers import pipeline
model_ckpt = "mrm8488/electricidad-base-finetuned-sst2-es"
classifier = pipeline("sentiment-analysis", model=model_ckpt)
classifier("Here your text in Spanish!")
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 0.2521 | 1.0 | 2105 | 0.2837 | 0.9002 | 0.9019 |
| 0.1694 | 2.0 | 4210 | 0.3175 | 0.8933 | 0.8920 |
| 0.1245 | 3.0 | 6315 | 0.3606 | 0.8945 | 0.8987 |
| 0.0934 | 4.0 | 8420 | 0.4419 | 0.9002 | 0.9037 |
| 0.069 | 5.0 | 10525 | 0.4377 | 0.9014 | 0.9034 |
Framework versions
- Transformers 4.21.2
- Pytorch 1.12.1+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1