metadata
license: mit
base_model: roberta-base
tags:
- generated_from_trainer
metrics:
- accuracy
- precision
- recall
- f1
model-index:
- name: irony_es_United_States
results: []
irony_es_United_States
This model is a fine-tuned version of roberta-base on part of the MultiPICo dataset. It achieves the following results on the evaluation set:
- Loss: 0.0041
- Accuracy: 0.6530
- Precision: 0.4709
- Recall: 0.3716
- F1: 0.4154
Model description
The model is trained considering the annotation of Spanish-speaking annotators from the United States only, on instances in Spanish (all linguistic varieties). The US annotations from these annotators are aggregated using majority voting and then used to train the model.
Training and evaluation data
The model has been trained on the annotation from annotators from the US from the MultiPICo dataset (instances in Spanish). The data has been randomly split into a train and a validation set.
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 16
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 0.0044 | 1.0 | 124 | 0.0043 | 0.5266 | 0.3652 | 0.5780 | 0.4476 |
| 0.0043 | 2.0 | 248 | 0.0043 | 0.4414 | 0.3553 | 0.8394 | 0.4993 |
| 0.0043 | 3.0 | 372 | 0.0042 | 0.6377 | 0.4398 | 0.3349 | 0.3802 |
| 0.0042 | 4.0 | 496 | 0.0041 | 0.4414 | 0.3513 | 0.8073 | 0.4896 |
| 0.004 | 5.0 | 620 | 0.0040 | 0.4840 | 0.3676 | 0.7706 | 0.4978 |
| 0.0036 | 6.0 | 744 | 0.0036 | 0.6499 | 0.4764 | 0.5550 | 0.5127 |
| 0.0031 | 7.0 | 868 | 0.0040 | 0.5297 | 0.3959 | 0.7936 | 0.5282 |
| 0.0024 | 8.0 | 992 | 0.0041 | 0.6530 | 0.4709 | 0.3716 | 0.4154 |
Framework versions
- Transformers 4.34.1
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.14.1