fedcsis-intent_baseline-xlm_r-all (en,es,pl)

This model is a fine-tuned version of xlm-roberta-base on the leyzer-fedcsis dataset. It was trained to predict intents and it was trained and evaluated on all three languages.

Results on test set:

• Accuracy: 0.950414

It achieves the following results on the evaluation set:

• Accuracy: 0.975555

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: 16
• 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
0.7642	1.0	2552	0.6074	0.8696	0.8696
0.2846	2.0	5104	0.2371	0.9464	0.9464
0.1251	3.0	7656	0.1486	0.9662	0.9662
0.0749	4.0	10208	0.1226	0.9731	0.9731
0.0503	5.0	12760	0.1159	0.9756	0.9756

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2

Citation

If you use this model, please cite the following:

@inproceedings{kubis2023caiccaic,
    author={Marek Kubis and Paweł Skórzewski and Marcin Sowański and Tomasz Ziętkiewicz},
    pages={1319–1324},
    title={Center for Artificial Intelligence Challenge on Conversational AI Correctness},
    booktitle={Proceedings of the 18th Conference on Computer Science and Intelligence Systems},
    year={2023},
    doi={10.15439/2023B6058},
    url={http://dx.doi.org/10.15439/2023B6058},
    volume={35},
    series={Annals of Computer Science and Information Systems}
}