xlm-r-base-amazon-massive-domain
This model is a fine-tuned version of xlm-roberta-base on the Amazon Massive dataset (only en-US subset). It achieves the following results on the evaluation set:
- Loss: 0.3788
- Accuracy: 0.9213
- F1: 0.9213
Model description
Domain classifier trained from Amazon Massive dataset.
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 |
|---|---|---|---|---|---|
| 1.382 | 1.0 | 720 | 0.4533 | 0.8795 | 0.8795 |
| 0.4598 | 2.0 | 1440 | 0.3448 | 0.9026 | 0.9026 |
| 0.2547 | 3.0 | 2160 | 0.3762 | 0.9065 | 0.9065 |
| 0.1986 | 4.0 | 2880 | 0.3748 | 0.9139 | 0.9139 |
| 0.1358 | 5.0 | 3600 | 0.3788 | 0.9213 | 0.9213 |
Framework versions
- Transformers 4.22.1
- Pytorch 1.12.1+cu113
- Datasets 2.5.1
- Tokenizers 0.12.1
Citation
@article{kubis2023back,
title={Back Transcription as a Method for Evaluating Robustness of Natural Language Understanding Models to Speech Recognition Errors},
author={Kubis, Marek and Sk{\'o}rzewski, Pawe{\l} and Sowa{\'n}ski, Marcin and Zi{\k{e}}tkiewicz, Tomasz},
journal={arXiv preprint arXiv:2310.16609},
year={2023}
eprint={2310.16609},
archivePrefix={arXiv},
}