language_identification_model

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0297
• Accuracy: 0.9955

Model description

This model is a fine-tuned version of FacebookAI/xlm-roberta-base for multilingual language identification.

It can identify 20 different languages from text input. The model uses transfer learning and was trained using the Hugging Face Transformers library.

Intended uses & limitations

This model is intended to identify the language of text. It supports 20 languages. Its predictions may be less reliable for very short text, mixed-language text, or text containing spelling errors.

Training and evaluation data

The model was trained on the papluca/language-identification dataset from Hugging Face. The dataset contains 90,000 samples across 20 languages:

• Training: 70,000 samples
• Validation: 10,000 samples
• Testing: 10,000 samples

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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.0237	1.0	4375	0.0213	0.9964
0.0051	2.0	8750	0.0249	0.9962

Framework versions

• Transformers 5.22.1
• Pytorch 2.14.1+cu130
• Datasets 4.8.5
• Tokenizers 0.23.3