wikiann-xlm-v-base

This model is a fine-tuned version of facebook/xlm-v-base on the Spanish and Guarani sets of unimelb-nlp/wikiann dataset. It achieves the following results on the evaluation set:

• Loss: 0.1872
• Precision: 0.9064
• Recall: 0.9120
• F1: 0.9092
• Accuracy: 0.9525

Model description

More information needed

Intended uses & limitations

• NER (PER, LOC, ORG)
• Spanish, Guarani & Jopará languages

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: 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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.255	1.0	1764	0.1951	0.8888	0.9017	0.8952	0.9465
0.1882	2.0	3528	0.1656	0.9014	0.9112	0.9063	0.9544
0.1454	3.0	5292	0.1599	0.9070	0.9191	0.9130	0.9575

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1