modernbert-italian-finetuned-ner

This model is a fine-tuned version of DeepMount00/Italian-ModernBERT-base on tner/wikiann dataset. It achieves the following results on the evaluation set:

• Loss: 0.0422
• Precision: 0.9339
• Recall: 0.9452
• F1: 0.9395
• Accuracy: 0.9909

Model description

Token classification for italian language experiment, NER.

Example

from transformers import pipeline
ner_pipeline = pipeline("ner", model="nickprock/modernbert-italian-finetuned-ner", aggregation_strategy="simple")
text = "La sede storica della Olivetti è ad Ivrea"
output = ner_pipeline(text)

Intended uses & limitations

The model can be used on token classification, in particular NER. It is fine tuned on italian language.

Training and evaluation data

The dataset used is wikiann

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH 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
0.0277	1.0	11050	0.0324	0.9233	0.9362	0.9297	0.9899
0.0139	2.0	22100	0.0341	0.9327	0.9428	0.9377	0.9907
0.0052	3.0	33150	0.0422	0.9339	0.9452	0.9395	0.9909

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0