---
datasets:
- squad_it
language:
- it
metrics:
- f1
- exact_match
pipeline_tag: question-answering
tags:
- question-answering
---
# bert-italian-xxl-cased_squad-it_v1

BERT-based model for extractive Question-Answering.

## Model Details
### Model Description

- **Developed by:** fabgraziano
- **Model type:** Question-Answering
- **Language(s) (NLP):** Italian
- **Finetuned from model:** [dbmdz/bert-base-italian-xxl-cased](https://huggingface.co/dbmdz/bert-base-italian-xxl-cased)
- **Datased used:** [squad_it](https://huggingface.co/datasets/squad_it)



## Training Details
#### Training Hyperparameters

- **Training regime:** fp16 mixed precision
- **MAX_LENGTH:** 384
- **STRIDE:** 128
- **BATCH_SIZE:** 8
- **LEARNING_RATE:** 2e-5
- **EPOCHS:** 3



## Evaluation
#### Metrics

- **EM:** 63.87
- **F1:** 75.34