srebora/MultItLit

Model Card

The model is a sequence classifier, finetuned on a corpus of Italian literature sentences to identify representations of multilingualism.

Model Details

Model Description

With this model, we offer a tool capable of automatically extract from Italian literary texts those sentences that contain en explicit representation of multilingualism. This can help researchers interested in multilingualism and plurilingualism in the literary context.

• Developed by: Simone Rebora
• Model type: AutoModelForSequenceClassification
• Finetuned from: dbmdz/bert-base-italian-xxl-cased
• Repository: [More Information Needed]
• Paper: [More Information Needed]

How to Get Started with the Model

Use the code below to get started with the model.

Important The model does not include a tokenizer, we recomend using the one from dbmdz/bert-base-italian-xxl-cased.

# import libraries
import torch
from transformers import AutoModelForSequenceClassification, AutoTokenizer

# define function for prediction
def predict_text_class(input_text, labels, model, tokenizer):

    input_ids = tokenizer(input_text, truncation=True, padding=True, return_tensors="pt")

    # Perform inference
    with torch.no_grad():
        output = model(**input_ids)

    predicted_labels = output.logits.argmax(dim=1)

    return labels[predicted_labels.item()]

# Define the tokenizer
checkpoint = "dbmdz/bert-base-italian-xxl-cased"

# Define the path to your model
checkpoint_path = "<MODEL_PATH>"

# Load the tokenizer
tokenizer = AutoTokenizer.from_pretrained(checkpoint)

# Load the model
model_new = AutoModelForSequenceClassification.from_pretrained(checkpoint_path)

# define sentence and labels
sentence = "La lingua italiana e la lingua francese convivevano bene assieme"
labels = ['No', 'Sì']

# make a prediction
predict_text_class(sentence, labels=labels, model=model_new, tokenizer=tokenizer)

Training Details

Training Data

[More Information Needed]

Training Hyperparameters

train_ratio = 0.8 validation_ratio = 0.2 batch_size: 16 num_train_epochs=3 metric_for_best_model='accuracy' weight_decay=0.01

Evaluation

              precision    recall  f1-score   support

          No       0.97      0.89      0.93       302
          Sì       0.83      0.96      0.89       167

    accuracy                           0.91       469
   macro avg       0.90      0.92      0.91       469
weighted avg       0.92      0.91      0.92       469

Citation

BibTeX:

[More Information Needed]

APA:

[More Information Needed]