src/inference_t5.py

"""
 Allows to predict the summary for a given entry text
"""
import re
import string
import os
os.environ['TRANSFORMERS_CACHE'] = './.cache'

import contractions
import torch
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer


def clean_text(texts: str) -> str:
    texts = texts.lower()
    texts = texts.translate(str.maketrans("", "", string.punctuation))
    texts = re.sub(r"\n", " ", texts)
    return texts



def inference_t5(text: str) -> str:
    """
    Predict the summary for an input text
    --------
    Parameter
        text: str
            the text to sumarize
    Return
        str
            The summary for the input text
    """

    # On défini les paramètres d'entrée pour le modèle
    text = clean_text(text)
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    hf_token = "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"
    tokenizer = AutoTokenizer.from_pretrained("Linggg/t5_summary", use_auth_token=hf_token )
    # load local model
    model = (AutoModelForSeq2SeqLM
             .from_pretrained("Linggg/t5_summary", use_auth_token = hf_token )
             .to(device))


    text_encoding = tokenizer(
        text,
        max_length=1024,
        padding="max_length",
        truncation=True,
        return_attention_mask=True,
        add_special_tokens=True,
        return_tensors="pt",
    )
    generated_ids = model.generate(
        input_ids=text_encoding["input_ids"],
        attention_mask=text_encoding["attention_mask"],
        max_length=128,
        num_beams=8,
        length_penalty=0.8,
        early_stopping=True,
    )

    preds = [
        tokenizer.decode(
            gen_id, skip_special_tokens=True, clean_up_tokenization_spaces=True
        )
        for gen_id in generated_ids
    ]
    return "".join(preds)


# if __name__ == "__main__":
#     text = input('Entrez votre phrase à résumer : ')
#     print('summary:', inferenceAPI_T5(text))