MiVaCod/Misogynystic_correction

1# import torch2# from transformers import AutoModelForSeq2SeqLM, AutoTokenizer3# import gradio as gr4 5# # Load your custom model and tokenizer6# model_name = "MiVaCod/mbart-neutralization"7# tokenizer = AutoTokenizer.from_pretrained(model_name)8# model = AutoModelForSeq2SeqLM.from_pretrained(model_name)9 10# # Function to correct sentences11# def predict(sentence):12#     inputs = tokenizer.encode("correction: " + sentence, return_tensors="pt", max_length=512, truncation=True)13#     outputs = model.generate(inputs, max_length=128, num_beams=4, early_stopping=True)14#     corrected_sentence = tokenizer.decode(outputs[0], skip_special_tokens=True)15#     return corrected_sentence16 17# # Gradio Interface18# iface = gr.Interface(19#     fn=correct_sentence,20#     inputs="text",21#     outputs="text",22#     title="Sentence Correction",23#     description="Enter a sentence to be corrected:",24#     theme="compact"25# )26 27# # Launch the interface28# gr.Interface(fn=predict, inputs=gr.inputs.Textbox, outputs=gr.outputs.Textbox).launch(share=False)29 30from transformers import MBartForConditionalGeneration, MBart50Tokenizer31import gradio as grad32 33model_name = "MiVaCod/mbart-neutralization"34text2text_tkn= MBart50Tokenizer.from_pretrained(model_name)35mdl = MBartForConditionalGeneration.from_pretrained(model_name)36 37def text2text_paraphrase(sentence1):38     inp1 = "rte sentence1: "+sentence139     enc = text2text_tkn(inp1, return_tensors="pt")40     tokens = mdl.generate(**enc)41     response=text2text_tkn.batch_decode(tokens)42     return response43 44sent1=grad.Textbox(lines=1, label="Frase misógina", placeholder="Introduce una frase misógina")45out=grad.Textbox(lines=1, label="Frase corregida")46grad.Interface(text2text_paraphrase, inputs=[sent1], outputs=out).launch()47