Modelpublic

paragon-analytics/t5_para

sourceHugging Facemitupdated 4y agoView on Hugging Face

0likes8downloads

Model Card

This model takes text as input and returns the top five paraphrased versions of the input text. The T5 model is fine-tuned using persuasive ad transcripts.

Example usage:

python

import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("paragon-analytics/t5_para")  
model = AutoModelForSeq2SeqLM.from_pretrained("paragon-analytics/t5_para").to(device)

sentence = "This is something"

text =  "paraphrase: " + sentence + " </s>"

encoding = tokenizer.encode_plus(text,pad_to_max_length=True, return_tensors="pt")
input_ids, attention_masks = encoding["input_ids"].to("cuda"), encoding["attention_mask"].to("cuda")

outputs = model.generate(
    input_ids=input_ids, attention_mask=attention_masks,
    max_length=256,
    do_sample=True,
    top_k=120,
    top_p=0.95,
    early_stopping=True,
    num_return_sequences=5
)

for output in outputs:
    line = tokenizer.decode(output, skip_special_tokens=True,clean_up_tokenization_spaces=True)
    print(line)