mt5-multilingual-sentiment

This model predicts the sentiment('Negative'/'Positive') for the input sentence. It is fine-tuned mt5-small

The present model supports 6 languages -

1. English
2. Hindi
3. German
4. Korean
5. Japanese
6. Portuguese

Here is how to use this model

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
model = AutoModelForSeq2SeqLM.from_pretrained("Chirayu/mt5-multilingual-sentiment")
tokenizer = AutoTokenizer.from_pretrained("Chirayu/mt5-multilingual-sentiment")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = model.to(device)

def get_sentiment(text, num_beams=2,max_length=512, repetition_penalty=2.5, length_penalty=1, early_stopping=True,top_p=.95, top_k=50, num_return_sequences=1):
  
  input_ids = tokenizer.encode(
    text, return_tensors="pt", add_special_tokens=True
  )
  
  input_ids = input_ids.to(device)
  generated_ids = model.generate(
      input_ids=input_ids,
     
      num_beams=num_beams,
      max_length=max_length,
      repetition_penalty=repetition_penalty,
      length_penalty=length_penalty,
      early_stopping=early_stopping,
      top_p=top_p,
      top_k=top_k,
      num_return_sequences=num_return_sequences,
  )
  sentiment = [tokenizer.decode(generated_id,skip_special_tokens=True,clean_up_tokenization_spaces=True,) for generated_id in generated_ids]
  return sentiment