app.py

import streamlit as st 
import numpy as np
import torch
import transformers


@st.cache_resource
def get_model():
    model = transformers.AutoModelForSequenceClassification.from_pretrained("Kwasiasomani/Finetuned-Distilbert-base-model")
    tokenizer = transformers.AutoTokenizer.from_pretrained("Kwasiasomani/Finetuned-Distilbert-base-model")
    return tokenizer,model

tokenizer, model = get_model()
button = st.button('analyze')
user_input = st.text_area('''
                          How Positive or Negative is your Text?,Enter some text and we'll tell you if it has a positive, negative, or neutral sentiment!''')

# Define the Helper function    
label = {

    0: 'Negative',
    1: 'Neutral',
    2: 'Positive'
}

if user_input and button:
    test_input = tokenizer([user_input],return_tensors='pt')
    
    # Test output
    output = model(**test_input)
    st.write('Logits:',output.logits)
    predicted_class = np.argmax(output.logits.detach().numpy())
    st.write('prediction:',label[predicted_class[0]])