from transformers import AutoModelForSequenceClassification
from transformers import TFAutoModelForSequenceClassification
from transformers import AutoTokenizer, AutoConfig
import numpy as np
from scipy.special import softmax
import gradio as gr

#model_path=f'sotseth/output'
model_path=f'https://huggingface.co/sotseth/output'

model=AutoModelForSequenceClassification.from_pretrained(model_path)
#tokenizer=AutoTokenizer.from_pretrained(f"cardiffnlp/twitter-roberta-base-sentiment-latest")
tokenizer=AutoTokenizer.from_pretrained(model_path)

#from huggingface_hub import notebook_login
#notebook_login()

#tokenizer.push_to_hub('sotseth/output')

def predict_tweet(tweet):

    inputs = tokenizer(tweet, return_tensors="pt", padding=True)
    outputs = model(**inputs)
    probs = outputs.logits.softmax(dim=-1)
    sentiment_classes = ['Negative', 'Neutral', 'Positive']

    return {sentiment_classes[i]: float(probs[0, i]) for i in range(len(sentiment_classes))}

iface=gr.Interface(
    fn=predict_tweet,
    inputs="text",
    outputs="label",
    title="Vaccine Sentiment Classifier",
    description="Enter your thought on vaccines",
    examples=[
        ["Vaccines are a game-changer in addressing public health"],
        ["Vaccines are profit making"],
        ["Vaccines are dangerous"]
    ]

)

iface.launch(share=True)