Update app.py

app.py

@@ -1,29 +1,28 @@
 import streamlit as st
-from transformers import AutoModelForSequenceClassification, AutoTokenizer
+from tensorflow.keras.models import load_model
+from tensorflow.keras.preprocessing.text import Tokenizer  # Assuming you used Tokenizer during training
 
-# Replace with your actual model ID or access token from Hugging Face Hub
-model_id = "your-model-id"
+# Load the pre-trained sentiment analysis model
+model = load_model("sentiment_analysis_model.h5")
 
-# Load the pre-trained sentiment analysis model and tokenizer from Hugging Face Hub
-model = AutoModelForSequenceClassification.from_pretrained(model_id)
-tokenizer = AutoTokenizer.from_pretrained(model_id)
+# Define a tokenizer (replace with your actual tokenization logic from training)
+tokenizer = Tokenizer(num_words=5000)  # Adjust vocabulary size based on your training
+
+def preprocess_text(text):
+  # Preprocess text based on your training process (e.g., tokenization)
+  tokens = tokenizer.texts_to_sequences([text])
+  padded_sequence = pad_sequences(tokens, maxlen=200)  # Adjust max length based on your training
+  return padded_sequence
 
 def classify_sentiment(text):
   """
   Function to preprocess text, make predictions using the loaded model,
   and return the predicted sentiment.
   """
-  # Preprocess text (tokenization)
-  encoded_text = tokenizer(text, return_tensors="pt")
-
-  # Make prediction using the loaded model
-  output = model(**encoded_text)
-  predictions = output.logits.argmax(-1)
-
-  # Map predicted class label to sentiment category
+  preprocessed_text = preprocess_text(text)
+  prediction = model.predict(preprocessed_text)
   sentiment_mapping = {0: "Negative", 1: "Neutral", 2: "Positive"}
-  sentiment = sentiment_mapping[predictions.item()]
-
+  sentiment = sentiment_mapping[prediction.argmax()]
   return sentiment
 
 def main():