Update app.py

app.py

@@ -1,39 +1,35 @@
 import streamlit as st
-from transformers import pipeline
+from transformers import BertTokenizer, BertForSequenceClassification
+import torch
 
-# Load the sentiment analysis pipeline
-pipe = pipeline("text-classification", model="finiteautomata/bertweet-base-sentiment-analysis")
+# Load the fine-tuned model and tokenizer
+model_path = 'path/to/your/fine-tuned/model'
+tokenizer = BertTokenizer.from_pretrained(model_path)
+model = BertForSequenceClassification.from_pretrained(model_path)
 
-# Define a function to perform sentiment analysis
+# Set the model to evaluation mode
+model.eval()
+
+# Function to perform sentiment analysis
 def analyze_sentiment(text):
-    try:
-        results = pipe(text)
-        sentiment = results[0]['label']
-        confidence = results[0]['score']
-        return sentiment, confidence
-    except Exception as e:
-        return "ERROR", 0.0  # Handle errors gracefully
+    inputs = tokenizer(text, return_tensors='pt', truncation=True, padding=True)
+    with torch.no_grad():
+        outputs = model(**inputs)
+    logits = outputs.logits
+    predicted_class = logits.argmax().item()
+    return predicted_class
 
 # Create a Streamlit app
 st.title("Sentiment Analysis App")
 
 # Get the user input
-text = st.text_input("Enter text for sentiment analysis")
-
-if text:
-    # Perform sentiment analysis
-    sentiment, confidence = analyze_sentiment(text)
-
-    # Set a confidence threshold
-    confidence_threshold = 0.5
+text = st.text_area("Enter text for sentiment analysis")
 
-    # Display the output based on the sentiment and confidence
-    if confidence >= confidence_threshold:
-        if sentiment == "POSITIVE":
-            st.success(f"Sentiment: Positive, Confidence: {confidence:.2f}")
-        elif sentiment == "NEGATIVE":
-            st.error(f"Sentiment: Negative, Confidence: {confidence:.2f}")
-        else:
-            st.info(f"Sentiment: Neutral, Confidence: {confidence:.2f}")
+if st.button("Analyze Sentiment"):
+    if text:
+        sentiment_class = analyze_sentiment(text)
+        sentiment_labels = ['Negative', 'Neutral', 'Positive']  # Adjust as needed
+        sentiment = sentiment_labels[sentiment_class]
+        st.write(f"Sentiment: {sentiment}")
     else:
-        st.warning("Low confidence. Sentiment result may not be reliable.")
+        st.warning("Please enter text for analysis.")