Update app.py

app.py

@@ -3,8 +3,8 @@ import torch
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import matplotlib.pyplot as plt
 
-# Initialize tokenizer and model once to avoid reloading them on every interaction
-@st.cache(allow_output_mutation=True)
+# Cache the loading of the tokenizer and model to speed up the app
+@st.cache_resource
 def load_model():
     tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
     model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
@@ -16,24 +16,29 @@ st.title("Sentiment Analysis App")
 
 text = st.text_input("Enter text to analyze:")
 if st.button("Analyze"):
-    encoding = tokenizer.encode_plus(text, return_tensors="pt", padding=True, truncation=True)
-    input_ids = encoding["input_ids"]
-    attention_mask = encoding["attention_mask"]
+    if text:  # Ensure that the text is not empty
+        encoding = tokenizer.encode_plus(text, return_tensors="pt", padding=True, truncation=True)
+        input_ids = encoding["input_ids"]
+        attention_mask = encoding["attention_mask"]
 
-    with torch.no_grad():
-        output = model(input_ids, attention_mask)
-    prediction = int(torch.argmax(output.logits))
+        with torch.no_grad():
+            output = model(input_ids, attention_mask)
+        prediction = int(torch.argmax(output.logits))
 
-    # Detailed sentiment output
-    sentiment = ["Negative", "Neutral", "Positive"][prediction]
-    st.write(f"Sentiment: {sentiment}")
+        # Define sentiments
+        sentiments = ["Negative", "Neutral", "Positive"]
+        sentiment = sentiments[prediction]
+        st.write(f"Sentiment: {sentiment}")
 
-    values = output.logits.squeeze().tolist()  # Flatten the logits tensor to a list
-    labels = ["Negative", "Neutral", "Positive"]
-    
-    # Plotting
-    fig, ax = plt.subplots()
-    ax.bar(labels, values, color=['red', 'blue', 'green'])
-    ax.set_title("Sentiment Analysis Scores")
-    ax.set_ylabel("Score")
-    st.pyplot(fig)
+        # Flatten the logits tensor to a list and check dimensions
+        values = output.logits.squeeze().tolist()
+        if len(values) != len(sentiments):
+            st.error(f"Mismatch in the number of sentiments and values. Expected {len(sentiments)}, got {len(values)}")
+        else:
+            fig, ax = plt.subplots()
+            ax.bar(sentiments, values, color=['red', 'blue', 'green'])
+            ax.set_title("Sentiment Analysis Scores")
+            ax.set_ylabel("Score")
+            st.pyplot(fig)
+    else:
+        st.error("Please enter some text to analyze.")