Create app.py

app.py

@@ -0,0 +1,122 @@
+import streamlit as st
+import torch
+import joblib
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+
+# -------------------------------------
+# 🎨 STREAMLIT PAGE CONFIG
+# -------------------------------------
+st.set_page_config(
+    page_title="StackOverflow Tag Predictor",
+    page_icon="🎯",
+    layout="centered",
+)
+
+# -------------------------------------
+# 🌈 CUSTOM CSS FOR BEAUTIFUL UI
+# -------------------------------------
+st.markdown("""
+<style>
+body {
+    background-color: #F2F2F7;
+}
+.big-title {
+    font-size: 40px;
+    font-weight: 900;
+    text-align: center;
+    margin-bottom: -10px;
+    color: #4A4AFC;
+}
+.subtitle {
+    text-align: center;
+    color: #666;
+    font-size: 18px;
+}
+.result-tag {
+    background-color: #4A4AFC;
+    padding: 10px 18px;
+    border-radius: 12px;
+    color: white;
+    display: inline-block;
+    font-size: 20px;
+    margin: 5px;
+    animation: fadeIn 0.6s ease-out;
+}
+@keyframes fadeIn {
+    from {opacity: 0; transform: translateY(10px);}
+    to {opacity: 1; transform: translateY(0);}
+}
+</style>
+""", unsafe_allow_html=True)
+
+# -------------------------------------
+# 📦 LOAD MODEL + TOKENIZER
+# -------------------------------------
+@st.cache_resource
+def load_model():
+    model = AutoModelForSequenceClassification.from_pretrained(".")
+    tokenizer = AutoTokenizer.from_pretrained(".")
+    return model, tokenizer
+
+model, tokenizer = load_model()
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model = model.to(device)
+
+# -------------------------------------
+# 🔤 LOAD LABEL ENCODER
+# -------------------------------------
+label_encoder = joblib.load("label_encoder.joblib")
+id2label = {i: label for i, label in enumerate(label_encoder.classes_)}
+
+# -------------------------------------
+# 🔮 PREDICTION FUNCTION
+# -------------------------------------
+def predict_tag(text):
+    encoding = tokenizer(
+        text,
+        truncation=True,
+        padding=True,
+        max_length=128,
+        return_tensors="pt"
+    )
+    encoding = {k: v.to(device) for k, v in encoding.items()}
+
+    with torch.no_grad():
+        outputs = model(**encoding)
+
+    pred_id = torch.argmax(outputs.logits, dim=-1).item()
+    tag = id2label[pred_id]
+    confidence = torch.softmax(outputs.logits, dim=-1).max().item()
+
+    return tag, confidence
+
+# -------------------------------------
+# 🖥️ UI LAYOUT
+# -------------------------------------
+st.markdown("<p class='big-title'>🎯 StackOverflow Tag Predictor</p>", unsafe_allow_html=True)
+st.markdown("<p class='subtitle'>Powered by DistilBERT • Predicts the most likely tag from a question title</p>", unsafe_allow_html=True)
+
+st.write("")
+
+user_input = st.text_area(
+    "💬 Enter a StackOverflow question title:",
+    height=120,
+    placeholder="Example: \"How to fix NullPointerException in Java?\""
+)
+
+if st.button("🔍 Predict Tag", use_container_width=True):
+    if user_input.strip() == "":
+        st.warning("Please enter a question title.")
+    else:
+        with st.spinner("Analyzing text using AI magic... ✨"):
+            tag, confidence = predict_tag(user_input)
+
+        st.success("Prediction complete!")
+
+        st.markdown(f"<div class='result-tag'>{tag}</div>", unsafe_allow_html=True)
+        st.markdown(
+            f"### 🔥 Confidence: **{confidence*100:.2f}%**"
+        )
+
+        st.info("Try another title!")