Create app.py

app.py

@@ -0,0 +1,59 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch
+import numpy as np
+
+# ✅ Paths to your hosted models on Hugging Face Hub
+MODEL_PATHS = [
+    "Basavians/youtube-comment-sentiment-1",
+    "Basavians/youtube-comment-sentiment-2",
+    "Basavians/youtube-comment-sentiment-3"
+]
+
+# Load models and tokenizers (once at startup)
+models = []
+tokenizers = []
+for path in MODEL_PATHS:
+    tokenizer = AutoTokenizer.from_pretrained(path)
+    model = AutoModelForSequenceClassification.from_pretrained(path)
+    model.eval()
+    tokenizers.append(tokenizer)
+    models.append(model)
+
+# Class labels (update if different)
+LABELS = ["negative", "neutral", "positive"]
+
+def predict_sentiment(text):
+    if not text.strip():
+        return "Please enter some text", None
+
+    probs = []
+    for model, tokenizer in zip(models, tokenizers):
+        inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
+        with torch.no_grad():
+            outputs = model(**inputs)
+            logits = outputs.logits
+            prob = torch.nn.functional.softmax(logits, dim=-1)
+            probs.append(prob.numpy())
+
+    # 🎯 Ensemble by averaging probabilities
+    avg_prob = np.mean(probs, axis=0)
+    pred_class = int(np.argmax(avg_prob, axis=1)[0])
+    pred_label = LABELS[pred_class]
+    confidence = float(avg_prob[0][pred_class])
+
+    return pred_label, {label: float(avg_prob[0][i]) for i, label in enumerate(LABELS)}
+
+# Gradio UI
+demo = gr.Interface(
+    fn=predict_sentiment,
+    inputs=gr.Textbox(lines=4, placeholder="Paste a YouTube comment here..."),
+    outputs=[
+        gr.Label(num_top_classes=1, label="Predicted Sentiment"),
+        gr.Label(label="Confidence Scores"),
+    ],
+    title="YouTube Comment Sentiment Classifier (Ensemble)",
+    description="Enter a comment to see sentiment prediction based on an ensemble of 3 models."
+)
+
+demo.launch()