Create app.py

app.py

@@ -0,0 +1,31 @@
+import gradio as gr
+import numpy as np
+import json
+from sentence_transformers import SentenceTransformer, util
+
+# Load precomputed QA pairs and embeddings
+with open("qa.json", encoding="utf-8") as f:
+    QA_PAIRS = json.load(f)
+
+QA_EMBEDDINGS = np.load("qa.npy")
+hf_model = SentenceTransformer('all-MiniLM-L6-v2')
+
+def find_similar(question):
+    user_embedding = hf_model.encode(question, convert_to_tensor=True)
+    similarities = util.cos_sim(user_embedding, QA_EMBEDDINGS)[0]
+    best_score = float(similarities.max())
+    best_index = int(similarities.argmax())
+
+    if best_score >= 0.6:
+        return QA_PAIRS[best_index]["answer"]
+    else:
+        return "Sorry, I could not find a relevant answer."
+
+iface = gr.Interface(
+    fn=find_similar,
+    inputs=gr.Textbox(label="Your Question"),
+    outputs=gr.Textbox(label="Answer"),
+    title="Similarity Service"
+)
+
+iface.launch()