Create app.py

app.py

@@ -0,0 +1,117 @@
+import gradio as gr
+from huggingface_hub import InferenceClient
+import os
+import faiss
+import numpy as np
+from sentence_transformers import SentenceTransformer
+import pickle
+
+# --- Configuration ---
+MODEL_NAME = "openai/gpt-oss-20b"
+SECURE_HF_TOKEN = os.environ.get("HF_TOKEN")
+
+if not SECURE_HF_TOKEN:
+    raise ValueError("HF_TOKEN environment variable not set. Add a Secret in Space settings.")
+
+client = InferenceClient(token=SECURE_HF_TOKEN, model=MODEL_NAME)
+embedder = SentenceTransformer("all-MiniLM-L6-v2")
+
+# --- Notes folder (TXT training files) ---
+notes_folder = "notes"   # <== create this folder in repo and upload TXT files inside
+os.makedirs(notes_folder, exist_ok=True)
+cache_file = os.path.join(notes_folder, "embeddings_cache.pkl")
+
+chunks, sources = [], []
+
+# --- Load from cache if exists ---
+if os.path.exists(cache_file):
+    with open(cache_file, "rb") as f:
+        chunks, sources, embeddings = pickle.load(f)
+    dim = embeddings.shape[1]
+    index = faiss.IndexFlatL2(dim)
+    index.add(np.array(embeddings).astype("float32"))
+else:
+    # --- Read all TXT files from notes/ ---
+    for file in os.listdir(notes_folder):
+        if file.endswith(".txt"):
+            subject = os.path.splitext(file)[0]
+            with open(os.path.join(notes_folder, file), "r", encoding="utf-8", errors="ignore") as f:
+                text = f.read()
+
+            # Split into chunks
+            file_chunks = [text[i:i+500] for i in range(0, len(text), 500)]
+            chunks.extend(file_chunks)
+            sources.extend([subject] * len(file_chunks))
+
+    if chunks:
+        embeddings = embedder.encode(chunks)
+        dim = embeddings.shape[1]
+        index = faiss.IndexFlatL2(dim)
+        index.add(np.array(embeddings).astype("float32"))
+
+        with open(cache_file, "wb") as f:
+            pickle.dump((chunks, sources, embeddings), f)
+    else:
+        index = None
+
+# --- Respond function ---
+def respond(message, history: list, system_message, max_tokens, temperature, top_p):
+    context = ""
+    source_names = set()
+
+    if index is not None and len(chunks) > 0:
+        query_emb = embedder.encode([message])
+        query_emb = np.array(query_emb).astype("float32")
+        k = min(3, len(chunks))
+        D, I = index.search(query_emb, k=k)
+        retrieved_chunks = [chunks[i] for i in I[0] if i != -1]
+        retrieved_sources = [sources[i] for i in I[0] if i != -1]
+
+        if retrieved_chunks:
+            context = "\n".join(retrieved_chunks)
+            source_names.update(retrieved_sources)
+
+    messages = [{"role": "system", "content": system_message}]
+    messages.extend(history)
+
+    source_text = ""
+    if source_names:
+        source_text = "Sources: " + ", ".join(sorted(source_names)) + "\n\n"
+
+    prompt_content = f"{source_text}Answer using the following notes if relevant:\n{context}\n\nQuestion: {message}"
+    messages.append({"role": "user", "content": prompt_content})
+
+    response = ""
+    for message_chunk in client.chat_completion(
+        messages,
+        max_tokens=max_tokens,
+        stream=True,
+        temperature=temperature,
+        top_p=top_p,
+    ):
+        choices = message_chunk.choices
+        token = ""
+        if len(choices) and choices[0].delta.content:
+            token = choices[0].delta.content
+        response += token
+        yield response
+
+# --- Gradio Chat Interface ---
+chatbot = gr.ChatInterface(
+    respond,
+    type="messages",
+    additional_inputs=[
+        gr.Textbox(value="Hey, need help?", label="System message"),
+        gr.Slider(1, 5000, value=3000, step=1, label="Max new tokens"),
+        gr.Slider(0.1, 4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p"),
+    ],
+)
+
+# --- Launch (mobile-friendly, no sidebar) ---
+with gr.Blocks(css=".gradio-container {max-width: 800px; margin:auto;}") as demo:
+    gr.Markdown("<h2 style='text-align:center;'>📚 AI Tutor (Trained on Notes)</h2>")
+    chatbot.render()
+
+if __name__ == "__main__":
+    demo.launch()