Update app.py

app.py

@@ -1,8 +1,7 @@
 """
 SchoolSpirit AI – Granite‑3.3‑2B chatbot Space
 ----------------------------------------------
-• Uses IBM Granite‑3.3‑2B‑Instruct (public, no access token).
-• Fits HF CPU Space (2‑B params, bfloat16).
+• Uses IBM Granite‑3.3‑2B‑Instruct (Apache‑2).
 • Keeps last MAX_TURNS exchanges.
 • “Clear chat” button resets context.
 • Robust error handling & logging.
@@ -17,7 +16,7 @@ from transformers import (
 )
 from transformers.utils import logging as hf_logging
 
-# ────────────────────────── Config ──────────────────────────────────────────
+# ─────────────── Config ────────────────────────────────────────────────────
 hf_logging.set_verbosity_error()
 LOG = hf_logging.get_logger("SchoolSpirit")
 
@@ -33,13 +32,11 @@ SYSTEM_MSG = (
     "say so and suggest contacting a human. Do not ask for personal data."
 )
 
-# ────────────────────────── Model Load ──────────────────────────────────────
+# ─────────────── Model Load ────────────────────────────────────────────────
 try:
     tok = AutoTokenizer.from_pretrained(MODEL_ID)
     model = AutoModelForCausalLM.from_pretrained(
-        MODEL_ID,
-        device_map="auto",
-        torch_dtype="auto",          # bfloat16/float16 under the hood
+        MODEL_ID, device_map="auto", torch_dtype="auto"
     )
     generator = pipeline(
         "text-generation",
@@ -55,20 +52,17 @@ except Exception as exc:  # noqa: BLE001
     generator = None
     LOG.error(MODEL_ERR)
 
-# ────────────────────────── Helpers ────────────────────────────────────────
+# ─────────────── Helpers ───────────────────────────────────────────────────
 def truncate(hist):
-    """Return last MAX_TURNS (u,a) pairs."""
     return hist[-MAX_TURNS:] if len(hist) > MAX_TURNS else hist
 
 
 def clean(text: str) -> str:
-    """Collapse whitespace; never return empty string."""
-    out = re.sub(r"\s+", " ", text.strip())
-    return out or "…"
+    return re.sub(r"\s+", " ", text.strip()) or "…"
 
-# ────────────────────────── Chat Callback ───────────────────────────────────
+# ─────────────── Chat Callback ─────────────────────────────────────────────
 def chat(history, user_msg):
-    history = list(history)  # Gradio ensures list of tuples
+    history = list(history)
 
     if MODEL_ERR:
         history.append((user_msg, MODEL_ERR))
@@ -84,7 +78,6 @@ def chat(history, user_msg):
 
     history = truncate(history)
 
-    # Build prompt
     prompt_lines = [SYSTEM_MSG]
     for u, a in history:
         prompt_lines += [f"User: {u}", f"AI: {a}"]
@@ -92,7 +85,7 @@ def chat(history, user_msg):
     prompt = "\n".join(prompt_lines)
 
     try:
-        completion = generator(prompt, truncate=4096)[0]["generated_text"]
+        completion = generator(prompt)[0]["generated_text"]
         reply = clean(completion.split("AI:", 1)[-1])
     except Exception as err:  # noqa: BLE001
         LOG.error(f"Inference error: {err}")
@@ -101,16 +94,16 @@ def chat(history, user_msg):
     history.append((user_msg, reply))
     return history, ""
 
-# ────────────────────────── Clear Chat ──────────────────────────────────────
+# ─────────────── Clear Chat ────────────────────────────────────────────────
 def clear_chat():
     return [], ""
 
-# ────────────────────────── UI Launch ───────────────────────────────────────
+# ─────────────── UI Launch ────────────────────────────────────────────────
 with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue")) as demo:
     gr.Markdown("# SchoolSpirit AI Chat")
-    chatbot  = gr.Chatbot()
-    msg_box  = gr.Textbox(placeholder="Ask me anything about SchoolSpirit AI…")
-    send_btn = gr.Button("Send")
+    chatbot   = gr.Chatbot(type="tuple")  # legacy tuple format
+    msg_box   = gr.Textbox(placeholder="Ask me anything about SchoolSpirit AI…")
+    send_btn  = gr.Button("Send")
     clear_btn = gr.Button("Clear Chat", variant="secondary")
 
     send_btn.click(chat, [chatbot, msg_box], [chatbot, msg_box])