Update app.py

app.py

@@ -3,26 +3,24 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 from peft import PeftModel
 import streamlit as st
 
-# Load base model and tokenizer
+# Load tokenizer
 base_model = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 tokenizer = AutoTokenizer.from_pretrained(base_model)
 
-# Load base model in CPU-only mode
+# Load base model in empty (meta) state and move to CPU
 model = AutoModelForCausalLM.from_pretrained(
     base_model,
-    device_map="auto",  # Use 'auto' or manually move later
     torch_dtype=torch.float32,
-    low_cpu_mem_usage=True  # Ensures meta device usage
+    low_cpu_mem_usage=True,
+    device_map="auto"
 )
-
-# Move the model to CPU safely
 model = model.to_empty(device=torch.device("cpu"))
 
-# Now load the LoRA adapter
-from peft import PeftModel
+# Load LoRA adapter and move to CPU
 model = PeftModel.from_pretrained(model, "lora_adapter", device_map="cpu")
 model.eval()
 
+# Format prompt for Python tutoring
 def format_prompt(instruction):
     return f"""### SYSTEM:
 You are a helpful and expert Python programming tutor.
@@ -30,18 +28,13 @@ You only answer questions related to Python programming.
 If the question is unrelated to Python, say:
 "Sorry, I can only answer Python-related questions."
 
-### USER:
-What is a tuple in Python?
-
-### ASSISTANT:
-A tuple in Python is an ordered and immutable collection of elements. Tuples are defined using parentheses, like (1, 2, 3). They are useful when you want to store multiple items and ensure they cannot be changed.
-
 ### USER:
 {instruction}
 
 ### ASSISTANT:
 """
 
+# Generate answer
 def chat(instruction):
     prompt = format_prompt(instruction)
     inputs = tokenizer(prompt, return_tensors="pt").to(model.device)