Update app.py

app.py

@@ -1,26 +1,18 @@
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig
+from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
 print("Loading model...")
 MODEL_ID = "newtechdevng/qwen-math-tutor"
 
 tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
-
-# Load and fix config — remove quantization completely
-config = AutoConfig.from_pretrained(MODEL_ID)
-if hasattr(config, "quantization_config"):
-    delattr(config, "quantization_config")  # ← delete it entirely
-
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_ID,
-    config        = config,
-    torch_dtype   = torch.float32,
-    low_cpu_mem_usage = True,
+    torch_dtype = torch.float16,
+    device_map  = "cpu",
 )
-model = model.to("cpu")
 model.eval()
-print("Model loaded!")
+print("✅ Model loaded!")
 
 def solve(question, history):
     messages = [