Override quantization settings and remove bitsandbytes

app.py

@@ -5,11 +5,19 @@ import accelerate
 print("Accelerate version:", accelerate.__version__)
 
 # Load the original model
-original_model = AutoModelForCausalLM.from_pretrained("unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit",load_in_4bit=False)
+original_model = AutoModelForCausalLM.from_pretrained(
+    "unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit",
+    load_in_4bit=False,  # Disable 4-bit quantization
+    quantization_config=None  # Explicitly disable quantization
+)
 original_tokenizer = AutoTokenizer.from_pretrained("unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit")
 
 # Load the fine-tuned model
-fine_tuned_model = AutoModelForCausalLM.from_pretrained("kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1",load_in_4bit=False)
+fine_tuned_model = AutoModelForCausalLM.from_pretrained(
+    "kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1",
+    load_in_4bit=False,  # Disable 4-bit quantization
+    quantization_config=None  # Explicitly disable quantization
+)
 fine_tuned_tokenizer = AutoTokenizer.from_pretrained("kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1")
 
 # Function to generate responses from both models