Update README.md

README.md

@@ -24,7 +24,17 @@ questions = [
 Ask the same questions to o3-mini-high and DeepSeek R1 ( Deepseek Think mode ).
 
 # Eval @ AIME2024 
-70 %
+
+This model has been evaluated on AIME2024 with below configurations
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,  # 4-bit quantization
+    bnb_4bit_compute_dtype=torch.float16,  # Use FP16 for computation
+    bnb_4bit_use_double_quant=True,  # Enable double quantization
+    bnb_4bit_quant_type="nf4"  # Most memory-efficient format
+)
+
+and it scored 70 % correct answers in 1st attempt.
 
 ## How to Use