Update README.md

README.md

@@ -15,7 +15,7 @@ USER: prompt
 ASSISTANT:
 ```
 
-*The quality of the 3-bit model is higher than the 2-bit model, but the interface is slower. The 3-bit model (type q3_K_S) barely fits into 16 gigabytes of RAM, but it works.*
+*The quality of the 3-bit model is higher than the 2-bit model, but the interface is slower. The 3-bit model (type q3_K_S) barely fits into 16 GB of RAM, but it works.*
 ``` 
 llama_model_load_internal: mem required  = 15716.00 MB (+ 3124.00 MB per state) 
 ```