Update README.md

README.md

@@ -123,7 +123,7 @@ The training takes about 4 days on four GPUs to complete.
 
 We use the following hyperparameters: 
   - Q-LoRA rank (r): 64 
-  - Q-LoRA scaling factor ($alpha$): 4 
+  - Q-LoRA scaling factor (alpha): 4 
   - Q-LoRA dropout: 0 
   - Optimizer: Paged AdamW 
   - Learning rate scheduler: linear