Update README.md

README.md

@@ -53,7 +53,7 @@ if __name__ == "__main__":
 ```
 
 𝐑𝐞𝐬𝐮𝐥𝐭𝐬:
-Firslty, a higher learning rate was required for the model to train the model on a dataset, training methods such as SFT and ORPO 
+Firstly, a higher learning rate was required for the model to train the model on a dataset, training methods such as SFT and ORPO 
 with PEFT failed to restore convergence.Secondly, training only using PEFT helped to restore partial model convergence.
 
 Lastly, future experiments to restore model convergence will require a systematic training and eval strategy.