Update README.md

README.md

@@ -9,7 +9,8 @@ language:
 We made a LLM model with yanolja/KoSOLAR-10.7B-v0.3 
 
 ## Model Description
-We use QLoRA and NEFTune on LLM Fine-tuning
+We use QLoRA(64, 16) and NEFTune on LLM Fine-tuning
+lr = 2e-4
 
 ## Train Detail
 Our Korean Wiki QA data used, and 3 epoch train