Update README.md

README.md

@@ -19,7 +19,7 @@ As such, this model will be updated soon when we have trained it on the full dat
 
 ## Sampling Parameters
 
-We've had success with top_p=0.95, top_k=30, repetition_penalty=1.15 and temperature=0.6 and it's also recommended to append \<think\> to force reasoning like this:
+We've had success with top_p=0.95, top_k=30, repetition_penalty=1.15 and temperature=0.4 and it's also recommended to append \<think\> to force reasoning like this:
 
 ```python
 input_text = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)