Update README.md

README.md

@@ -20,7 +20,7 @@ Based on our latest technological advancements, we have trained a `GLM-4-0414` s
 | ------------ | ----------------- | -------------------------------------------- |
 | temperature  | **0.6**           | Balances creativity and stability            |
 | top_p        | **0.95**          | Cumulative probability threshold for sampling|
-| top_k        | **20–40**         | Filters out rare tokens while maintaining diversity |
+| top_k        | **40**         | Filters out rare tokens while maintaining diversity |
 | max_new_tokens        | **30000**         | Leaves enough tokens for thinking |
 
 ### II. Enforced Thinking