Top k (LLM Attribute): Difference between revisions

Latest revision as of 12:52, 1 July 2026

Attribute	Unit	Description	Default value
TOP_K	number	Top-K sampling limits the LLM model to choosing from the K most likely tokens at each step.	64

@@ Line 1: / Line 1: @@
 {{LLM attributes
 | <onlyinclude>
-{{LLM attributes|type=row|attribute=TOP_K|unit=number|defaultvalue=64|description=Top-k sampling is an adaptive version of [[Top p (LLM Attribute)]]. Instead of fixed number of tokens, it cuts off based on cumulative probability of the [[Large Language Model (LLM)|LLM]].}} </onlyinclude>
+{{LLM attributes|type=row|attribute=TOP_K|unit=number|defaultvalue=64|description=Top-K sampling limits the [[Large Language Model (LLM)|LLM]] model to choosing from the K most likely tokens at each step.}} </onlyinclude>
 }}
 {{article end