|
en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
|
|
en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
|