Token Limit #220
-
Use case Description : |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Hi @Gowtham-Sivam ! Only the top 3 "relevant chunks" are included in the prompt, and their maximum length is limited (default 400 characters - https://github.com/NVIDIA/NeMo-Guardrails/blob/develop/nemoguardrails/kb/utils.py#L22). So, the short answer is no, the token limit should not be reached. As per the above, that type of query would not work out of the box. Most likely, only 3 rows would be included in the prompt and you would get an incorrect result. |
Beta Was this translation helpful? Give feedback.
Hi @Gowtham-Sivam !
Only the top 3 "relevant chunks" are included in the prompt, and their maximum length is limited (default 400 characters - https://github.com/NVIDIA/NeMo-Guardrails/blob/develop/nemoguardrails/kb/utils.py#L22). So, the short answer is no, the token limit should not be reached.
As per the above, that type of query would not work out of the box. Most likely, only 3 rows would be included in the prompt and you would get an incorrect result.