Add --enable-auto-tool-choice and suggest a higher value for max_tokens with reasoning on, following @venkats-nvidia 's suggestion.

suhara changed pull request status to merged

Sign up or log in to comment