●Change8

b8144

📦 llama-cppView on GitHub →
✨ 1 features⚡ 1 deprecations🔧 2 symbols

Summary

The server API has been updated to support the new "max_completion_tokens" request property, deprecating the older "max_tokens" parameter.

Migration Steps

  1. Replace usage of the deprecated request property "max_tokens" with "max_completion_tokens".

✨ New Features

  • Server now supports the "max_completion_tokens" request property to set the upper bound for reasoning and output tokens.

Affected Symbols

max_completion_tokensmax_tokens

⚡ Deprecations

  • The request property "max_tokens" is deprecated in favor of "max_completion_tokens".