Skip to contents

Changelog

Source: NEWS.md

tidyprompt (development version)

  • Added new prompt wraps: answer_as_category() & answer_as_multi_category()
  • Added llm_break_soft(), which can be used to interrupt prompt evaluation without causing an error
  • Safer handling of chat history objects (‘NA’ rows could cause errors in specific cases)
  • Fixed default Google Gemini LLM provider
  • More flexible lenient extraction for final answer extraction in chain-of-thought
  • Use message() instead of cat() for printing LLM responses
  • Added ‘num_ctx’ parameter for Ollama LLM provider to more easily control the context window size
  • Added set_option() & `set_options() to Ollama LLM provider to more easily set options
  • Added a more detailed error message in case a LLM provider cannot be reached

tidyprompt 0.0.1

CRAN release: 2025-01-08

  • Initial CRAN release

tidyprompt 0.0.0.9000

  • Initial development version available on GitHub