Skip to content

Instantly share code, notes, and snippets.

@joshreini1

joshreini1/3H_evaluations.py

Last active November 2, 2023 13:19
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save joshreini1/10510e7703bc38c183041cf4e1d4402e to your computer and use it in GitHub Desktop.
Save joshreini1/10510e7703bc38c183041cf4e1d4402e to your computer and use it in GitHub Desktop.
Download ZIP
Raw
3H_evaluations.py
honest_feedbacks = [qa_relevance, qs_relevance, f_embed_dist, f_groundedness]
harmless_feedbacks = [f_controversiality, f_criminality, f_harmfulness, f_insensitivity,
f_maliciousness, f_misogyny, f_stereotypes, f_hate,
f_hatethreatening, f_violent, f_violentgraphic, f_selfharm]
helpful_feedbacks = [f_langmatch, f_conciseness]
feedback_suite = honest_feedbacks + harmless_feedbacks + helpful_feedbacks
tru_recorder = TruLlama(
query_engine,
app_id='Query Engine',
feedbacks=feedback_suite
)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment