Detect LLM hallucinations using uncertainty quantification techniques with UQLM
\n \t
\n
<\/BR>
\n
\n \t
\n
<\/BR><\/p>\n
UQLM (uncertainty quantification for language models) is an open source Python package for generation time, zero-resource hallucination detection. It leverages state-of-the-art uncertainty quantification (UQ) techniques from the academic literature to compute response-level confidence scores based on response consistency (in multiple responses to the same prompt), token probabilities, LLM-as-a-Judge, or ensembles of these.<\/p>\n
UQLM aims to democratize state-of-the-art UQ techniques. By integrating generation and UQ-scoring processes with a user-friendly API, UQLM makes these methods accessible to non-specialized practitioners with minimal engineering effort.<\/p>\n
Check it out, share feedback, and contribute if you are interested!<\/p>\n
Link: https:\/\/github.com\/cvs-health\/uqlm<\/a><\/p>\n<\/p><\/div>\n submitted by \/u\/Opposite_Answer_287 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span>\n<\/div>\n