components generation_safety_quality_signal_monitor - Azure/azureml-assets GitHub Wiki

Generation Safety & Quality - Signal Monitor

generation_safety_quality_signal_monitor

Overview

Computes the content generation safety metrics over LLM outputs.

Version: 0.5.21

View in Studio: https://ml.azure.com/registries/azureml/components/generation_safety_quality_signal_monitor/version/0.5.21

Inputs

Name Description Type Default Optional Enum
monitor_name string
signal_name string
production_data mltable
metric_names a comma-separated list of metric names to compute string False
model_deployment_name model name, e.g., name of the deployment for Azure OpenAI string False
sample_rate sample rate for the input dataset, should be greater than 0 and at most 1 number True
prompt_column_name string prompt True
completion_column_name string completion True
ground_truth_column_name string ground_truth True
context_column_name string context True
groundedness_rating_threshold integer 4 True
similarity_rating_threshold integer 4 True
relevance_rating_threshold integer 4 True
fluency_rating_threshold integer 4 True
coherence_rating_threshold integer 4 True
thresholds a comma-separated dictionary of threshold:value for each metric string True
groundedness_passrate_threshold number 0.7 True
similarity_passrate_threshold number 0.7 True
relevance_passrate_threshold number 0.7 True
fluency_passrate_threshold number 0.7 True
coherence_passrate_threshold number 0.7 True
workspace_connection_arm_id string False
monitor_current_time string
notification_emails string True
instance_type string standard_e4s_v3 True

Outputs

Name Description Type
signal_output uri_folder
signal_scored_data mltable
⚠️ **GitHub.com Fallback** ⚠️