Hello Haoyu, thanks for posting!
The score describes the % score of the match with the training datasets. Though the number must be below 0 i.e., 0.9756 or 0.96322 etc., I see some incorrect match scores in your screenshot. We did notice it as well and raised a bug internally to solve it. Let me check if it was resolved.
------------------------------
Chetan Hireholi
------------------------------
Original Message:
Sent: Wed March 06, 2024 03:41 PM
From: Haoyu Wang
Subject: what is the value of the score included in the Ansible Lightspped Training Matches?
Hello,
I'm preparing a client demo for WatsonX Code Assistant for Ansible Lightspeed. After generating the code, it displays information about the data source and the path associated with the code recommendation. However, I'm curious about the significance of the score provided. Additionally, I noticed that the scores vary (1.96 vs. 1.92) even when the remaining context appears consistent. Please advise. Thanks!
------------------------------
Haoyu Wang
------------------------------