1. Introduction to LLM evaluations in 10 key ideas скачать в хорошем качестве

1. Introduction to LLM evaluations in 10 key ideas

00:03 Intro 00:24 LLM evals ≠ benchmarking 01:03 LLM evals are a tool, not a task 02:26 LLM evals ≠ software testing 03:36 Manual + automated evals 04:31 Use reference-based and -free evals 05:40 Think in datasets, not unit tests 06:30 LLM-as-a-judge is a key method 07:30 Use custom criteria, not generic metrics 09:12 Start with analytics 10:05 Evaluation is a moat LINKS Intro playlist mentioned in the video:    • LLM evaluation course   LLM evaluation guides: LLM Benchmarks https://www.evidentlyai.com/llm-guide... Intro to LLM evals https://www.evidentlyai.com/llm-guide... Test datasets https://www.evidentlyai.com/llm-guide... COURSE PLAYLIST Full playlist:    • Course: LLM evaluation for builders   Instructor: Elena Samuylova, CEO Evidently AI. EVIDENTLY Sign up for Evidently Cloud https://www.evidentlyai.com/register Support Evidently on GitHub https://github.com/evidentlyai/evidently