))
Constitutional AI Harmlessness from AI Feedback
סדרה בארכיון ("עדכון לא פעיל" status)
When?
This feed was archived on February 21, 2025 21:08 (
Why? עדכון לא פעיל status. השרתים שלנו לא הצליחו לאחזר פודקאסט חוקי לזמן ממושך.
What now? You might be able to find a more up-to-date version using the search function. This series will no longer be checked for updates. If you believe this to be in error, please check if the publisher's feed link below is valid and contact support to request the feed be restored or if you have any other concerns about this.
Manage episode 429711879 series 3498845
תוכן מסופק על ידי BlueDot Impact. כל תוכן הפודקאסטים כולל פרקים, גרפיקה ותיאורי פודקאסטים מועלים ומסופקים ישירות על ידי BlueDot Impact או שותף פלטפורמת הפודקאסט שלהם. אם אתה מאמין שמישהו משתמש ביצירה שלך המוגנת בזכויות יוצרים ללא רשותך, אתה יכול לעקוב אחר התהליך המתואר כאן https://he.player.fm/legal.
This paper explains Anthropic’s constitutional AI approach, which is largely an extension on RLHF but with AIs replacing human demonstrators and human evaluators.
Everything in this paper is relevant to this week's learning objectives, and we recommend you read it in its entirety. It summarises limitations with conventional RLHF, explains the constitutional AI approach, shows how it performs, and where future research might be directed.
If you are in a rush, focus on sections 1.2, 3.1, 3.4, 4.1, 6.1, 6.2.
A podcast by BlueDot Impact.
Learn more on the AI Safety Fundamentals website.
פרקים
1. Constitutional AI Harmlessness from AI Feedback (00:00:00)
2. Abstract (00:00:20)
3. 1 Introduction (00:03:27)
4. 1.1 Motivations (00:06:10)
5. 1.2 The Constitutional AI Approach (00:11:20)
6. 1.3 Contributions (00:14:29)
7. 2 Evaluating the Potential for AI Supervision of HHH (00:22:01)
8. 3 Constitutional AI: Critiques, Revisions, and Supervised Learning (00:24:20)
9. 3.1 Method (00:24:56)
10. 3.2 Datasets and Training (00:29:12)
11. 3.3 Main Results (00:30:23)
12. 3.4 Scaling Trends (00:34:10)
13. 4 Constitutional AI: Reinforcement Learning from AI Feedback (00:37:08)
14. 4.1 Method (00:37:40)
15. 4.2 Datasets and Training (00:42:21)
16. 4.3 Main Results (00:45:31)
17. 4.4 Harmlessness vs. Evasiveness (00:49:35)
18. 4.5 Absolute Harmfulness Score (00:52:12)
19. 5 Related Work (00:54:40)
20. 6 Discussion (00:56:29)
21. 6.1 Future Directions (00:58:13)
22. 6.2 Broader Impacts (01:00:16)
85 פרקים
שתפו
