Automated Evaluation of LLMs

Infinite Curiosity Pod with Prateek Joshi

Prateek Joshi에서 제공하는 콘텐츠입니다. 에피소드, 그래픽, 팟캐스트 설명을 포함한 모든 팟캐스트 콘텐츠는 Prateek Joshi 또는 해당 팟캐스트 플랫폼 파트너가 직접 업로드하고 제공합니다. 누군가가 귀하의 허락 없이 귀하의 저작물을 사용하고 있다고 생각되는 경우 여기에 설명된 절차를 따르실 수 있습니다 https://ko.player.fm/legal.

1+ y ago 36:03

MP3•에피소드 홈

Anand Kannappan is the cofounder and CEO of Patronus AI, an automated AI evaluation and security company. They have raised funding from Lightspeed Venture Partners, Replit CEO Amjad Masad, Gokul Rajaram, and Fortune 500 executives. He was previously at Meta and Vertis. He was also the cofounder of Kyber Technologies, which was a service to systematically predict market events using AI and remote sensing data. It evolved into a futures quant hedge fund managing $15M for partners.
Anand's favorite book: Harry Potter series (Author: JK Rowling)
(00:00) Introduction and Common Failure Modes of Large Language Models
(03:02) Challenges of Automated Evaluation in AI Models
(06:08) The Importance of Fine-Tuning and Retrieval Augmented Generation
(09:02) Addressing Copyright Detection in Language Models
(11:51) The Liability of Companies Using AI Models
(15:02) Advancements in Multimodal Models and State Space Models
(20:48) The Role of Fine-Tuning in the Evolution of Language Models
(23:51) The Significance of Adversarial Testing in AI
(25:56) The Role of Retrieval Augmented Generation in AI
(28:05) The Need for Continuous Function Optimization in Prompting
(29:02) Rapid Fire Round
--------
Where to find Prateek Joshi:
Newsletter: https://prateekjoshi.substack.com
Website: https://prateekj.com
LinkedIn: https://www.linkedin.com/in/prateek-joshi-91047b19
Twitter: https://twitter.com/prateekvjoshi