- 3
- 1 377
Julia Turc
Приєднався 9 вер 2022
How does OpenAI Operator work under the hood? | Tech deep dive
In this video, we discuss the science behind OpenAI's Operator and other browser / computer agents, including LMMs (Large Multimodal Models), Imitation Learning and Reinforcement Learning.
Relevant papers for web agents:
arxiv.org/abs/2306.06070 (Mind2Web)
arxiv.org/abs/2310.11441 (Set-of-Mark Prompting)
arxiv.org/abs/2207.01206 (WebShop)
arxiv.org/abs/2401.01614 (SeeAct)
arxiv.org/abs/2401.13919 (WebVoyager)
arxiv.org/abs/2303.08774 (GPT-4 Technical Report)
arxiv.org/abs/2112.09332 (WebGPT)
Relevant blogposts:
openai.com/index/computer-using-agent/
www.anthropic.com/news/developing-computer-use
00:00 Intro
00:57 Demo
01:57 Other agents
03:16 Large Multimodal Models
05:38 Imitation and Reinforcement Learning
09:01 Outro
Relevant papers for web agents:
arxiv.org/abs/2306.06070 (Mind2Web)
arxiv.org/abs/2310.11441 (Set-of-Mark Prompting)
arxiv.org/abs/2207.01206 (WebShop)
arxiv.org/abs/2401.01614 (SeeAct)
arxiv.org/abs/2401.13919 (WebVoyager)
arxiv.org/abs/2303.08774 (GPT-4 Technical Report)
arxiv.org/abs/2112.09332 (WebGPT)
Relevant blogposts:
openai.com/index/computer-using-agent/
www.anthropic.com/news/developing-computer-use
00:00 Intro
00:57 Demo
01:57 Other agents
03:16 Large Multimodal Models
05:38 Imitation and Reinforcement Learning
09:01 Outro
Переглядів: 694
Відео
How does DeepSeek actually work? | Full technical review
Переглядів 775День тому
In this video, we dive into the technical innovations behind DeepSeek-R1: scaling with compute (Reasoning-Oriented Reinforcement Learning, Chain-of-Thought, GRPO, Distillation). While knowledge about AI is helpful, general software engineers should still get great value out of it. 00:00 Intro 00:57 Scaling using compute instead of data 02:18 Overview of LLM training 03:20 Training DeepSeek 04:0...
Discover How LLMs Work by Dissecting Llama
Переглядів 645День тому
In this video, I explain AI models from first principles, at a level of abstraction that is easy to understand if you're a generalist software engineer or just a little tech-savvy. In only 10 minutes, we'll cover AI models & LLMs and dive deep into the Llama model. We'll even change its internals to generate special Haiku's, which the original Llama 1B model can't do. You can follow along using...