Все публикации

0:24:02

Deliberative Alignment: Reasoning Enables Safer Language Models

0:24:04

Alignment Faking in Large Language Models

0:16:00

RE-Bench: measuring AI agents at AI R&D vs human experts

0:05:17

NeurIPS 2024 Poster - On scalable oversight

0:01:20

NeurIPS 2024 Poster - No 'Zero-Shot' Without Exponential Data

0:03:31

Still a long way to go for Computer Vision? The GRAB Benchmark

0:11:29

Gemini 1.5 Pro has a massive context window

0:16:02

Challenges with unsupervised LLM knowledge discovery

0:19:35

Anthropic - AI sleeper agents?

0:16:01

Mamba - a replacement for Transformers?

0:15:24

How does Gemini compare to GPT-4?

0:28:48

Self-supervised vision

0:30:49

Vision Transformer Basics

0:25:31

Is Chain of Thought faithful?

0:12:49

How strong is Claude 2?

0:15:49

What does AI believe is true?

0:17:10

Can we verify training data?

0:08:54

What is Superalignment?

0:08:53

What is SDXL 0.9?

0:25:52

Eliciting Latent Knowledge

0:14:47

What is KOSMOS-2?

0:37:27

Possible catastrophic AI risks?

0:13:40

Textbooks Are All You Need

0:00:54

What is Gaussian Elimination?