Uploaded December 2022 | Updated September 2026, 7 hours ago
Let's talk about all the NeurIPS 2022 outstanding paper awards. There were a good mix of papers, but especially a lot of work in diffusion models and optimization with both empirical work and theory. You can find links to all of these papers at: blog.neurips.cc/2022/11/21/announcing-the-neurips-2022-awards
ClearML: bit.ly/3GtCsj5
Outline
0:00 - Intro
1:03 - ClearML
2:08 - OOD Detection
4:21 - Imagen
5:19 - Diffusion Model Design
7:07 - ProcTHOR
8:58 - Instilling Human Biases
10:46 - Neural Document Retrieval
12:35 - High Dimensional SGD
13:14 - Gradient Descent: The Ultimate Optimizer
15:00 - Riemannian Diffusion Models
16:09 - Gradient Estimation for Discrete Values
17:27 - Chinchilla
19:03 - Beyond Neural Scaling Laws via Pruning
20:23 - On-Demand Multi-Distribution Learning
22:18 - LAION-5B
23:15 - MineDOJO
Social Media
YouTube - youtube.com/c/EdanMeyer
Twitter - twitter.com/ejmejm1
Sources:
Outstanding Papers - blog.neurips.cc/2022/11/21/announcing-the-neurips-2022-awards
Let's talk about all the NeurIPS 2022 outstanding paper awards. There were a good mix of papers, but especially a lot of work in diffusion models and optimization with both empirical work and theory. You can find links to all of these papers at: blog.neurips.cc/2022/11/21/announcing-the-neurips-2022-awards
ClearML: bit.ly/3GtCsj5
Outline
0:00 - Intro
1:03 - ClearML
2:08 - OOD Detection
4:21 - Imagen
5:19 - Diffusion Model Design
7:07 - ProcTHOR
8:58 - Instilling Human Biases
10:46 - Neural Document Retrieval
12:35 - High Dimensional SGD
13:14 - Gradient Descent: The Ultimate Optimizer
15:00 - Riemannian Diffusion Models
16:09 - Gradient Estimation for Discrete Values
17:27 - Chinchilla
19:03 - Beyond Neural Scaling Laws via Pruning
20:23 - On-Demand Multi-Distribution Learning
22:18 - LAION-5B
23:15 - MineDOJO
Social Media
YouTube - youtube.com/c/EdanMeyer
Twitter - twitter.com/ejmejm1
Sources:
Outstanding Papers - blog.neurips.cc/2022/11/21/announcing-the-neurips-2022-awards










