2: data2vec

Update: 2022-03-07

Description

Todays paper: data2vec (https://arxiv.org/abs/2202.03555)

Summary of the paper
A multimodal SSL algorithm that predicts latent representation of different types of input.

Highlights of discussion

What are the motivations of SSL and multimodal
How does the student teacher learning work?
What are similarities and differences between ViT, BYOL, and Reinforcement Learning algorithms.

Comments

In Channel

LoRA

2023-09-0201:02:56

15: InstructGPT

2023-03-2857:27

14: Whisper

2023-03-1749:14

13: AlphaTensor

2023-03-1149:05

12: SIRENs

2022-10-2554:17

11: CVPR Workshop on Autonomous Driving Keynote by Ashok Elluswamy, a Tesla engineer

2022-09-3048:51

10: Outracing champion Gran Turismo drivers with deep reinforcement learning

2022-08-2354:50

8: GATO (A Generalist Agent)

2022-07-2944:51

7: Deep Unsupervised Learning Using Nonequilibrium Thermodynamics (Diffusion Models)

2022-06-1430:55

6: Deep Reinforcement Learning at the Edge of the Statistical Precipice

2022-06-0601:01:08

5: QMIX

2022-04-2642:06

4: Can Neural Nets Learn the Same Model Twice?

2022-04-0655:23

3: VICReg

2022-03-2144:46

2: data2vec

2022-03-0753:23

1: Reward is Enough

2022-02-2154:36

Mixture of Experts

2024-10-0854:46

9: Heads-Up Limit Hold'em Poker Is Solved

2022-07-2947:55

00:00

2: data2vec

Vahe Hagopian, Taka Hasegawa, Farrukh Rahman

#box-pro-ellipsis-175749094861235{-webkit-line-clamp:2;}2: data2vec

2: data2vec

Vahe Hagopian, Taka Hasegawa, Farrukh Rahman

2: data2vec