A Closer Look at Learned Optimization: Stability, Robustness, and Inductive Biases

google/learned_optimization 22 Sep 2022

We apply the resulting learned optimizer to a variety of neural network training tasks, where it outperforms the current state of the art learned optimizer -- at matched optimizer computational overhead -- with regard to optimization performance and meta-training speed, and is capable of generalization to tasks far different from those it was meta-trained on.

Inductive Bias

461
0.49 stars / hour

LiT: Zero-Shot Transfer with Locked-image text Tuning

mlfoundations/open_clip CVPR 2022

This paper presents contrastive-tuning, a simple method employing contrastive training to align image and text models while still taking advantage of their pre-training.

Image Classification Retrieval +2

2,430
0.48 stars / hour

Shape, Pose, and Appearance from a Single Image via Bootstrapped Radiance Field Inversion

google-research/nerf-from-image 21 Nov 2022

Neural Radiance Fields (NeRF) coupled with GANs represent a promising direction in the area of 3D reconstruction from a single view, owing to their ability to efficiently model arbitrary topologies.

3D Reconstruction Pose Estimation

72
0.46 stars / hour
185
0.45 stars / hour

DeepPrivacy2: Towards Realistic Full-Body Anonymization

hukkelas/deep_privacy2 17 Nov 2022

Generative Adversarial Networks (GANs) are widely adapted for anonymization of human figures.

Face Anonymization

70
0.43 stars / hour

High-fidelity 3D GAN Inversion by Pseudo-multi-view Optimization

jiaxinxie97/hfgi3d 28 Nov 2022

We present a high-fidelity 3D generative adversarial network (GAN) inversion framework that can synthesize photo-realistic novel views while preserving specific details of the input image.

Novel View Synthesis

53
0.42 stars / hour

Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation

peract/peract 12 Sep 2022

With this formulation, we train a single multi-task Transformer for 18 RLBench tasks (with 249 variations) and 7 real-world tasks (with 18 variations) from just a few demonstrations per task.

97
0.39 stars / hour

Generate rather than Retrieve: Large Language Models are Strong Context Generators

wyu97/GenRead 21 Sep 2022

A common approach for knowledge-intensive tasks is to employ a retrieve-then-read pipeline that first retrieves a handful of relevant contextual documents from an external corpus such as Wikipedia and then predicts an answer conditioned on the retrieved documents.

Fact Checking Language Modelling +3

58
0.39 stars / hour

Towards Robust Blind Face Restoration with Codebook Lookup Transformer

sczhou/codeformer 22 Jun 2022

In this paper, we demonstrate that a learned discrete codebook prior in a small proxy space largely reduces the uncertainty and ambiguity of restoration mapping by casting blind face restoration as a code prediction task, while providing rich visual atoms for generating high-quality faces.

Blind Face Restoration

1,530
0.35 stars / hour

MineDojo: Building Open-Ended Embodied Agents with Internet-Scale Knowledge

MineDojo/MineDojo 17 Jun 2022

Autonomous agents have made great strides in specialist domains like Atari games and Go.

Atari Games

830
0.35 stars / hour