Explore cutting-edge research presented at the NeurIPS 2023 conference in this 44-minute video covering seven papers from the Wednesday morning poster session. Dive into topics including bifurcations in RNN training, limitless image editing with text-to-image models, lexinvariant language models, transformers learning preconditioned gradient descent, user preferences in text-to-image generation, worldwide geo-localization using CLIP-inspired techniques, and hardware resilience of text-guided image classifiers. Gain insights into the latest advancements in machine learning, computer vision, and natural language processing as each paper is discussed in detail with timestamps provided for easy navigation.
NeurIPS 2023 Research Highlights - Poster Session 2
Overview
Syllabus
- Bifurcations and loss jumps in RNN training https://arxiv.org/abs/2310.17561
- LEDITS++: Limitless Image Editing using Text-to-Image Models https://arxiv.org/abs/2311.16711
- Lexinvariant Language Models https://arxiv.org/abs/2305.16349
- Transformers learn to implement preconditioned gradient descent for in-context learning https://arxiv.org/abs/2306.00297
- Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation https://arxiv.org/abs/2305.01569
- GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization https://arxiv.org/abs/2309.16020
- Hardware Resilience Properties of Text-Guided Image Classifiers https://arxiv.org/abs/2311.14062
Taught by
Yannic Kilcher
