Context Learning

In-context learning (ICL) is a paradigm shift in machine learning, focusing on enabling models to adapt to new tasks using only a few examples provided within the input, without requiring parameter updates. Current research emphasizes understanding ICL's mechanisms, particularly within transformer-based large language models, and improving its effectiveness through techniques like enhanced example selection, chain-of-thought prompting, and addressing issues such as spurious correlations and copy bias. This research is significant because ICL offers a more efficient and adaptable approach to many machine learning problems, impacting fields ranging from natural language processing and computer vision to scientific computing and beyond.

Papers

December 5, 2023

December 4, 2023

December 3, 2023

December 1, 2023

Manipulating the Label Space for In-Context Classification
Haokun Chen, Xu Yang, Yuhang Huang, Zihan Wu, Jing Wang, Xin Geng
Language Model Vision Language Model Context Learning Language Generation Label Space

November 30, 2023

Positional Information Matters for Invariant In-Context Learning: A Case Study of Simple Function Classes
Yongqiang Chen, Binghui Xie, Kaiwen Zhou, Bo Han, Yatao Bian, James Cheng
Case Study Context Learning Context Demonstration Function Class

November 29, 2023

Understanding and Improving In-Context Learning on Vision-language Models
Shuo Chen, Zhen Han, Bailan He, Mark Buckley, Philip Torr, Volker Tresp, Jindong Gu
Vision Language Model Context Learning Human Understanding Context Example Multimodal Demonstration

November 28, 2023

Eliciting In-Context Learning in Vision-Language Models for Videos Through Curated Data Distributional Properties
Keunwoo Peter Yu, Zheyuan Zhang, Fengyuan Hu, Shane Storks, Joyce Chai
Vision Language Model Context Learning Large Vision Language Model Egocentric Video Context Example Large Scale Naturalistic Egocentric Video Datasets Efficient in Context Learning

November 27, 2023

November 22, 2023

November 21, 2023

In-Context Learning Functions with Varying Number of Minima
David Oniani, Yanshan Wang
Context Learning Numerical Data Two Layer Neural Network Function Approximation

November 20, 2023

November 17, 2023

Exploring the Relationship between In-Context Learning and Instruction Tuning
Hanyu Duan, Yixuan Tang, Yi Yang, Ahmed Abbasi, Kar Yan Tam
Context Learning Instruction Tuning Human Relationship Inference Time

November 16, 2023

Hijacking Large Language Models via Adversarial In-Context Learning
Yao Qiang, Xiangyu Zhou, Dongxiao Zhu
Large Language Model Adversarial Attack Context Learning Token Embeddings Adversarial Suffix

Context Learning

Papers

Machine Vision Therapy: Multimodal Large Language Models Can Enhance Visual Robustness via Denoising In-Context Learning

Towards More Unified In-context Visual Understanding

IMProv: Inpainting-based Multimodal Prompting for Computer Vision Tasks

How to Configure Good In-Context Sequence for Visual Question Answering

The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning

The mechanistic basis of data dependence and abrupt learning in an in-context classification task

Improving In-Context Learning in Diffusion Models with Visual Context-Modulated Prompts

Manipulating the Label Space for In-Context Classification

Positional Information Matters for Invariant In-Context Learning: A Case Study of Simple Function Classes

Understanding and Improving In-Context Learning on Vision-language Models

Eliciting In-Context Learning in Vision-Language Models for Videos Through Curated Data Distributional Properties

WorldSense: A Synthetic Benchmark for Grounded Reasoning in Large Language Models

FreeAL: Towards Human-Free Active Learning in the Era of Large Language Models

AlignedCoT: Prompting Large Language Models via Native-Speaking Demonstrations

Multi-modal In-Context Learning Makes an Ego-evolving Scene Text Recognizer

In-Context Learning Functions with Varying Number of Minima

On the Potential and Limitations of Few-Shot In-Context Learning to Generate Metamorphic Specifications for Tax Preparation Software

Adapt in Contexts: Retrieval-Augmented Domain Adaptation via In-Context Learning

Exploring the Relationship between In-Context Learning and Instruction Tuning

Hijacking Large Language Models via Adversarial In-Context Learning