paper-with-me

홈 › Papers

Instruction Diversity Drives Generalization To Unseen Tasks

2024-02-16 · Dylan Zhang, Justin Wang, Francois Charton

Instruction tuning -- fine-tuning a large language model (LLM) on pairs of instructions and desired outcomes -- is an approach that enables pre-trained language models to perform real-world tasks and follow human instructions. Its practical success depends on the model learning a broader set of instructions than those it was trained on. Yet the factors that determine model generalization to such \emph{unseen tasks} are not well understood. %To understand the driving factors of generalization, In this paper, we experiment with string rewrites, a symbolic task that serves as a building block for Turing complete Markov algorithms while allowing experimental control of "inputs" and "instructions". We investigate the trade-off between the number of instructions the model is trained on and the number of training samples provided for each instruction and observe that the diversity of the instruction set determines generalization. Generalization emerges once a diverse enough set of tasks is provided, even though very few examples are provided for each task. Instruction diversity also ensures robustness with respect to non-uniform distributions of instructions in the training set.

📄 PDF Abstract BibTeX arXiv:2402.10891

Code (0)

등록된 구현이 없습니다.

Tasks

DiversityLanguage ModelingLanguage ModellingLarge Language Model

Methods 이 논문이 사용한 방법론

SET Dynamic Sparse Training method where weight mask is updated randomly periodically

Similar Papers 제목 키워드 기반

Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning

2025-05-26 · JaeHun Jung, Seungju Han, Ximing Lu, Skyler Hallinan 외

Effective generalization in language models depends critically on the diversity of their training data. Yet existing diversity metrics often fall short of this goal, relying on surface-level heuristics that are decoupled…

DiversityMathNatural Language Inference

Zero-Shot Task Generalization with Multi-Task Deep Reinforcement Learning

2017-06-15 · ICML 2017 8 · Junhyuk Oh, Satinder Singh, Honglak Lee, Pushmeet Kohli

As a step towards developing zero-shot task generalization capabilities in reinforcement learning (RL), we introduce a new RL problem where the agent should learn to execute sequences of instructions after learning usefu…

Deep Reinforcement Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Unseen from Seen: Rewriting Observation-Instruction Using Foundation Models for Augmenting Vision-Language Navigation

2025-03-23 · Ziming Wei, Bingqian Lin, Yunshuang Nie, Jiaqi Chen 외

Data scarcity is a long-standing challenge in the Vision-Language Navigation (VLN) field, which extremely hinders the generalization of agents to unseen environments. Previous works primarily rely on additional simulator…

DiversityImage GenerationText to Image GenerationText-to-Image Generation+1

Chain-of-Instructions: Compositional Instruction Tuning on Large Language Models

2024-02-18 · Shirley Anugrah Hayati, Taehee Jung, Tristan Bodding-Long, Sudipta Kar 외

Fine-tuning large language models (LLMs) with a collection of large and diverse instructions has improved the model's generalization to different tasks, even for unseen tasks. However, most existing instruction datasets …

OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalization

2022-12-22 · Srinivasan Iyer, Xi Victoria Lin, Ramakanth Pasunuru, Todor Mihaylov 외

Recent work has shown that fine-tuning large pre-trained language models on a collection of tasks described via instructions, a.k.a. instruction-tuning, improves their zero and few-shot generalization to unseen tasks. Ho…

Language ModelingLanguage ModellingMeta-LearningNatural Language Inference+1