# The Right Time Matters: Data Arrangement Affects Zero-Shot Generalization in Instruction Tuning

arXiv: https://arxiv.org/abs/2406.11721 · HTML: https://mlsanity.com/paper/2406.11721/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 41 references that are in the map.
- First posted on arXiv 17 Jun 2024.

## Builds on (most cited first, 30 of 30)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [Multitask Prompted Training Enables Zero-Shot Task Generalization](https://mlsanity.com/paper/2110.08207/) (2021)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks](https://mlsanity.com/paper/2204.07705/) (2022)
- [Enhancing Chat Language Models by Scaling High-quality Instructional Conversations](https://mlsanity.com/paper/2305.14233/) (2023)
- [The Flan Collection: Designing Data and Methods for Effective Instruction Tuning](https://mlsanity.com/paper/2301.13688/) (2023)
- [Taskonomy: Disentangling Task Transfer Learning](https://mlsanity.com/paper/1804.08328/) (2018)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [Are Emergent Abilities of Large Language Models a Mirage?](https://mlsanity.com/paper/2304.15004/) (2023)
- [Scaling Laws for Autoregressive Generative Modeling](https://mlsanity.com/paper/2010.14701/) (2020)
- [The False Promise of Imitating Proprietary LLMs](https://mlsanity.com/paper/2305.15717/) (2023)
- [Unified Scaling Laws for Routed Language Models](https://mlsanity.com/paper/2202.01169/) (2022)
- [OpenChat: Advancing Open-source Language Models with Mixed-Quality Data](https://mlsanity.com/paper/2309.11235/) (2023)
- [OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalization](https://mlsanity.com/paper/2212.12017/) (2022)
- [Advancing LLM Reasoning Generalists with Preference Trees](https://mlsanity.com/paper/2404.02078/) (2024)
- [Exploring and Predicting Transferability across NLP Tasks](https://mlsanity.com/paper/2005.00770/) (2020)
- [Understanding Emergent Abilities of Language Models from the Loss Perspective](https://mlsanity.com/paper/2403.15796/) (2024)
- [Long Is More for Alignment: A Simple but Tough-to-Beat Baseline for Instruction Fine-Tuning](https://mlsanity.com/paper/2402.04833/) (2024)
- [ROUGE 2.0: Updated and Improved Measures for Evaluation of Summarization Tasks](https://mlsanity.com/paper/1803.01937/) (2018)
- [Deep Model Transferability from Attribution Maps](https://mlsanity.com/paper/1909.11902/) (2019)
- [Flacuna: Unleashing the Problem Solving Power of Vicuna using FLAN Fine-Tuning](https://mlsanity.com/paper/2307.02053/) (2023)
- [Using Similarity Measures to Select Pretraining Data for NER](https://mlsanity.com/paper/1904.00585/) (2019)
- [TaskWeb: Selecting Better Source Tasks for Multi-task NLP](https://mlsanity.com/paper/2305.13256/) (2023)
- [Data Similarity is Not Enough to Explain Language Model Performance](https://mlsanity.com/paper/2311.09006/) (2023)
- [D2PO: Discriminator-Guided DPO with Response Evaluation Models](https://mlsanity.com/paper/2405.01511/) (2024)

## Built on by (most cited first, 4 of 4)

- [Rethinking On-Policy Distillation of Large Language Models II: One Training Example](https://mlsanity.com/paper/2609.04172/) (2026)
- [Training-Free Model Merging for Multi-target Domain Adaptation](https://mlsanity.com/paper/2407.13771/) (2024)
- [Supervised Fine-Tuning Achieve Rapid Task Adaption Via Alternating Attention Head Activation Patterns](https://mlsanity.com/paper/2409.15820/) (2024)
- [Adapting Time Series Foundation Models through Data Mixtures](https://mlsanity.com/paper/2603.02840/) (2026)

## Abstract (arXiv)

Understanding alignment techniques begins with comprehending zero-shot generalization brought by instruction tuning, but little of the mechanism has been understood. Existing work has largely been confined to the task level, without considering that tasks are artificially defined and, to LLMs, merely consist of tokens and representations. To bridge this gap, we investigate zero-shot generalization from the perspective of the data itself. We first demonstrate that zero-shot generalization happens very early during instruction tuning, with loss serving as a stable indicator. Next, we investigate training data arrangement through similarity and granularity perspectives, confirming that the timing of exposure to certain training examples may greatly facilitate generalization on unseen tasks. Finally, we propose a more grounded training data arrangement framework, Test-centric Multi-turn Arrangement, and show its effectiveness in promoting continual learning and further loss reduction. For the first time, we show that zero-shot generalization during instruction tuning is a form of similarity-based generalization between training and test data at the instance level. Our code is released at https://github.com/thunlp/Dynamics-of-Zero-Shot-Generalization.