keyword
instruction embeddings
Instruction embeddings are dense vector representations that capture the semantic meaning, task objectives, and functional intent of natural language prompts or directives. In machine learning and natural language processing, these embeddings map written instructions into a continuous mathematical space, enabling algorithms to evaluate and compare tasks based on their underlying goals rather than surface-level wording. By converting task descriptions into measurable vectors, models can effectively quantify instruction complexity, measure task similarity across datasets, guide data selection during model fine-tuning, and optimize continual learning workflows to preserve capabilities across diverse tasks.
2 items
From Quantity to Quality: Boosting LLM Performance with Self-Guided Data Selection for Instruction Tuning
Ming Li, Yong Zhang, Zhitao Li, Jiuhai Chen, Lichang Chen, Ning Cheng, Jianzong Wang, Tianyi Zhou, Jing Xiao
Why you should read this
Proposes the Instruction-Following Difficulty metric to let large language models autonomously select high-impact training samples, outperforming full-dataset fine-tuning on Alpaca and WizardLM while using only ten percent of the data.
In the realm of Large Language Models (LLMs), the balance between instruction data quality and quantity is a focal point. Recognizing this, we introduce a self-guided methodology for LLMs to autonomously discern and select cherry samples from open-source datasets, effectively minimizing manual curation and potential cost for instruction tuning an LLM. Our key innovation, the Instruction-Following Difficulty (IFD) metric, emerges as a pivotal metric to identify discrepancies between a model's expected responses and its intrinsic generation capability. Through the application of IFD, cherry samples can be pinpointed, leading to a marked uptick in model training efficiency. Empirical validations on datasets like Alpaca and WizardLM underpin our findings; with a mere 10% of original data input, our strategy showcases improved results. This synthesis of self-guided cherry-picking and the IFD metric signifies a transformative leap in the instruction tuning of LLMs, promising both efficiency and resource-conscious advancements. Codes, data, and models are available.
Added
2026-09-28
InsCL: A Data-efficient Continual Learning Paradigm for Fine-tuning Large Language Models with Instructions
Yifan Wang, Yafei Liu, Chufan Shi, Haoling Li, Chen Chen, Haonan Lu, Yujiu Yang
Why you should read this
Introduces InsCL, a continual learning framework for large language models that dynamically allocates replay size using Wasserstein task similarity and prioritizes high-quality training examples via instruction complexity metrics to mitigate catastrophic forgetting.
Instruction tuning effectively optimizes Large Language Models (LLMs) for downstream tasks. Due to the changing environment in real-life applications, LLMs necessitate continual task-specific adaptation without catastrophic forgetting. Considering the heavy computational cost, replay-based Continual Learning (CL) methods are the simplest and most widely used for LLMs to address the forgetting issue. However, traditional replay-based methods do not fully utilize instructions to customize the replay strategy. In this work, we propose a novel paradigm called Instruction-based Continual Learning (InsCL). InsCL dynamically replays previous data based on task similarity, calculated by Wasserstein Distance with instructions. Moreover, we further introduce an Instruction Information Metric (InsInfo) to quantify the complexity and diversity of instructions. According to InsInfo, InsCL guides the replay process more inclined to high-quality data. We conduct extensive experiments over 16 tasks with different training orders, observing consistent performance improvements of InsCL. When all tasks have been trained, InsCL achieves performance gains of 3.0 Relative Gain compared with Random Replay, and 27.96 Relative Gain compared with No Replay.
Added
2026-09-26