On the Effect of Instruction Tuning Loss on Generalization

• URL: http://arxiv.org/abs/2507.07817v2
• Date: Tue, 15 Jul 2025 11:42:05 GMT
• Title: On the Effect of Instruction Tuning Loss on Generalization
• Authors: Anwoy Chatterjee, H S V N S Kowndinya Renduchintala, Sumit Bhatia, Tanmoy Chakraborty,
• Abstract summary: We show that the standard instruction tuning loss often yields suboptimal performance and limited robustness to input prompt variations.<n>We find that a low-to-moderate weight for prompt tokens coupled with a moderate-to-high weight for response tokens yields the best-performing models across settings.
• Score: 22.288479270814484
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Instruction Tuning has emerged as a pivotal post-training paradigm that enables pre-trained language models to better follow user instructions. Despite its significance, little attention has been given to optimizing the loss function used. A fundamental, yet often overlooked, question is whether the conventional auto-regressive objective - where loss is computed only on response tokens, excluding prompt tokens - is truly optimal for instruction tuning. In this work, we systematically investigate the impact of differentially weighting prompt and response tokens in instruction tuning loss, and propose Weighted Instruction Tuning (WIT) as a better alternative to conventional instruction tuning. Through extensive experiments on five language models of different families and scale, three finetuning datasets of different sizes, and five diverse evaluation benchmarks, we show that the standard instruction tuning loss often yields suboptimal performance and limited robustness to input prompt variations. We find that a low-to-moderate weight for prompt tokens coupled with a moderate-to-high weight for response tokens yields the best-performing models across settings and also serve as better starting points for the subsequent preference alignment training. These findings highlight the need to reconsider instruction tuning loss and offer actionable insights for developing more robust and generalizable models. Our code is open-sourced at https://github.com/kowndinya-renduchintala/WIT.

Related papers

• IGD: Token Decisiveness Modeling via Information Gain in LLMs for Personalized Recommendation [70.2753541780788]
  We introduce an Information Gain-based Decisiveness-aware Token handling (IGD) strategy that integrates token decisiveness into both tuning and decoding.<n>IGD consistently improves recommendation accuracy, achieving significant gains on widely used ranking metrics compared to strong baselines.
  arXiv  Detail & Related papers  (2025-06-16T08:28:19Z)
• The First Few Tokens Are All You Need: An Efficient and Effective Unsupervised Prefix Fine-Tuning Method for Reasoning Models [69.798277882245]
  We introduce Unsupervised Prefix Fine-Tuning (UPFT) to enhance large language models' reasoning efficiency.<n>UPFT removes the need for labeled data or exhaustive sampling.<n> Experiments show that UPFT matches the performance of supervised methods.
  arXiv  Detail & Related papers  (2025-03-04T18:56:03Z)
• Context-aware Prompt Tuning: Advancing In-Context Learning with Adversarial Methods [69.36397993451742]
  This work introduces Context-aware Prompt Tuning (CPT), a method inspired by ICL, PT, and adversarial attacks. We modify specific context tokens, considering the unique structure of input and output formats. Inspired by adversarial attacks, we adjust the input based on the labels present in the context, focusing on minimizing, rather than maximizing, the loss.
  arXiv  Detail & Related papers  (2024-10-22T17:45:47Z)
• Hard Prompts Made Interpretable: Sparse Entropy Regularization for Prompt Tuning with RL [29.01858866450715]
  We present RLPrompt, which aims to find optimal prompt tokens leveraging soft Q-learning. While the results show promise, we have observed that the prompts frequently appear unnatural, which impedes their interpretability. We address this limitation by using sparse Tsallis entropy regularization, a principled approach to filtering out unlikely tokens from consideration.
  arXiv  Detail & Related papers  (2024-07-20T03:10:19Z)
• Revisiting the Power of Prompt for Visual Tuning [50.11465784194896]
  This study explores the correlation evolvement between prompts and patch tokens during proficient training. Inspired by the observation that the prompt tokens tend to share high mutual information with patch tokens, we propose initializing prompts with downstream token prototypes. Our method significantly advances the adaptation for self-supervised pretraining, achieving impressive task performance gains of at least 10% to 30%.
  arXiv  Detail & Related papers  (2024-02-04T07:49:02Z)
• Instructive Decoding: Instruction-Tuned Large Language Models are Self-Refiner from Noisy Instructions [26.192531184689763]
  This paper presents Instructive Decoding (ID), a simple yet effective approach that augments the efficacy of instruction-tuned models. ID adjusts the logits for next-token prediction in a contrastive manner, utilizing predictions generated from a manipulated version of the original instruction. We conduct experiments across a spectrum of such noisy instructions, ranging from those that insert semantic noise via random words to others like 'opposite' that elicit deviated responses.
  arXiv  Detail & Related papers  (2023-11-01T02:31:35Z)
• Self-supervised Meta-Prompt Learning with Meta-Gradient Regularization for Few-shot Generalization [40.45470744120691]
  Self-sUpervised meta-Prompt learning framework with MEta-gradient Regularization for few-shot generalization (SUPMER) This paper proposes a novel Self-sUpervised meta-Prompt learning framework with MEta-gradient Regularization for few-shot generalization (SUPMER)
  arXiv  Detail & Related papers  (2023-03-22T05:04:21Z)
• Gradient-Regulated Meta-Prompt Learning for Generalizable Vision-Language Models [137.74524357614285]
  We introduce a novel Gradient-RegulAted Meta-prompt learning framework. It helps pre-training models adapt to downstream tasks in a parameter -- and data -- efficient way. GRAM can be easily incorporated into various prompt tuning methods in a model-agnostic way.
  arXiv  Detail & Related papers  (2023-03-12T05:03:37Z)
• The Wisdom of Hindsight Makes Language Models Better Instruction Followers [84.9120606803906]
  Reinforcement learning has seen wide success in finetuning large language models to better align with instructions via human feedback. In this paper, we consider an alternative approach: converting feedback to instruction by relabeling the original one and training the model for better alignment in a supervised manner. We propose Hindsight Instruction Relabeling (HIR), a novel algorithm for aligning language models with instructions.
  arXiv  Detail & Related papers  (2023-02-10T12:16:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.