Patches of Nonlinearity: Instruction Vectors in Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Despite the success of instruction-tuned language models, little is known about how they process instructions internally. |
| Approach: | They propose a method to localize instruction processing in language models that is free from patching assumptions. |
| Outcome: | The proposed method disentangles the implicit linear assumptions of patching-based techniques. |
Similar Papers
Demystifying Instruction Mixing for Fine-tuning Large Language Models (2024.acl-srw)
Copied to clipboard
| Challenge: | Instruction tuning is effective for aligning large language models with human instructions, but the procedure to optimizing the mixing of instruction datasets is still unclear. |
| Approach: | They categorize instructions into three primary types: NLP downstream tasks, coding, and general chat. |
| Outcome: | The proposed method improves performance of large language models (LLMs) but it is difficult to combine different instruction datasets to optimize overall performance. |
Layer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Pre-trained large language models retain task-specific knowledge, but where and to what extent they retain it remains unexplored. |
| Approach: | They investigate the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across over 60 NLP tasks. |
| Outcome: | The results show that pre-trained models retain task-specific knowledge . some tasks are already encoded in pre-train models, but others benefit from instruction tuning. |
Fine-Tuning Large Language Models with Sequential Instructions (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing instruction-tuned models struggle to adhere to a query with multiple intentions, which impairs their performance when the completion of several tasks is demanded by a single command. |
| Approach: | They develop an automatic process that turns existing data into diverse and complex task chains and a new benchmark to evaluate a model’s ability to follow all the instructions in a sequence. |
| Outcome: | The proposed model can follow instructions better and deliver higher results in coding, maths, and open-ended generation. |
From Language Modeling to Instruction Following: Understanding the Behavior Shift in LLMs after Instruction Tuning (2024.naacl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have achieved remarkable success in aligning with user intentions. |
| Approach: | They develop local and global explanation methods and a feed-forward-based method for input-output attribution to investigate the impact of instruction tuning on user intentions. |
| Outcome: | The proposed method compares explanations from pre-trained and instruction-tuned models . it empowers LLMs to recognize the instruction parts of user prompts, it encourages response generation . |
Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks (2025.acl-srw)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) excel in math reasoning problemsolving, text generation, summarization, creative writing, among other tasks. |
| Approach: | They evaluate Direct Preference Optimization and its variants for aligning Large Language Models with human preferences. |
| Outcome: | The proposed alignment methods achieve near-optimal performance even with smaller subsets of training data. |
Instruction Position Matters in Sequence Generation with Large Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) can perform conditional sequence generation tasks, such as translation or summarization, through instruction fine-tuning. |
| Approach: | They propose to shift the position of task instructions after the input sentences to enhance the model's instruction-following capability. |
| Outcome: | The proposed method outperforms traditional settings across various model scales (1B / 7B & 13B) and different sequence generation tasks (translation and summarization) without any additional data or annotation costs. |
SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe (2026.acl-long)
Copied to clipboard
| Challenge: | Efforts to improve instruction tuning often focus on higher-quality supervised fine-tuning datasets, typically requiring data filtering with proprietary LLMs or human annotation. |
| Approach: | They propose a Mixup-based recipe that elevates LLM instruction tuning without relying on well-curated datasets. |
| Outcome: | The proposed model improves instruction-following and healthcare-specific tasks with consistent improvements across LLM families and SFT datasets. |
A Deep Dive into the Trade-Offs of Parameter-Efficient Preference Alignment Techniques (2024.acl-long)
Copied to clipboard
| Challenge: | Large language models are pre-trained on trillions of tokens and instruction-tuned or aligned to specific preferences. |
| Approach: | They propose guidelines to help researchers perform more effective parameter-efficient LLM alignment. |
| Outcome: | The proposed methods outperform preference optimization and outperformed pre-trained models on three key axes. |
Revealing the Inherent Instructability of Pre-Trained Language Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Pre-trained large language models perform multitask learning during their pre-training . a new technique, Response Tuning, removes the instruction and its corresponding mapping to the response from instruction tuning. |
| Approach: | They propose a method which removes the instruction and its mapping to the response from instruction tuning. |
| Outcome: | The proposed model can respond to a wide range of instructions . it can recognize and reject unsafe queries after learning from response data. |
Interpret and Improve In-Context Learning via the Lens of Input-Label Mappings (2025.acl-long)
Copied to clipboard
| Challenge: | Large language models excel at downstream NLP tasks through in-context learning . however, the internal mechanisms behind ICL remain under-explored . |
| Approach: | They propose a PC patching approach to identify modules where input-label mappings function . they observe and verify that key heads utilize input-labeled mappings to generate target labels for new queries. |
| Outcome: | The proposed approach detects modules where input-label mappings function . it also detects that key heads use the mappings to generate labels for new queries . |