Papers with Alignment
Automatic Pair Construction for Contrastive Post-training (2024.findings-naacl)
Copied to clipboard
Canwen Xu, Corby Rosset, Ethan Chau, Luciano Corro, Shweti Mahajan, Julian McAuley, Jennifer Neville, Ahmed Awadallah, Nikhil Rao
| Challenge: | Large language models (LLMs) have unprecedented proficiency in a wide array of tasks. |
| Approach: | They propose a way to construct contrastive data using preference pairs from multiple models of varying strengths using SLiC and DPO. |
| Outcome: | The proposed method outperforms existing models like Orca in the comparison of SLiC and DPO with SFT baselines. |
Aligning Language Models to Explicitly Handle Ambiguity (2024.emnlp-main)
Copied to clipboard
Hyuhng Joon Kim, Youna Kim, Cheonbok Park, Junyeob Kim, Choonghyun Park, Kang Min Yoo, Sang-goo Lee, Taeuk Kim
| Challenge: | Large language models (LLMs) are not specifically trained to deal with ambiguous utterances . ambiguity can lead to varying interpretations of the same input based on different assumptions or background knowledge . |
| Approach: | They propose a pipeline that aligns large language models to manage ambiguous queries . they propose to use their own assessment of perceived ambiguity to detect and manage queries a . |
| Outcome: | Experimental results show that APA empowers LLMs to detect and manage ambiguous queries while retaining the ability to answer clear questions. |
Alignment Quality Index (AQI) : Beyond Refusals: AQI as an Intrinsic Alignment Diagnostic via Latent Geometry, Cluster Divergence, and Layer wise Pooled Representations (2025.emnlp-main)
Copied to clipboard
Abhilekh Borah, Chhavi Sharma, Danush Khanna, Utkarsh Bhatt, Gurpreet Singh, Hasnat Md Abdullah, Raghav Kaushik Ravi, Vinija Jain, Jyoti Patel, Shubham Singh, Vasu Sharma, Arpita Vats, Rahul Raja, Aman Chadha, Amitava Das
| Challenge: | a new metric measures the quality of large language models (LLMs) that detects hidden misalignments and jailbreak risks. |
| Approach: | They propose a decoding-invariant metric that measures latent safety failures . they propose 'Alignment Quality Index' to measure latent activations in latent space . |
| Outcome: | The proposed metric detects latent safety failures overlooked by behavioral benchmarks and jailbreaks. |
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. |
ARM: Alignment with Residual Energy-Based Model (2024.naacl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) acquire a wide range of abilities and abilities, but their behavior does not align with human preferences. |
| Approach: | They propose to minimize a forward Kullback–Leibler divergence from a target policy to a parameteric policy instead of a reverse KL as in RLHF methods. |
| Outcome: | The proposed method can learn an aligned policy by minimizing a forward Kullback–Leibler divergence from a target policy to a parameteric policy instead of a reverse KL as in RLHF methods. |
LIONs: An Empirically Optimized Approach to Align Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have focused on aligning large language models with pre-trained datasets. |
| Approach: | They conduct a rigorous analysis of a three-stage training pipeline using sequence packing, loss masking and increasing the preference dataset size in DPO to improve the performance of language models. |
| Outcome: | The proposed models outperform the official instruct models tuned with closed-source data and algorithms. |
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models (2026.findings-acl)
Copied to clipboard
Hengyuan Zhang, Zhihao Zhang, Ercong Nie, Mingyang Wang, Zunhai Su, Yiwei Wang, Qianli Wang, Shuzhou Yuan, Xufeng Duan, Qibo Xue, Zeping Yu, Chenming Shang, Xiao Liang, Jing Xiong, Hui Shen, Chaofan Tao, Zhengwu Liu, Senjie Jin, Zhiheng Xi, Dongdong Zhang, Sophia Ananiadou, Tao Gui, Ruobing Xie, Hayden Kwok-Hay So, Hinrich Schuetze, Xuanjing Huang, Qi Zhang, Ngai Wong
| Challenge: | Existing literature on mechanistic interpretation (MI) treats it as an observational science, leaving practical applications underexplored. |
| Approach: | They propose a survey structured around the pipeline to identify and improve MI models. |
| Outcome: | The proposed framework enables tangible improvements in Alignment, Capability, and Efficiency. |
AlignBench: Benchmarking Chinese Alignment of Large Language Models (2024.acl-long)
Copied to clipboard
Xiao Liu, Xuanyu Lei, Shengyuan Wang, Yue Huang, Andrew Feng, Bosi Wen, Jiale Cheng, Pei Ke, Yifan Xu, Weng Lam Tam, Xiaohan Zhang, Lichao Sun, Xiaotao Gu, Hongning Wang, Jing Zhang, Minlie Huang, Yuxiao Dong, Jie Tang
| Challenge: | Effective evaluation of alignment for emerging Chinese LLMs is still significantly lacking, calling for real-scenario grounded, open-ended, challenging and automatic evaluations tailored for alignment. |
| Approach: | They propose a multi-dimensional benchmark for evaluating LLMs’ alignment in Chinese with 8 main categories, 683 real-scenario rooted queries and corresponding human verified references. |
| Outcome: | The benchmark uses a human-in-the-loop data curation pipeline, 683 real-scenario rooted queries and human verified references. |
Aligners: Decoupling LLMs and Alignment (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) need to be aligned with human expectations to ensure their safety and utility in most applications. |
| Approach: | They propose to decouple LLMs and alignment by training *aligner* models that can be used to align any LLM on an as-needed basis. |
| Outcome: | The proposed model can be used to align any LLM for a given criteria on an as-needed basis. |
Enhancing Multimodal Retrieval via Complementary Information Extraction and Alignment (2025.acl-long)
Copied to clipboard
| Challenge: | Existing studies focus on capturing information in multimodal data that is similar to their paired texts, but often ignores the complementary information contained in multimodule data. |
| Approach: | They propose a multimodal retrieval approach that employs Complementary Information Extraction and Alignment to capture complementary information in multimodal data. |
| Outcome: | The proposed approach achieves significant improvements over divide-and-conquer models and universal dense retrieval models. |
Robustness and Confounders in the Demographic Alignment of LLMs with Human Perceptions of Offensiveness (2025.findings-acl)
Copied to clipboard
| Challenge: | despite evidence of demographic bias, reports with whom they align best are hard to generalize or contradictory . confounders introduced in the annotation process account for more variation in alignment patterns than demographic traits . |
| Approach: | They examine the alignment of large language models with human annotations in offensive language datasets. |
| Outcome: | The results show that LLMs align better with human annotations than other models. |
PLLuM-Align: Polish Preference Dataset for Large Language Model Alignment (2025.emnlp-main)
Copied to clipboard
Karolina Seweryn, Anna Kołos, Agnieszka Karlińska, Katarzyna Lorenc, Katarzyna Dziewulska, Maciej Chrabaszcz, Aleksandra Krasnodebska, Paula Betscher, Zofia Cieślińska, Katarzyna Kowol, Julia Moska, Dawid Motyka, Paweł Walkowiak, Bartosz Żuk, Arkadiusz Janz
| Challenge: | Large language models generate preferred responses while avoiding harmful or inappropriate outputs, despite their ability to generate cross-language transferability. |
| Approach: | They introduce the first Polish preference dataset PLLuM-Align, created entirely through human annotation to reflect Polish language and cultural nuances. |
| Outcome: | The proposed dataset lays the groundwork for more aligned Polish LLMs and contributes to the broader goal of multilingual alignment in underrepresented languages. |
Segment, Embed, and Align: A Universal Recipe for Aligning Subtitles to Signing (2026.acl-long)
Copied to clipboard
| Challenge: | Existing approaches for aligning spoken language text to sign language videos rely on end-to-end training tied to a specific language or dataset. |
| Approach: | They propose a universal approach for aligning spoken language text with corresponding timestamps to sign language videos using a lightweight dynamic programming procedure. |
| Outcome: | The proposed method can be used on four sign language datasets and is highly efficient on CPU. |