Steering Language Models in Multi-Token Generation: A Case Study on Tense and Aspect (2025.emnlp-main)
Copied to clipboard
| Challenge: | Prior work has focused largely on binary grammatical contrasts, but how do they encode their syntactic knowledge internally? |
| Approach: | They propose to use a multidimensional hierarchical grammar phenomenon to identify distinct, orthogonal directions in residual space to demonstrate causal control over both grammatical features. |
| Outcome: | The proposed model can encode tense and aspect in human-like ways, but effective steering during generation is sensitive to multiple factors and requires manual tuning or automated optimization. |
Similar Papers
Large Language Models Share Representations of Latent Grammatical Concepts Across Typologically Diverse Languages (2025.naacl-long)
Copied to clipboard
| Challenge: | In the brains of human bilinguals, syntax processing may occur in similar regions for their first and second language, depending on factors like when the second language was learned and language proficiency. |
| Approach: | They propose to use sparse autoencoders to train Llama-3-8B and Aya-23-8B models to train multilingual models that share morphsyntactic representations of grammatical concepts. |
| Outcome: | The proposed model can predict plural verbs in different languages by activating the same plural feature. |
Style Vectors for Steering Generative Large Language Models (2024.findings-eacl)
Copied to clipboard
Kai Konen, Sophie Jentzsch, Diaoulé Diallo, Peer Schütt, Oliver Bensch, Roxanne El Baff, Dominik Opitz, Tobias Hecking
| Challenge: | Large language models (LLMs) can be trained on vast corpora and can generate text in a nuanced and parameterisable way. |
| Approach: | They propose to add style vectors to the activations of hidden layers during text generation to steer output towards specific styles. |
| Outcome: | The proposed approach differs from prompt engineering in that it can be nuanced and parameterisable. |
Dynamic Steering With Episodic Memory For Large Language Models (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing activation steering methods apply a single sentence-level steering vector uniformly across all tokens, ignoring LLMs’ token-wise, auto-regressive nature. |
| Approach: | They propose a framework that aligns LLMs to given demonstrations by steering at the token level conditioned on the input query. |
| Outcome: | The proposed framework surpasses baselines across safety, style transfer, and role-playing tasks, demonstrating improved alignment as demonstration scales. |
Are Large Language Models Capable of Generating Human-Level Narratives? (2024.emnlp-main)
Copied to clipboard
Yufei Tian, Tenghao Huang, Miri Liu, Derek Jiang, Alexander Spangher, Muhao Chen, Jonathan May, Nanyun Peng
| Challenge: | a recent HCI study has pointed to gaps in machine storytelling ability at the global level . authors show that LLMs have less suspense and less tension than human stories . |
| Approach: | They propose a computational framework to analyze narratives through three discourse-level aspects. |
| Outcome: | The proposed framework analyzes narratives through three discourse-level aspects . it shows that LLMs fall short of human abilities in discourse understanding . |
Step-by-Step: Controlling Arbitrary Style in Text with Large Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods for autoregressive text generation have low controllability and accumulating errors. |
| Approach: | They propose a three-stage prompt-based approach to express autoregressive text in a specific region editing task using a word frequency-based strategy. |
| Outcome: | Experiments on publicly competitive datasets confirm that the proposed approach achieves state-of-the-art performance. |
Beyond Linear Steering: Unified Multi-Attribute Control for Language Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Empirical results show that K-Steering outperforms strong baselines in accurately steering multiple behaviors. |
| Approach: | They propose a method that trains a single classifier on hidden activations and computes intervention directions via gradients at inference time. |
| Outcome: | The proposed method outperforms strong baselines in steering multiple behaviors. |
Bridging Internal Consistency and External Alignment: A Causal and Dynamic Interpretability Framework for LLM Generation (2026.acl-long)
Copied to clipboard
| Challenge: | Existing interpretability methods focus on internal and external aspects of the model . existing explanations often focus on surface correlations or static dependencies . |
| Approach: | They propose a causal and dynamic interpretability framework for Large Language Models . they characterize backdoor-adjusted causal effects of generated prefix and prompt . |
| Outcome: | The proposed framework provides a unified causal view of internal consistency and external alignment in LLM generation dynamics. |
Are the Values of LLMs Structurally Aligned with Humans? A Causal Perspective (2025.findings-acl)
Copied to clipboard
Yipeng Kang, Junqi Wang, Yexin Li, Mengmeng Wang, Wenming Tu, Quansen Wang, Hengli Li, Tingjun Wu, Xue Feng, Fangwei Zhong, Zilong Zheng
| Challenge: | Current approaches to value alignment focus on a few core values, such as helpfulness, harmlessness, and honesty. |
| Approach: | They propose to use latent causal value graphs to guide two lightweight value-steering methods . role-based prompting and sparse autoencoder (SAE) steering are also used . |
| Outcome: | Experiments on Gemma-2B-IT and Llama3-8B- IT show that the proposed methods are effective and controllable. |
Grammar-Based Code Representation: Is It a Worthy Pursuit for LLMs? (2025.findings-acl)
Copied to clipboard
Qingyuan Liang, Zhao Zhang, Zeyu Sun, Zheng Lin, Qi Luo, Xiao Yueyi, Yizhou Chen, Yuqun Zhang, Haotian Zhang, Lu Zhang, Chenbin Chenbin, Yingfei Xiong
| Challenge: | Existing research demonstrates the effectiveness of grammar-based code representations in small-scale models, showing their ability to reduce syntax errors and enhance performance. |
| Approach: | They develop a series of billion-scale grammar-based code representations that incorporate grammar rules into the code generation process. |
| Outcome: | Experiments on HumanEval and MBPP show that grammar-based representations reduce syntax errors and improve performance even in billion-scale models. |
How LLMs Comprehend Temporal Meaning in Narratives: A Case Study in Cognitive Evaluation of LLMs (2025.acl-long)
Copied to clipboard
| Challenge: | Large language models exhibit increasingly sophisticated linguistic capabilities, yet the extent to which these models reflect human-like cognition versus advanced pattern recognition remains an open question. |
| Approach: | They conduct a series of targeted experiments to assess whether LLMs construct semantic representations and pragmatic inferences in a human-like manner. |
| Outcome: | The proposed framework can be used to assess the cognitive and linguistic capabilities of large language models (LLMs). |