Yftah Ziser
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yftah Ziser has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Shay B. Cohen
School of Informatics University of Edinburgh
3 shared publications - Zheng Zhao
University of Edinburgh
3 shared publications - Amrita Bhattacharjee
Arizona State University
2 shared publications - Anum Afzal
Technical University of Munich Technical University of Munich Nvidia Research & Bar-Ilan University Nvidia Research
2 shared publications - Christopher Parisien
School of Computing and AI Arizona State University
2 shared publications - Florian Matthes
Technical University of Munich
2 shared publications
Latest Publications
- Open Access
Knowing Before Saying: LLM Representations Encode Information About Chain-of-Thought Success Before Completion
We investigate whether the success of a zero-shot Chain-of-Thought (CoT) process can be predicted before completion. We discover that a probing classifier, based on LLM representations, performs well \emph{even before a single token is generated}, suggesting that crucial information about the reasoning process is already present in the initial steps representations. In contrast, a strong BERT-based baseline, which relies solely on the generated tokens, performs worse, likely because it depends on shallow linguistic cues rather than deeper reasoning dynamics. Surprisingly, using later reasoning steps does not always improve classification. When additional context is unhelpful, earlier representations resemble later ones more, suggesting LLMs encode key information early. This implies reasoning can often stop early without loss. To test this, we conduct early stopping experiments, showing that truncating CoT reasoning still improves performance over not using CoT at all, though a gap remains compared to full reasoning. However, approaches like supervised learning or reinforcement learning designed to shorten CoT chains could leverage our classifier's guidance to identify when early stopping is effective. Our findings provide insights that may support such methods, helping to optimize CoT's efficiency while preserving its benefits.
Open AccessKnowing Before Saying: LLM Representations Encode Information About Chain-of-Thought Success Before Completion
We investigate whether the success of a zero-shot Chain-of-Thought (CoT) process can be predicted before completion. We discover that a probing classifier, based on LLM representations, performs well \emph{even before a single token is generated}, suggesting that crucial information about the reasoning process is already present in the initial steps representations. In contrast, a strong BERT-based baseline, which relies solely on the generated tokens, performs worse, likely because it depends on shallow linguistic cues rather than deeper reasoning dynamics. Surprisingly, using later reasoning steps does not always improve classification. When additional context is unhelpful, earlier representations resemble later ones more, suggesting LLMs encode key information early. This implies reasoning can often stop early without loss. To test this, we conduct early stopping experiments, showing that truncating CoT reasoning still improves performance over not using CoT at all, though a gap remains compared to full reasoning. However, approaches like supervised learning or reinforcement learning designed to shorten CoT chains could leverage our classifier's guidance to identify when early stopping is effective. Our findings provide insights that may support such methods, helping to optimize CoT's efficiency while preserving its benefits.
- Open Access
SafeSteer: Interpretable Safety Steering with Refusal-Evasion in LLMs
Fine-tuning large language models (LLMs) to adapt to evolving safety policies is costly and impractical. Mechanistic interpretability enables inference-time control through latent activation steering, yet its potential for precise, customizable safety adjustments remains largely untapped. This paper investigates an approach called SafeSteer for guiding the outputs of LLMs by: (i) leveraging category-specific steering vectors for more precise control, (ii) employing a simple, gradient-free unsupervised method to enhance safety steering while preserving text quality, topic relevance, and without explicit refusal, and (iii) accomplishing this without a hard requirement of contrastive pairwise safe data. We also highlight that our method, being simple and effective, aligns with recent studies suggesting that simple techniques often outperform more complex ones in activation steering. We showcase the effectiveness of our approach across various LLMs, datasets, and risk categories, demonstrating its ability to provide precise control, prevent blanket refusals, and guide models toward generating safe content while maintaining topic relevance.
Open AccessSafeSteer: Interpretable Safety Steering with Refusal-Evasion in LLMs
Fine-tuning large language models (LLMs) to adapt to evolving safety policies is costly and impractical. Mechanistic interpretability enables inference-time control through latent activation steering, yet its potential for precise, customizable safety adjustments remains largely untapped. This paper investigates an approach called SafeSteer for guiding the outputs of LLMs by: (i) leveraging category-specific steering vectors for more precise control, (ii) employing a simple, gradient-free unsupervised method to enhance safety steering while preserving text quality, topic relevance, and without explicit refusal, and (iii) accomplishing this without a hard requirement of contrastive pairwise safe data. We also highlight that our method, being simple and effective, aligns with recent studies suggesting that simple techniques often outperform more complex ones in activation steering. We showcase the effectiveness of our approach across various LLMs, datasets, and risk categories, demonstrating its ability to provide precise control, prevent blanket refusals, and guide models toward generating safe content while maintaining topic relevance.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 29218-29243
Iterative Multilingual Spectral Attribute Erasure
Shun Shao, Yftah Ziser, Zheng Zhao, Yifu Qiu, Shay B Cohen, Anna Korhonen. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 29218-29243
Open AccessIterative Multilingual Spectral Attribute Erasure
Shun Shao, Yftah Ziser, Zheng Zhao, Yifu Qiu, Shay B Cohen, Anna Korhonen. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 35116-35136
A Simple Yet Effective Method for Non-Refusing Context Relevant Fine-grained Safety Steering in LLMs
Shaona Ghosh, Amrita Bhattacharjee, Yftah Ziser, Christopher Parisien. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 35116-35136
Open AccessA Simple Yet Effective Method for Non-Refusing Context Relevant Fine-grained Safety Steering in LLMs
Shaona Ghosh, Amrita Bhattacharjee, Yftah Ziser, Christopher Parisien. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 12791-12806
Knowing Before Saying: LLM Representations Encode Information About Chain-of-Thought Success Before Completion
Anum Afzal, Florian Matthes, Gal Chechik, Yftah Ziser. Findings of the Association for Computational Linguistics: ACL 2025. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 12791-12806
Open AccessKnowing Before Saying: LLM Representations Encode Information About Chain-of-Thought Success Before Completion
Anum Afzal, Florian Matthes, Gal Chechik, Yftah Ziser. Findings of the Association for Computational Linguistics: ACL 2025. 2025.
- Open Access
Layer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models
Fine-tuning pre-trained large language models (LLMs) on a diverse array of tasks has become a common approach for building models that can solve various natural language processing (NLP) tasks. However, where and to what extent these models retain task-specific knowledge remains largely unexplored. This study investigates the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across a diverse set of over 60 NLP tasks. We use a set of matrix analysis tools to examine the differences between the way pre-trained and instruction-tuned LLMs store task-specific information. Our findings reveal that while some tasks are already encoded within the pre-trained LLMs, others greatly benefit from instruction tuning. Additionally, we pinpointed the layers in which the model transitions from high-level general representations to more task-oriented representations. This finding extends our understanding of the governing mechanisms of LLMs and facilitates future research in the fields of parameter-efficient transfer learning and multi-task learning.
Open AccessLayer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models
Fine-tuning pre-trained large language models (LLMs) on a diverse array of tasks has become a common approach for building models that can solve various natural language processing (NLP) tasks. However, where and to what extent these models retain task-specific knowledge remains largely unexplored. This study investigates the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across a diverse set of over 60 NLP tasks. We use a set of matrix analysis tools to examine the differences between the way pre-trained and instruction-tuned LLMs store task-specific information. Our findings reveal that while some tasks are already encoded within the pre-trained LLMs, others greatly benefit from instruction tuning. Additionally, we pinpointed the layers in which the model transitions from high-level general representations to more task-oriented representations. This finding extends our understanding of the governing mechanisms of LLMs and facilitates future research in the fields of parameter-efficient transfer learning and multi-task learning.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 15195-15214
Layer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models
Zheng Zhao, Yftah Ziser, Shay B Cohen. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 15195-15214
Open AccessLayer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models
Zheng Zhao, Yftah Ziser, Shay B Cohen. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.