arXiv subjects

Joschka Braun

Publications and source records attributed to Joschka Braun.

5 recordsLinked to original sources

arXiv2026

Beyond Multiple Choice: Evaluating Steering Vectors for Summarization

Steering vectors are a lightweight method for controlling text properties by adding a learned bias to language model activations at inference time. While predominantly studied for multiple-choice and toy tasks, their effectiveness in free-form generation remains largely unexplored. Moving "Beyond Multiple Choice," we evaluate steering vectors for controlling topical focus, sentiment, toxicity, and readability in abstractive summaries across the SAMSum, NEWTS, and arXiv datasets. We find that steering effectively controls targeted properties, but high steering strengths consistently induce degenerate repetition and factual hallucinations. Prompting alone preserves summary quality but offers weaker control. Combining both methods yields the strongest control and the most favorable efficacy-quality trade-off at moderate steering strengths. Our work demonstrates that steering vectors face a critical control-quality trade-off in free-form generation, and that hybrid approaches offer the best balance in practice.

cs.LG↗

arXiv2026

Exploration Hacking: Can LLMs Learn to Resist RL Training?

Reinforcement learning (RL) has become essential to the post-training of large language models (LLMs) for reasoning, agentic capabilities and alignment. Successful RL relies on sufficient exploration of diverse actions by the model during training, which creates a potential failure mode: a model could strategically alter its exploration during training to influence the subsequent training outcome. In this paper we study this behavior, called exploration hacking. First, we create model organisms of selective RL resistance by fine-tuning LLMs to follow specific underperformance strategies; these models can successfully resist our RL-based capability elicitation in agentic biosecurity and AI R&D environments while maintaining performance on related tasks. We then use our model organisms to evaluate detection and mitigation strategies, including monitoring, weight noising, and SFT-based elicitation. Finally, we show that current frontier models can exhibit explicit reasoning about suppressing their exploration when provided with sufficient information about their training context, with higher rates when this information is acquired indirectly through the environment. Together, our results suggest exploration hacking is a possible failure mode of RL on sufficiently capable LLMs.

cs.LG↗

arXiv2026

Understanding Unreliability of Steering Vectors in Language Models: Geometric Predictors and the Limits of Linear Approximations

Steering vectors are a lightweight method for controlling language model behavior by adding a learned bias to the activations at inference time. Although effective on average, steering effect sizes vary across samples and are unreliable for many target behaviors. In my thesis, I investigate why steering reliability differs across behaviors and how it is impacted by steering vector training data. First, I find that higher cosine similarity between training activation differences predicts more reliable steering. Second, I observe that behavior datasets where positive and negative activations are better separated along the steering direction are more reliably steerable. Finally, steering vectors trained on different prompt variations are directionally distinct, yet perform similarly well and exhibit correlated efficacy across datasets. My findings suggest that steering vectors are unreliable when the latent target behavior representation is not effectively approximated by the linear steering direction. Taken together, these insights offer a practical diagnostic for steering unreliability and motivate the development of more robust steering methods that explicitly account for non-linear latent behavior representations.

cs.CL↗

arXiv2025

Logit Reweighting for Topic-Focused Summarization

Generating abstractive summaries that adhere to a specific topic remains a significant challenge for language models. While standard approaches, such as fine-tuning, are resource-intensive, simpler methods like prompt engineering often struggle to maintain topical focus, particularly with smaller models. To address this, we propose a lightweight method that enhances topical relevance by directly reweighting the logits of topic-relevant tokens during generation. We evaluate three such reweighting techniques: Constant Shift, which adds a constant value to logits; Factor Scaling, which multiplies them by a factor; and Threshold Selection, which selectively boosts logits that exceed a probability threshold. Experiments on the NEWTS topical summarization dataset, using both Gemma-2B and Llama-3-8B models, show that these techniques effectively increase the use of topic-relevant vocabulary. Notably, the Threshold Selection method successfully improves topical focus without compromising summary quality-a trade-off often seen in other approaches. Our findings demonstrate that directly reweighting logits is a practical and resource-efficient alternative to fine-tuning, offering a promising pathway for precisely controlling the thematic content of generated text.

cs.LG↗

arXiv2025

Understanding (Un)Reliability of Steering Vectors in Language Models

Steering vectors are a lightweight method to control language model behavior by adding a learned bias to the activations at inference time. Although steering demonstrates promising performance, recent work shows that it can be unreliable or even counterproductive in some cases. This paper studies the influence of prompt types and the geometry of activation differences on steering reliability. First, we find that all seven prompt types used in our experiments produce a net positive steering effect, but exhibit high variance across samples, and often give an effect opposite of the desired one. No prompt type clearly outperforms the others, and yet the steering vectors resulting from the different prompt types often differ directionally (as measured by cosine similarity). Second, we show that higher cosine similarity between training set activation differences predicts more effective steering. Finally, we observe that datasets where positive and negative activations are better separated are more steerable. Our results suggest that vector steering is unreliable when the target behavior is not represented by a coherent direction.

cs.LG↗