5 papers
Broken Chains: The Cost of Incomplete Reasoning in LLMs
Ian Su, Gaurav Purushothaman, Jey Narayan +5
Reasoning-specialized models like OpenAI's 5.1 and DeepSeek-V3.2 allocate substantial inference compute to extended chain-of-thought (CoT) traces, yet reasoning tokens incur signif…
Calibrated Reasoning: An Explanatory Verifier for Dynamic and Efficient Problem-Solving
Anisha Garg, Engin Tekin, Yash More +3
Advanced test-time computing strategies are essential for scaling reasoning models, but their effectiveness is capped by the models' poor self-evaluation. We propose a pairwise Exp…
Towards More Realistic Extraction Attacks: An Adversarial Perspective
Yash More, Prakhar Ganesh, Golnoosh Farnadi
Language models are prone to memorizing their training data, making them vulnerable to extraction attacks. While existing research often examines isolated setups, such as a single…
Beyond the Safety Bundle: Auditing the Helpful and Harmless Dataset
Khaoula Chehbouni, Jonathan Colaço Carr, Yash More +2
In an effort to mitigate the harms of large language models (LLMs), learning from human feedback (LHF) has been used to steer LLMs towards outputs that are intended to be both less…
Combining Domain and Alignment Vectors to Achieve Better Knowledge-Safety Trade-offs in LLMs
Megh Thakkar, Quentin Fournier, Matthew Riemer +4
There is a growing interest in training domain-expert LLMs that excel in specific technical fields compared to their general-purpose instruction-tuned counterparts. However, these…