8 papers
On-Policy Attention Linearization
Arian Raje, Anupam Nayak, Anthony Fei +3
Hybrid transformer architectures that replace most softmax attention layers with linear attention offer transformer-level quality at a fraction of the memory cost. Rather than pret…
Emergent and Subliminal Misalignment Through the Lens of Data-Mediated Transfer
Baris Askin, Muhammed Ustaomeroglu, Anupam Nayak +3
Fine-tuning LLMs on narrow harmful datasets can induce Emergent Misalignment (EM), where models exhibit misaligned behavior far beyond the fine-tuning distribution. We argue that e…
PubSwap: Public-Data Off-Policy Coordination for Federated RLVR
Anupam Nayak, Baris Askin, Muhammed Ustaomeroglu +2
Reasoning post-training with reinforcement learning from verifiable rewards (RLVR) is typically studied in centralized settings, yet many realistic applications involve decentraliz…
Not All Turns Are Equally Hard: Adaptive Thinking Budgets For Efficient Multi-Turn Reasoning in Agents
Neharika Jali, Anupam Nayak, Gauri Joshi
As LLM reasoning performance plateaus, improving inference-time compute efficiency is crucial to mitigate overthinking and long thinking traces even for simple queries. Prior appro…
MELINOE: Fine-Tuning Enables Memory-Efficient Inference for Mixture-of-Experts Models
Arian Raje, Anupam Nayak, Gauri Joshi
Mixture-of-Experts (MoE) model architectures can significantly reduce the number of activated parameters per token, enabling computationally efficient training and inference. Howev…
Federate the Router: Learning Language Model Routers with Sparse and Decentralized Evaluations
Baris Askin, Shivam Patel, Anupam Nayak +4
Large language models (LLMs) are increasingly accessed as remotely hosted services by edge and enterprise clients that cannot run frontier models locally. Since models vary widely…
Global PIQA: Evaluating Commonsense Reasoning Across 100+ Languages and Cultures
Tyler A. Chang, Catherine Arnett, Abdelrahman Sadallah +377
To date, there exist almost no culturally-specific evaluation benchmarks for large language models (LLMs) that cover a large number of languages and cultures. In this paper, we pre…
Achieving Logarithmic Regret in KL-Regularized Zero-Sum Markov Games
Anupam Nayak, Tong Yang, Osman Yagan +2
Reverse Kullback-Leibler (KL) divergence-based regularization with respect to a fixed reference policy is widely used in modern reinforcement learning to preserve the desired trait…