Large language models (LLMs) deployed as conversational agents are increasingly expected to follow procedural logic. While factual hallucination has received extensive attention, a different failure mode has been overlooked: procedural hallucination, in which an agent produces individually fluent and factually plausibl...
Muon's gains in data efficiency motivate evaluating its performance under explicit data, time, and tuning budgets. On a single A100 GPU, we compare a reference single-device Muon implementation with AdamW across 17.7M- and 33.6M-parameter causal language models and 2.70M-parameter vision transformers. A common 23-candi...
M. Akhtar, H. Nounou, M. Mansouri et al.· 0 citations
Instead of learning a mapping from instances to outcomes in the standard way, the key idea of pairwise difference learning (PDL) is to learn a function that takes two instances as input and predicts the difference between the respective outcomes. Given a function of this kind, predictions for a query instance are deriv...
Mohamed Karim Belaid, Maximilian Rabus, Eyke H\"ullermeier· 0 citations
Muon replaces a matrix gradient by an approximate polar factor and has shown strong empirical performance in neural-network training. The Isotropic Curvature Model (ICM) provides a curvature-based explanation for this update by asking when gradient orthogonalization minimizes a one-step local-loss surrogate. We investi...
Ao Xu· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
SpecAgent is introduced, a collaboration protocol that turns the strong model from the author of each step into a reviewer of steps drafted by a weak model, suggesting that a strong model need not author every step of an agent: reviewing grounded drafts retains most of its quality while cutting both latency and memory...
Zi-Lin Wang, Yinhong Liu, Chao Chen et al.· 0 citations
In reinforcement learning, advantage-based value-learning methods, unlike standard Q-learning, share information across actions by decomposing the action-value function into a state-dependent value component and action-specific advantages. While this mechanism has proven effective in both tabular and deep reinforcement...
Kintan Saha, Hamed Masoudi, Xinyi Sheng et al.· 0 citations
To improve task performance while encouraging desirable response properties, LLM post-training often combines a primary reward (e.g., correctness) with auxiliary rewards (e.g., brevity or efficiency). However, how these rewards are aggregated can critically affect training outcomes. We show that symmetric aggregation,...
Yu-Shu Li, Jiin Woo, Xiao-Xiao Li et al.· 0 citations
Memory enables large language models (LLMs) to use information beyond their parameters, such as past user conversations, long documents, or an agent's earlier actions. For a frozen LLM, access to this accumulated knowledge is limited to what a memory system places in its context window. However, gathering and combining...
Adibvafa Fallahpour, Parsa Idehpour, Vignesh Kothapalli et al.· 0 citations
On-policy distillation (OPD) transfers teacher knowledge to a student through feedback on student-generated rollouts, but generating complete rollouts and computing per-token teacher feedback are computationally expensive. Existing truncation methods save computation by stopping rollouts early. However, they do not com...
Zheng-Xin Zhang, Mao-Sen Tang, Ning Wang et al.· 0 citations
Sparse autoencoders (SAEs) represent model activations with sparse, linearly decoded features, but the downstream effects of those features need not compose additively. We introduce SAE Additivity Residuals (SAR), which measure the behavioral variation left unexplained above a chosen interaction order. Under independen...
Large Language Models (LLMs) demand efficient inference, where the multiply-accumulate (MAC) operations in linear projections dominate compute and energy cost. Post-training quantization (PTQ) reduces this cost by mapping weights and activations to low-bit integers. Power-of-Two (PoT) weight quantization promises effic...
Sunghyun Wee, Geunjae Choi, Hyeonjin Kim et al.· 0 citations
The largest System-1 models match or exceed fast closed LLMs in non-agentic accuracy, with better-calibrated probabilities at one-quarter of the latency, and a reasoning LLM is the most accurate, at about 70x the cost of a fast LLM.
Tian-Yu Chen, Xin-Yang Liu, Ru-Yi An et al.· 1 citation
A machine-learning approach called CheMeleon provides a practical way to develop new drugs or other molecules with much less detailed, expensive experimentation.
Research Manager Mike Walker reflects on a career spanning hacker culture, cybersecurity, and AI, sharing lessons about trust, teamwork, leadership, and why solving difficult problems requires curiosity, judgment, agility—and an understanding of the rules. The post Learn the rules: The hacker mindset appeared first on Microsoft Research.
Learn how Google Antigravity and our Cheminformatics Skills accelerate drug discovery workflows, search compounds, and propose novel rigidified macrocycles.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
