CORTEXA
← Browse
arxivcs.DCcs.AIcs.LG2026-06-26

Optimizing Teacher-Student Partitioning for Scalable Knowledge Distillation on HPC Systems

Adrian P. Dieguez, Victor Conchello Vendrell, Alex Batlle, Vinnam Kim, Jordi Ros-Giralt, Harris Teague

Knowledge Distillation (KD) enables training smaller student models under the guidance of larger teacher models, and the widely adopted TRL library implements it. Yet, TRL treats both models symmetrically, missing opportunities to exploit their pronounced asymmetry in memory footprint, and communication requirements. This paper presents an HPC-aware methodology for KD that decouples teacher and student partitioning efficiently. Our approach achieves up to 67% higher samples-per-second than TRL by avoiding unnecessary teacher-model data structures and selecting the best split strategy. We combine vertical and horizontal partitioning of models, deriving an analytical expression that identifies the existence of inflection points between splitting regimes. These results showed that exploiting teacher--student asymmetry through topology-aware parallelism notably accelerated GKD training on production HPC clusters at our company

View free PDFSource page

Related papers

arxivcs.LGcs.AIcs.ARcs.DCcs.PFstat.CO2026-07-24

Optimizing Transformer Neural Network for Real-Time Outlier Detection on FPGAs

Ilia Sobakinskikh, Paul Alexander Bilokon

In this work, we explore how the inference time of a Transformer Neural Network can be efficiently optimized with applications to real-time anomaly detection in financial time series. The financial time series are price series such as asset prices. Unfortunately, the data is ofte…

View free PDFSource page
arxivcs.LGcs.AIcs.CL2026-07-24

Cross-Tokenizer On-Policy Distillation via Byte-Prefix Marginalization

Hao Wang, Kun Yuan, Wenlin Zhong, Minglei Zhang, Han Xiao, Ming Sun, et al.

Open-weight language models from different families exhibit complementary capabilities, motivating their consolidation into a compact student through on-policy distillation (OPD). However, full-vocabulary OPD typically assumes a shared tokenizer, while existing cross-tokenizer me…

View free PDFSource page
arxivcs.LGcs.AI2026-07-31

SAF-OPD: Stable Advantage Fusion for On-Policy Distillation

Yifan Ding, Xincheng Wei, Yoshua Y. Li, Ziheng Li, Yuquan Lu, Siyu Zhang, et al.

Reinforcement learning with verifiable rewards (RLVR) broadcasts a single response-level reward to every token, while on-policy distillation (OPD) scores each token against a stronger teacher for a dense advantage but caps performance at teacher quality and discourages exploratio…

View free PDFSource page
arxivcs.LGcs.AIcs.CV2026-07-31

SERUM: State Extraction and Refinement for User Modeling

Andy J. Phu, James Mooney, Karin de Langis, Khanh Chi Le, Dongyeop Kang

Agentic assistants capable of proactive, personalized interactions require structured models of user intent and workflow. However, building these models from raw, unstructured screen activity remains an open challenge. We present SERUM, a multi-pass framework that extracts finite…

View free PDFSource page
arxivcs.ETcs.AIcs.ARcs.DC2026-07-27

The SpiNNaker2 chip: a many-core platform for flexible and scalable brain-inspired computing

Stefan Scholze, Johannes Partzsch, Sebastian Höppner, Florian Kelber, Andreas Dixius, Marco Stolba, et al.

In deep learning, efficiency gets more and more important to compensate for the ongoing growth in model sizes and applications. Neuromorphic hardware has long been advocated as an upcoming alternative to deep networks, taking inspiration from the brain for achieving unprecedented…

View free PDFSource page