← Back to all news
arXiv Computation and Language By Juan S. Santillana

Conditional Calibration Contracts under Coupled Load: A Branch-Aware Audit of Switched Inference

Read the original on arXiv Computation and Language →

The paper investigates conditional calibration contracts for switched inference systems, proposing a branch‑aware audit framework called CA‑MHD that integrates load‑coupled distillation, closed‑loop replay, and a cost‑informed Schmitt router. Experiments across 40 cells reveal that the deployed wide branch consistently underperforms the original teacher in ordinal accuracy, while student‑only serving shows mixed improvements and routing exposes significant violation tails. Retrospective temporal recalibration narrows the ordinal gap in most cells, yet causal delayed‑label verification and incremental per‑mode benefits remain unproven.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

Related stories

arXiv AI
Oct 21

Building Telecom Language Models with RSOD: Domain Data, Specialist Post-Training, Memory-Conscious Distillation, and Networking Evaluation

The paper introduces RSOD, a framework for creating telecom‑specific language models that blend broad networking knowledge with specialized routing, broadband, and diagnostic capabilities. It generates over 212,000 synthetic training records from networking taxonomies, trains six frozen specialist LoRA teachers, and distills their behavior into a single deployable student adapter, Llama‑CalixCLM v1. The student achieves near‑teacher performance (within 0.05 micro‑score points) while maintaining a validated, auditable training‑to‑deployment path.

By Anil Srirangapatna Nagesh, Nitin Gupta, Anuj Taneja, Vivek Singh, Shane Eleniak, Theo Zhao, Moulidharan N, Shashank Keshetty
llmsfine-tuningefficiency
More like this →
arXiv Machine Learning
Sep 14

SCOPE-OPSD: Fisher-Conditioned Privileged Subspaces for On-Policy Self-Distillation

SCOPE-OPSD introduces a Fisher‑conditioned privileged subspace for on‑policy self‑distillation (OPSD) that projects the teacher‑student residual onto a frozen rank‑64 factor derived from residual covariance and language‑model‑head Fisher sensitivity. The method adds no extra rollouts or inference modules and, across multiple Qwen3 model checkpoints and trajectory lengths, consistently matches or surpasses pure OPSD and a matched random baseline, achieving significant gains in most settings. A cross‑fitted diagnostic shows a 4.40‑fold increase in captured privileged‑gap compared to the random orientation.

By Yunmeng Chen (Chongqing Ant Consumer Finance Co., Ltd), Kunyu Wang (Alibaba Cloud Computing Co., Ltd), Peihan Li (Chongqing Ant Consumer Finance Co., Ltd), Yi Wang (Chongqing Ant Consumer Finance Co., Ltd), Shuyin Xia (Chongqing University of Posts and Telecommunications), Yi Liu (Chongqing Ant Consumer Finance Co., Ltd), Xinyong Cheng (Alibaba Cloud Computing Co., Ltd), Dehui Wang (Alibaba Cloud Computing Co., Ltd), Xiangyong Zhai (Alibaba Cloud Computing Co., Ltd), Yanxing Liu (Chongqing Ant Consumer Finance Co., Ltd), Song Liu (Chongqing Ant Consumer Finance Co., Ltd)
efficiency
More like this →
arXiv Machine Learning
Oct 8

Composing What Each Teacher Learned: Multi-Teacher On-Policy Distillation through Teacher-Relative Shifts

The paper introduces Δ‑MOPD, a multi‑teacher on‑policy distillation method that transfers each teacher’s logit shift relative to its base policy, re‑anchored at the student’s frozen initialization. Δ‑MOPD is evaluated in two settings: common‑domain composition, where multiple teachers score a single student rollout, and routed‑domain distillation, where prompts are assigned to specialist teachers. Across experiments, Δ‑MOPD outperforms traditional endpoint supervision, especially when teacher signals are combined at a state, achieving higher accuracy and reducing performance gaps in phased routing scenarios.

By Hejian Sang, Zhengze Zhou, Shayan Mohajer Hamidi, Xiaomin Li, Rohit Jain, Alborz Geramifard
efficiencybenchmarks
More like this →