Smoothed decentralized Bregman proximal-gradient (SDBPG) methods combine relative-smoothness geometry with edge-dual consensus. We extend SDBPG to node-specific convex nonsmooth regularizers and unbiased stochastic local gradients while retaining its Bregman primal-dual smoothing mechanism. The analysis isolates tempor...
Sheng-Ze Xu, Chen-Han Jin, Qing-Song Wang et al.· 0 citations
Softmax routing in mixture-of-experts models can concentrate probability on a subset of experts, suppressing access to the others. We study how this concentration forms, why it persists, and how suppressed probabilities recover. For fixed experts, a stationary-loss barrier identifies the face supporting the optimal rou...
We introduce NAGO, a noise-aware generative operator for efficient sampling from invariant measures of stochastic partial differential equations (SPDEs) in function space. Our central idea is to treat the source measure as a design variable rather than a fixed Gaussian default. We establish admissibility conditions for...
Guan-Yu Chen, Sheng-Ze Xu, Peng-Wei Liu et al.· 0 citations
Automated generation of LLM harnesses promises to improve inference through task specialization. Yet additional answer coverage can arise from repeated execution of the same program, making specialization difficult to identify. We introduce a controlled evaluation that separates answer coverage, repeatable task advanta...
Zi-Yang Xu, Haitian Zhong, Hao Zhou et al.· 1 citation
TIDE (Token-level Independent Deficit-Excess correction), which applies bounded Hellinger shaping to suppress the most severe sampled excesses and an analytic teacher top-$K$ injection to restore deficient probability mass without requiring deficit tokens to be sampled, is proposed.
Zi-Chao Yu, Chengzhi Yu, Sheng-Ze Xu et al.· 4 citations
Ada-BPSG is introduced, a line-search-free BPSG method that couples the SAGA gradient table with a stabilized Barzilai--Borwein (BB) candidate and yields a direct analytical chain from relative smoothness and component-wise variance control to convergence in finite-dimensional normed spaces.
Chen-Han Jin, Sheng-Ze Xu, Bing-Hui Xie et al.· 0 citations
This work proposes Causal Localization and Retain-Aware Projection (CLRP), a lightweight training-free framework that uses activation patching to identify the layer that causally mediates target-attribute disclosure, then applies a retain-aware projection that removes the target-attribute subspace while preserving same...
Junkai Lin, Junkai Chen, Siqi Hou et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.