Q-Strata is proposed, a bi-level allocator that ranks within-block assignments with a cheap proxy and allocates across blocks with a model-level objective evaluated on the assembled quantized model, achieving lower WikiText2 perplexity than uniform-bitwidth GPTQ and the state-of-the-art MoE MPQ methods MxMoE and GEMQ in the low-bit regime.
This work proposes Neural Double Q-routing, which replaces destination-indexed tables with a shared state--action value network, and achieves the lowest mean completion time among all compared methods in the six 150- and 200-OHT settings, whereas Dijkstra remains best in the three 100-OHT settings.
Chen-Feng Gu, Qiu-Sheng Zhao, An-Bang Liu et al.· 0 citations
This survey provides a structured entry point to tensorized language models and clarifies when parameter savings can plausibly translate into memory efficiency, computational efficiency, or interpretability, and introduces a metric for the compression-realization gap between theoretical memory reduction and measured system-level speedup.
M. Tarasov, Salman Ahmadi-Asl, A. D. de Almeida et al.· 0 citations
Topo^2, a measurement framework that makes deep networks causally separable, measurable, and law-governed, and turns"memorization" from an ill-defined capacity into a measurable, separable, invertible topological layer.
Zhang Zhang, Ming Liu, Qing Wang· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
TopGQ, an accurate post-training GNN quantization framework, alleviating redundant quantization overhead, is presented, and dual-axis scale absorption is proposed, which enables activation quantization along both the outer and inner dimensions by merging one into the adjacency matrix.
Dain Kwon, Kanghyun Choi, Hyeyoon Lee et al.· 0 citations
Decay-Aware State Compression (DASC), which derives retention horizons from model weights, selects long-horizon state units, and packs them into a ragged state checkpoint layout to integrate efficiently with tensor-parallel inference engines.
Yanzhi Yu, Ping-Wei Sun, Jian-Chao Tan et al.· 0 citations
LFPG-RL is developed and evaluated, which integrates link-flow propagation guidance (LFPG) into proximal policy optimization (PPO), and results support the contention that the method is a more efficient and accurate online OD demand calibration method compared to existing ones.
Tail-Replay is presented, a prefix caching mechanism that enables unconstrained token-level prefix reuse in hybrid large language models and is evaluated on three Gated DeltaNet-based hybrid models using the LongBench and RULER benchmarks.
Yi-Rui Liu, Ruoling Qi, Xuan'er Wu et al.· 0 citations
This work discovers that certain attention heads exhibit sequential consistency in their attention patterns, which can be persistently identified using a coefficient-of-variation-based algorithm, and proposes CateKV, a hybrid KV cache method that retains only critical token information for consistent heads, thereby reducing KV cache size and computational overhead.
Hao-Yun Jiang, Hao-Lin Li, Jian-Wei Zhang et al.· International Conference on...· 2 citations
BCPPO (Bachelier-Inspired Constrained Proximal Policy Optimization), a proximal policy optimization (PPO) method, supports a practical balance among reward, caution around cost predictions that vary across trained critics, and policy-only deployment.
This work introduces a new architectural component that embeds structured inductive bias into deep learning: an attention mechanism operating over tensor-product representations (TPRs) that outperforms existing architectural components in combinatorial generalization.
Graph4BiLO is introduced, a graph neural network (GNN) approach for learning bilevel value functions from variable--constraint graph representations that obtains objective values comparable to Neur2BiLO across all tested sizes while avoiding size-specific neural networks.
Jessica D. Elrefaei, Kaixun Hua, Seungbae Kim et al.· 0 citations
