Quantization, distillation, pruning and serving work aimed at the same accuracy for less memory, latency and money.
arXiv:2608.30908v1 Announce Type: new Abstract: Fine-tuning Low-bit models aims to adapt a quantized model while keeping the final deployed checkpoint in the same low-bit form. This setting is practi...
arXiv:2608.30262v1 Announce Type: new Abstract: Scientific simulations generate collections of physical fields with heterogeneous statistics and dependencies, yet learned compressors often encode tho...
arXiv:2608.30699v1 Announce Type: cross Abstract: Long-tailed distributions are prevalent in real-world semi-supervised learning (SSL), where pseudo-labels tend to favor majority classes, leading to...
Modern explicit-time generative models, such as Flow Matching [Lipman et al., 2023] and Rectified Flow [Liu et al., 2023], are typically derived top-down via Optimal Transport and the continuity equat...
Efficient evaluation changes the protocol used to support claims about model behavior, yet it is rarely tested whether those claims remain stable after the evaluation itself is made cheaper. We stress...
Scientific discovery often requires reasoning over competing hypotheses that are consistent with experimental observations. For mixed-variable and combinatorial hypothesis spaces, however, constructin...
Multimodal models often build on architectures designed for generative vision-language modeling, typically combining separately pretrained vision encoders with causal language models. Visual document...
Rapid mineral characterization is essential for applications ranging from mineral exploration to industrial ore processing. To this end, Hyperspectral Imaging (HSI) has emerged as a promising sensing...
Existing GNN quantization methods suffer from considerable quantization overhead, which severely limits their practical usage in real-world scenarios. To this end, we present TopGQ, an accurate post-t...
Long-context LLM applications such as document summarization and multi-turn agents require generation from prefixes spanning tens of thousands of tokens, making decoding latency a major bottleneck. Sp...
The paper introduces VISTA, a method that enhances on‑policy self‑distillation (OPSD) by adapting the teacher model toward the student’s distribution using outcome‑verified rollouts. VISTA keeps the standard OPSD student update but selectively adjusts the teacher only on the top‑k positions with the largest teacher‑student KL divergence, without adding new sampling or reward objectives. Experiments on AIME24, AIME25, and HMMT25 with Qwen3 models show that VISTA outperforms OPSD across all scales, improving Avg@12 by up to 2.1 points.
The paper introduces PASK (Parser-Aware Structural KV Persistence), a method that leverages parser transitions to inform key‑value (KV) persistence decisions in structured generation tasks. By aligning KV compression with task‑level structured risk, PASK sets protection floors based on error sensitivity and allocates remaining KV capacity using attention‑output distortion, producing a lightweight, structure‑conditioned lookup policy. In experiments on Qwen3‑4B, PASK achieves a 17.39‑point accuracy gain over the best compressed baseline, while delivering up to 2.2× higher throughput, 3.3× lower TPOT, and 0.53× the peak GPU memory of full KV.
The paper investigates the intrinsic dimension (ID) of large language model (LLM) representations as an indicator of linguistic complexity. By comparing ID across model layers for coordination vs. subordination, right‑branching vs. center‑embedding, and unambiguous vs. ambiguous attachment, the authors find consistent ID differences that align with established complexity contrasts. Experiments across six LLMs, including representational similarity and layer pruning analyses, confirm that more complex phenomena produce higher ID profiles, with peaks occurring at different layers for each contrast.
The paper introduces a method for allocating bit-widths to individual layers in Gemma-3-1B to maximize performance (latency reduction) while staying within a specified quality-degradation budget. Using a layer sensitivity profile from SA-PTQ and TensorRT-LLM, the authors evaluate 13 W8A8 variants on an RTX 5090, finding that FFN 5+5 with lm_head yields an 11.0% latency reduction with negligible quality loss. They also discuss trade-offs for Attention layers and propose further optimizations such as fused INT8 attention kernels and FP8 usage.
The paper introduces VICT, a method that leverages the internal structure of verifiable tasks to perform fine‑grained credit assignment for long‑horizon LLM agents. VICT exposes executable or evidence‑backed atoms from a task’s terminal verifier and traces them back to actions via dependency‑valid proof edges, redistributing advantage only along these edges. This approach improves performance on ALFWorld and WebShop compared to outcome‑only training and matches recent fine‑grained credit methods without requiring additional critics, labels, or inference‑time verifier access.
HyperVision introduces the first ground‑based hyperspectral pre‑trained backbone, addressing challenges of varying spectral configurations, limited annotations, and dataset diversity. It employs a channel‑adaptive dynamic embedding to unify heterogeneous inputs, a multi‑source pseudo‑labeling strategy combining SAM2 spatial cues with HyperFree spectral details, and cross‑modal knowledge distillation from a pre‑trained RGB vision model. Trained on 15k images from 26 datasets, HyperVision achieves significant improvements—up to 16.3% relative gain in hyperspectral semantic segmentation, 2.1% in object tracking AUC, and 35.5% reduction in salient object detection MAE—while requiring only head‑only adaptation.
The paper introduces Speculative Probing, a method that repurposes the speculative‑decoding module of large language models for real‑time classification tasks. By appending a trained soft prompt to the target sequence, the approach leverages the already‑cached KV store during inference, adding negligible overhead while achieving higher accuracy than traditional hidden‑state probes. Experiments on four classification tasks across multiple models show that these lightweight probes outperform zero‑shot GPT‑5.4‑mini and rival or surpass specialized 8B safety classifiers without running a full LLM.
The paper introduces a distance‑guided decoding framework that guarantees context‑free grammar compliance for large language models. By precomputing bounded pushdown summaries with reachability labels and upper‑bound distances to acceptance, the decoder performs horizon‑aware pruning and beam search during generation. Experiments on JSON, SQL, and Linear Temporal Logic demonstrate consistent syntactic validity and improved completion quality compared to existing baselines.
The article surveys recent neural‑network optimizers, noting that the field has moved beyond simple Adam variants to encompass matrix‑ and layer‑level designs, time‑policy horizons, and state representations that survive sharding and low‑precision computation. It categorizes optimizers along four axes—temporal estimation, update geometry, horizon management, and representation & systems—highlighting methods such as Muon, Shampoo, SOAP, and quantized states. The survey concludes that while matrix‑aware methods are a genuine advance, no single optimizer universally replaces AdamW, and performance depends on model scale, data‑to‑parameter ratio, batch size, schedule, partitioning, tuning budget, and target metric.
The study presents a real‑time musculoskeletal surrogate for children with cerebral palsy, built from OpenSim parameters, joint kinematics, and muscle capacities. Using leave‑one‑subject‑out validation on nine pediatric gait recordings, the surrogate reproduces musculotendon lengths with high accuracy (R² ≈ 0.92–0.95, nRMSE < 8%) and achieves sub‑millisecond inference times, well below the 100 ms interactive‑rehabilitation target. A Monte Carlo credibility pilot reveals that small variations in anthropometry and muscle capacity lead to overconfident 90 % prediction intervals, highlighting the need for improved force modeling and uncertainty quantification.