2026-07-10
SLORR: Simple and Efficient In-Training Low-Rank Regularization
针对低秩正则化需SVD或修改架构的问题，提出无状态SLORR方法，提升压缩效率。
Authors: David González-Martínez, Shiwei Liu | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08754v1
Abstract
Low-rank factorization is widely used to compress neural networks, but modern models are often not naturally amenable to aggressive factorization without significant accuracy loss. Existing training-time low-rank regularizers can improve compressibility, but they often require SVDs of large weight matrices, modify the model architecture (introducing additional trainable parameters), or rely on stateful cached quantities. To address these limitations, we introduce SLORR, a simple, stateless, and architecture-preserving framework for in-training low-rank regularization, instantiated with two main variants based on the Hoyer sparsity metric and the nuclear norm. SLORR directly regularizes the original weight matrices using GPU-friendly approximations for the forward and backward passes of the regularizers, for which we provide approximation guarantees. We first evaluate SLORR on ImageNet-1K across short-horizon continued training of ResNet-50, ViT-B/16, and ViT-L/16, and pretraining of ResNet-18, where SLORR induces compressibility while introducing less than 8% training overhead. We further evaluate SLORR-Hoyer in LLM pretraining at 135M and 560M scales: SLORR-trained compressed models preserve performance substantially better than unregularized models while adding less than 1% average training overhead.The Illusion of Equivalency: Statistical Characterization of Quantization Effects in LLMs
指出量化评估仅用准确率不足，提出一致性指标揭示量化行为变化。
Authors: Baha Rababah, Cuneyt Gurcan Akcora, Carson K. Leung | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08734v1
Abstract
Post-training quantization is widely used to deploy large language models in resource-constrained settings, yet its evaluation relies almost exclusively on accuracy and perplexity. We show that these metrics fail to capture behavioral changes induced by quantization. We introduce correctness agreement, a decision-level metric that measures overlap in correct predictions between a base model and its quantized variants, independent of absolute accuracy. Across multiple models and quantization schemes from 8-bit to 2-bit, we find that behavioral divergence emerges under moderate quantization even when task performance appears preserved. To explain this effect, we analyze quantization as a structural operator on attention weights and quantify layer-wise distortions using statistical and distributional measures. Our results reveal non-linear breakpoints at low bit-widths and show that query and key projections are consistently more sensitive than value and output projections. These findings expose an illusion of equivalence between base and quantized models and motivate behavioral evaluation beyond conventional performance metrics.Super Weights in LLMs and the Failure of Selective Training
质疑超级权重普遍重要性，证明选择性训练无效，精度降至随机猜测。
Authors: Shreyas Subramanian, Adewale Akinfaderin, Akarsha Sehwag | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08733v1
Abstract
Recent work identified Super Weights, individual parameters whose removal degrades model performance by orders of magnitude. We show that this degradation due to pruning Super Weights does not universally apply to all LLMs. Furthermore, if these parameters are so important, Super Weight-aware training should be effective. We show the opposite. Training Super Weights in isolation (100 to 8,192 parameters) drops accuracy to random-guessing levels on both OLMo-1B and OLMo-7B, and expanding to local neighborhoods of up to 36K parameters provides no improvement. The failure is specific to Super Weight coordinates: training an equal number of randomly chosen positions in the same down_proj layers instead improves over the baseline, so the collapse comes from targeting Super Weights, not from sparsity itself. Vanilla LoRA, updating every position in attention weight matrices through low-rank structure, succeeds with only 0.16% of parameters, and applying the same low-rank update to down_proj succeeds as well. A 10-seed ablation confirms that constraining LoRA updates at positions corresponding to Super Weight coordinates yields statistically indistinguishable results. These findings establish that parameter importance does not imply parameter trainability in isolation, and that effective fine-tuning relies on structured decompositions over entire layers rather than targeting individually important weights.A Practical Investigation of Training-free Relaxed Speculative Decoding
研究无训练松弛推测解码，探索放松无损保证对速度与能力的影响。
Authors: Guoxuan Xia, Luka Ribar, Paul Balanca | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08690v1
Abstract
Speculative decoding accelerates sampling from an autoregressive LLM by using a faster auxiliary model to draft tokens which are then verified in parallel by the LLM. Standard speculative decoding is lossless: its rejection and resampling steps exactly preserve the LLM's sampling distribution. Recent work argues that relaxing this strict guarantee can yield further speed-ups, controlled capability-speed trade-offs, or even capability gains. We practically investigate training-free relaxed speculative decoding techniques, unify existing approaches within a shared framework, benchmark them on contemporary settings, and distil takeaways and empirical findings for practitioners. Important takeaways include: relaxation can require considerable capability evaluation unlike lossless speculative decoding, and many relaxed approaches rely on a drafter that is a good language model, making them unsuited for lightweight dedicated multi-token-prediction drafters.BiSCo-LLM: Lookup-Free Binary Spherical Coding for Extreme Low-Bit Large Language Model Compression
针对极低比特压缩，提出无查找二进制球面编码，平衡表示能力与效率。
Authors: Yuantian Shao, Peisong Wang, Zhilei Liu, Chuangyi Li, Yuanteng Chen, Pengcheng Xie, Yiwu Yao, Zhihui Wei, Jian Cheng | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08643v1
Abstract
Large language models (LLMs) are increasingly constrained by memory capacity, weight bandwidth, and checkpoint storage during deployment. Existing low-bit compression methods mainly follow two directions. Scalar or group-wise quantization is simple and compatible with efficient low-precision kernels, but its representation capacity becomes limited when the target budget approaches 2 bits per weight. Vector-quantized weight compression provides a richer block-level representation, but usually introduces explicit codebooks, index lookup, and additional storage accounting. This paper presents BiSCo-LLM, a codebook-free binary spherical coding framework for extreme low-bit LLM weight compression. The core pipeline is built on three components. First, local weight chunks are mapped onto a unit hypersphere and binarized into compact spherical codes, so that the main payload is a bit-packed sign stream rather than explicit VQ centroids. Second, a residual BSQ stage encodes the reconstruction error left by the base spherical codec, providing an explicit rate-distortion path without stored codebooks. Third, category-wise recovery distillation is performed after replacing each Transformer module category, reducing the mismatch between local weight reconstruction and assembled model behavior. A small 8-bit protected-channel path is used as an auxiliary stabilization mechanism for sensitive channels and is counted separately from the BSQ payload. The reported storage budget includes binary codes, neural decoders, protected-channel payloads, LoRA adapters, and metadata.It Takes a MAESTRO To Prune Bad Experts
针对MoE模型专家剪枝，提出MAESTRO方法，考虑路由依赖性提升效率。
Authors: Palaash Goel, Ayush Maheshwari, Tanmoy Chakraborty | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08601v1
Abstract
Sparsely-activated Mixture-of-Experts (MoE) language models achieve remarkable inference efficiency by activating only a small fraction of parameters per token, yet their full expert banks reside in memory at all times, creating a prohibitive deployment bottleneck. Existing structured pruning methods, largely designed for dense transformers, assess expert importance using locally derived heuristics that are blind to the interdependent nature of MoE routing. We introduce MAESTRO (Markov-chain Approximated Expert Sparsification via Transition-based ROuting), a structured pruning framework designed for MoE architectures that models autoregressive expert activation trajectories as Ergodic Markov chains whose stationary distributions encode cross-layer dependencies, yielding a globally aware importance heuristic. Evaluated across five diverse domains including Safety, Bias, and Ethics, MAESTRO outperforms state-of-the-art baselines by up to 10.61% in average performance retention under a strict 50% compression regime, while exhibiting substantially lower cross-task variance, indicating that global, routing-congruent pruning produces models that generalize more consistently across heterogeneous tasks.SMetric: Rethink LLM Scheduling for Serving Agents with Balanced Session-centric Scheduling
针对智能体服务调度，提出会话感知调度，优化吞吐量与KV复用。
Authors: Jiahao Wang, Kaizhan Lin, Kaixi Zhang, Jinbo Han, Xingda Wei, Sijie Shen, Chenguang Fang, Wenyuan Yu, Rong Chen, Haibo Chen | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08565v1
Abstract
LLM scheduling is critical to serving, yet it remains unclear how well existing designs fit agentic serving--with LLM requests issued by agents instead of humans. This shifts the workload in two ways: (1) agents act only on complete responses, making the cluster's tokens per second (TPS) the primary goal and relaxing--not eliminating--per-token latency requirements; and (2) requests share much of their KV\$-reuse exceeds 80% of request tokens in a production trace from BAILIAN, versus 54-62% in chat. This paper first contributes a systematic study of request scheduling for agents on two real-world traces. We find that to increase KV\$ reuse, existing schedulers overly prioritize routing requests to instances caching their KV\$, overloading a few while leaving the rest idle, capping TPS. We thus present two key insights: (1) load balance need not sacrifice all KV\$ reuse, thanks to the global-tier KV\$ store and (2) by utilizing the workload's intra-session locality, balancing a small fraction of requests--the first request in each agent session--suffices to balance the cluster without sacrificing most KV\$ reuse on local instances. SMETRIC realizes these insights with balanced session-centric scheduling: it routes each session's first request purely for load balance and its follow-up requests in a cache-aware manner, preserving load balance and local reuse while keeping demand on the global tier low. Using the session turn information as the scheduling metric is deliberate: it is derived efficiently and accurately from the user inputs alone, so the scheduler stays clean and stateless. SMETRIC improves cluster TPS by 10-16% under prefill-decode colocation with a global store and prefill TPS by 2-34% under disaggregation over state-of-the-art schedulers, also with a better per-token latency.Hidden Decoding at Scale: Latent Computation Scaling for Large Language Models
研究固定Transformer骨干通过增加每token计算提升性能，解决扩展瓶颈。
Authors: Aiwei Liu, Cheng Shi, Chuhan Wu, Ci Lei, Di Lu, Donald He, Fan Zhang, Fanhao Kong, Feifei Zhang, Guan Wang, Haicheng Wang, Haoyu Liu, Houjin Yu, Jiachen Ding, Jiayi Feng, Jie Zhou, Jijun Chi, Jindi Shi, Jing Lei, Junjie Zhang, Laiyi Li, Le Tian, Linhao Zhang, Miao Fan, Sijun Zhang, Wei Jia, Weiwei Shi, Wenhan Li, Wentao Zhao, Wenteng Liang, Xiao Zhou, Xiaojin Zhou, Xihuai Wang, Xinyu Gao, Xuanliang Wang, Xuyang Ao, Yang Yu, Yangxiu You, Yinuo Zhao, Yufei Kuang, Yufei Wang, Yuan Liu, Yuan Liu, Yuwen Chen, Zhencong Tian, Zhongyin Zhao, Zilin Yu, Zitao Wang | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08186v1
Abstract
Scaling Large Language Models (LLMs) has been driven mainly by enlarging the Transformer backbone, but for an already-strong model this requires another round of costly pretraining. We study whether an existing backbone can keep improving by allocating more computation to each token while leaving the Transformer backbone fixed. Depth-recurrent (looped) Transformers pursue this goal but are hard to scale, because looped computation does not fit naturally with the pipeline parallelism used to train the largest models. We add computation along the sequence-length dimension, where the extra computation is simply a longer input and stays compatible with standard large-model training. We propose Hidden Decoding, a sequence-length scaling method applied during continued pretraining (CPT). It expands each token into n streams with independent embedding tables and keeps the intermediate streams' key-value cache as context, so each token performs more internal computation without adding or widening Transformer layers. To keep this affordable at scale, we introduce Stream-Factorized Attention, in which most layers attend only within each stream and only a few layers mix across streams, reducing the attention cost from quadratic to roughly linear in n. Experiments support two scaling results. At frontier scale, we train WeLM-HD4-80B and WeLM-HD4-617B at n=4 and improve their matched non-HD baselines, making Hidden Decoding the first demonstrated sequence-length scaling method at the 100B+ MoE scale. Across expansion factors, the gains grow as n increases, showing that sequence-length expansion is a practical fixed-backbone scaling path for frontier-scale LLMs.SQuaD-SQL: Efficient Text-to-SQL with Small Language Models via LLM-Guided Knowledge Distillation
针对文本到SQL任务，提出知识蒸馏方法，使小模型接近大模型性能。
Authors: Wangyu Wu, Xiaojian Lin, Rong Fu, Zaiyang Yu, Xuhang Chen, Wenjun Yu, Zhenhong Chen | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08161v1
Abstract
Text-to-SQL is a fundamental task in natural language processing that enables users to interact with structured databases using natural language. While large language models (LLMs) have demonstrated remarkable performance on this task, their substantial computational requirements hinder deployment in resource-constrained settings. In this paper, we introduce SQuaD-SQL (Small-Qualified and Distilled for SQL), a novel approach that empowers small language models (SLMs) to approach the performance of LLMs on the Text-to-SQL task while significantly improving efficiency through knowledge distillation and synthetic data generation. Our method comprises three key components: (1) LLM-based synthetic data generation, where structured knowledge is extracted from LLMs via carefully designed prompting strategies; (2) parameter-efficient fine-tuning, enabling full model training on a single consumer-grade GPU; and (3) domain-adaptive fine-tuning, where domain-specific synthetic data further enhances performance in targeted domains. Experiments on the WikiSQL dataset demonstrate that SQuaD-SQL achieves an execution accuracy of 86.9% on the test set, approaching the performance of LLMs while offering faster inference and lower memory usage. These results suggest that, with proper training strategies, SLMs can serve as practical and efficient alternatives for Text-to-SQL applications in resource-limited environments.Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization
综述系统感知KV缓存优化，从执行、存储、调度三维度组织现有工作。
Authors: Jiantong Jiang, Peiyu Yang, Rui Zhang, Feng Liu | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08057v1
Abstract
Despite the rapid advancements of large language models (LLMs), LLM serving systems remain memory-intensive and costly. The key-value (KV) cache, which stores KV tensors during autoregressive decoding, is crucial for enabling low-latency, high-throughput LLM inference serving. In this survey, we focus on system-aware KV infrastructure for serving LLMs (abbreviated as sKis). We revisit recent work from a system behavior perspective, organizing existing efforts into three dimensions: execution and scheduling (temporal), placement and migration (spatial), and representation and retention (structural). Furthermore, we analyze cross-behavior co-design affinity and behavior-objective links, highlighting future opportunities. Our work systematizes a rapidly evolving area, providing a foundation for understanding and innovating KV cache designs in modern LLM serving infrastructure.What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents
从率失真视角统一LLM与智能体的记忆压缩研究，提出综合框架。
Authors: Ashwin Gerard Colaco, Nada Lahjouji | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08032v1
Abstract
Large language models, and the agents built on them, spend an ever-growing share of their compute and memory on remembering: caching attention keys and values, carrying long prompts, maintaining recurrent state, and storing what happened in previous turns and sessions. Because none of this memory is free, four largely separate research communities have each learned to compact it. They evict or quantize the KV cache, prune or distill prompts, bound architectural state, and consolidate agent memory. We argue that these are instances of one problem: a rate--distortion decision about what context-derived information to retain versus discard, at what fidelity, under a resource budget, so as to preserve downstream task utility. We make this lens precise with a single compaction objective and a layer-agnostic lower bound, use it to build a seven-axis taxonomy that classifies methods from across the stack uniformly, and use it to transfer mechanisms between layers that have never been connected, from serving-stack KV management to agent long-term memory. Two patterns hold across the survey. At every layer the signal that decides what to keep is attention magnitude or recency, and it fails in the same way everywhere, by discarding, before the query is known and with no way to undo it, information the query later needs. And while compression is measured carefully on single-turn long context, the repeated compaction that agents actually perform is almost never measured, and no benchmark holds one budget axis across all the layers at once. We turn both observations into a benchmark proposal, a small reference experiment, and a set of compaction-aware design principles, and we map the open problems.Rethinking Small VLM Quantization: From Component-Wise Analysis to Hardware-Aware Edge Deployment
系统评估小视觉语言模型量化，提出硬件感知部署框架优化边缘设备。
Authors: Hyeju Shin, Chorwon Kim, Ryangsoo Kim, Hark Yoo, Jaein Kim | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08029v1
Abstract
The emergence of vision language models with fewer than 3 billion parameters has accelerated the implementation of on-device multimodal intelligence. However, a detailed understanding of component-wise quantization remains a bottleneck for optimal deployment. This paper presents a systematic evaluation framework for empirically validating five hypotheses across six quantization configurations on the Jetson Orin NX and AGX. By separating the vision encoder, projector, and large language model backbone yields the following results: (1) Quantization sensitivity is governed by the structural paradigm (MoE vs. dense) rather than scale alone, with MoE backbones mitigating INT4 noise where dense backbones degrade; (2) SigLIP encoders incur disproportionate INT8 latency on Jetson Ampere--a deployment-specific encoder-kernel-hardware interaction, not a SigLIP flaw; (3) Although INT4 quantization of LLMs greatly reduces VRAM consumption, it also causes slower token generation due to dequantization overhead; (4) Composite quantization errors are largely additive, except along the modality-alignment path, which is architecture-dependent; (5) The intelligence-per-joule profile varies significantly across platforms owing to memory bandwidth constraints.Structured Pruning of Large Language Models via Power Transformation and Sign-Preserving Score Aggregation with Adaptive Feature Retention
改进结构化剪枝方法，解决异构分数分布不匹配与符号信息丢失问题。
Authors: Ryota Kobayashi, Tsubasa Hirakawa, Takayoshi Yamashita, Hironobu Fujiyoshi, Yasunori Ishii, Tomoyuki Okuno, Kazuki Kozuka | Date: 2026-07-09
Link: http://arxiv.org/abs/2607.08027v1
Abstract
This paper proposes an improved structured pruning method for large language models (LLMs) that addresses key challenges in adapting Adaptive Feature Retention (AFR), an unstructured pruning technique, to structured pruning. When applying AFR to structured pruning, three major problems arise: distribution mismatch between heterogeneous pruning scores, loss of sign information indicating optimization direction consistency, and influence of outliers. To address these issues, we propose a unified approach combining power transformation for nonlinear distribution alignment, sign-preserving score aggregation, and percentile-based outlier removal. Experiments on Llama-3-8B, Vicuna-v1.5-13B, and LLaVA-v1.5-13B demonstrate that our method maintains accuracy comparable to unstructured pruning while achieving practical inference speedup through structured pruning.KronQ: LLM Quantization via Kronecker-Factored Hessian
提出KronQ量化框架，引入梯度协方差改进二阶后训练量化目标。
Authors: Donghyun Lee, Yuhang Li, Ruokai Yin, Priyadarshini Panda | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07964v1
Abstract
Post-training quantization (PTQ) is a widely adopted technique for compressing large language models (LLMs) without retraining. Existing second-order PTQ methods, including GPTQ, construct quantization objectives exclusively from input activation statistics, effectively assuming that all output channels contribute equally to the layer-wise reconstruction objective. We propose KronQ, a PTQ framework that challenges this assumption by introducing the gradient covariance into the quantization pipeline. Under the Kronecker-factored Hessian approximation, the quantization loss depends jointly on both the activation and gradient covariances, and KronQ exploits this at two complementary levels. (1) KronQ introduces bidirectional incoherence processing, extending the existing input-side random rotation to the output dimension using the gradient covariance, reducing weight magnitude variance across both input and output dimensions. (2) KronQ derives a new sensitivity metric for inter-layer mixed-precision allocation, driven by the gradient and activation Hessian traces. Notably, in the case of 2-bit weight-only quantization on LLaMA-3-70B, while GPTQ and GPTAQ diverge or produce degenerate quantizations (>2000 perplexity on WikiText-2), KronQ achieves 7.93 perplexity.Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing
比较线性注意力机制，分析表达能力与内存衰减权衡，提出跨层路由。
Authors: Tommaso Cerruti, Tim Rieder, George Rowlands, Lingfeng Jin, Imanol Schlag | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07953v1
Abstract
Self-attention lets each token retrieve information from the full context, but its quadratic cost in sequence length limits training and inference at long context. This paper presents a comparative study of softmax attention and four recent recurrent linear-attention architectures: DeltaNet, Gated DeltaNet, Kimi Delta Attention, and Gated DeltaNet-2. We express these mechanisms in a common recurrent-memory notation, making explicit how they differ in expressivity, memory decay, erase and write control, training throughput, and implementation complexity. Our experiments center on 350M-parameter models trained for 15B tokens, and include optimizer and learning-rate comparisons, hybrid-versus-pure stack comparisons, sequence-length runtime measurements, larger DeltaNet runs at 1.3B and 3B parameters, and a small set of downstream evaluations. The reported speed results measure training throughput and iteration time; we do not provide an empirical inference-speed benchmark. Within the reported 350M-parameter, 15B-token sweep, Kimi Delta Attention with Muon reaches the lowest final validation loss, a pure Gated DeltaNet stack trained with AdamW has the highest normalized training throughput, hybrid stacks generally improve loss at a throughput cost, and Muon consistently lowers final validation loss relative to AdamW in the matched architecture settings we evaluate. We introduce and evaluate lightweight cross-layer routing mechanisms for DeltaNet-style memories. The most natural DeltaNet-inspired formulation, forwarding a lower layer's delta-rule write error into the next layer's value target, does not improve over matched baselines. Routing into the aligned hidden stream and forwarding the write value instead yields a modest improvement in the matched runs we report: Cross-Layer Value Routing (CLVR) lowers final validation loss for both DeltaNet and Gated DeltaNet.CTA-Pipelining: A Latency-Oriented Spatial Scaling Method for Multi-GPU Systems
针对多GPU系统延迟约束，提出CTA流水线方法，优化空间扩展。
Authors: Tingkai Liu, Muralidhar Andoorveedu, Sanjoy Das, Sanjay Patel, Volodymyr Kindratenko | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07862v1
Abstract
The evolution of compute infrastructure has transformed multi-GPU systems into tightly integrated shared-memory structures. However, current software still mostly treats these coherent interconnects simply as high-speed networks. Simultaneously, the demand for serving Large Language Models under latency constraints has shifted GPU workload optimization from being throughput-driven to latency-bound, necessitating latency-oriented scaling methods beyond Tensor Parallelism (TP). Thus, we introduce CTA-pipelining, an execution paradigm designed to exploit shared-memory multi-GPU systems. As a latency-oriented spatial scaling technique, CTA-pipelining leverages dependencies at the Cooperative Thread Array level, enabling concurrent execution of dependent kernels across GPUs. We demonstrate its capability using CUTLASS, cuBLAS, and NCCL libraries on 8-GPU H200 and B200 systems. Results show on 2-layer GEMM, representing the MLP operation, CTA-pipelining reduces latency by up to 31.8% compared to micro-batching, and 29.6% compared to TP. It can also be combined with TP as an orthogonal scaling dimension to further push the latency boundary.Co-LMLM: Continuous-Query Limited Memory Language Models
提出连续查询有限记忆语言模型，支持连续键值对知识检索。
Authors: Yair Feldman, Linxi Zhao, Nathan Godey, Dongyoung Go, Yilun Hua, Kilian Q. Weinberger, Jennifer J. Sun, Yoav Artzi | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07707v1
Abstract
Limited memory language models (LMLMs) externalize factual knowledge during pretraining to a knowledge base (KB), rather than memorizing it in their weights. During generation, the model then fetches knowledge from the KB as needed. This recently introduced paradigm provides multiple advantages, including knowledge control capabilities that remain beyond conventional LLMs. We propose continuous-query LMLM (CO-LMLM), where the KB pairs continuous keys with textual knowledge values, a significant departure from prior reliance on relational KB and queries. CO-LMLM generates flexible vector queries at minimal cost, while still integrating human-readable and attributable retrieved knowledge into its generation. We pair this design with an annotation pipeline that tags free-form factual spans in arbitrary text, removing prior work's restriction to Wikipedia. Across pretraining on Wikipedia and FineWeb-Edu and at multiple model scales, CO-LMLM outperforms prior LMLMs and vanilla LLMs in both perplexity and factual precision. At 360M scale, this includes lower perplexity than models pretrained on 40x more data, and SimpleQA-verified performance that is in line with gpt-4o-mini and higher than Claude Sonnet 4.5.The Key to Going Linear: Analysis-Driven Transformer Linearization
分析Transformer线性化，指出softmax依赖秩1投影，改进状态更新设计。
Authors: Anna Kuzina, Paul N. Whatmough, Babak Ehteshami Bejnordi | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07706v1
Abstract
The quadratic cost of causal self-attention severely bottlenecks long-context transformer inference. While numerous post hoc linearization pipelines exist, it is difficult to identify which components preserve model quality. This work isolates the effect of state update design in a strict frozen-backbone regime. We show that softmax relies on key-dependent, rank-1 orthogonal projections, elucidating why delta-style networks outperform purely gated accumulation. We identify a potential source of approximation errors and introduce structural interventions, specifically sink tokens, short convolutions, and fixed-budget cache routing, which reduces the remaining gap. We scale this linearization approach across LLaMA and Qwen models up to 32B parameters, outperforming prior post hoc baselines on MMLU and matching the long-context retrieval of complex adaptive-caching frameworks.Future Confidence Distillation in Large Language Models
研究LLM置信度估计的时间演化，提出未来置信度蒸馏方法。
Authors: Sahil Kale | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07626v1
Abstract
Reliable confidence estimation is essential for deploying large language models (LLMs) in confidence-aware systems, where downstream decisions such as retrieval, tool use, and adaptive computation depend on accurately estimating answer reliability. Existing approaches, however, largely treat confidence as a property of completed responses, overlooking how confidence-related information evolves throughout the answering process. In this work, we investigate confidence from a temporal perspective by comparing pre-solution Feeling-of-Knowing (FOK) and post-solution Judgement-of-Learning (JOL) confidence estimates across frontier and open-source LLMs. We show that post-solution confidence is consistently better calibrated and more discriminative than pre-solution confidence, while linear probes trained on hidden representations recover substantially richer confidence-related information than models explicitly verbalise. Building on this observation, we introduce future confidence distillation, which trains predictors operating on pre-solution hidden representations using teacher confidence estimates produced by post-solution correctness probes. Despite requiring only pre-solution representations for inference, distilled predictors recover much of the calibration improvement achieved by post-solution confidence, remain highly sample efficient, and transfer across datasets within the same domain. Together, our findings demonstrate that confidence-related information evolves throughout the answering process and can be anticipated before answer generation is complete, enabling significantly more reliable yet low-cost confidence estimation.PALS: Percentile-Aware Layerwise Sparsity for LLM Pruning
提出百分位感知层稀疏剪枝，自适应调整每层稀疏率提升性能。
Authors: Yazdan Jamshidi, Alexey Shvets | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07557v1
Abstract
One-shot pruning methods like Wanda and SparseGPT apply the same sparsity ratio to every layer of a transformer, ignoring known variation in layer importance. We propose PALS (Percentile-Aware Layerwise Sparsity), which adjusts per-layer sparsity based on the 99th percentile of activation magnitudes, bounded to $\pm 5\%$ around the target ratio. On LLaMA-2-7B at 50\% sparsity, PALS achieves 10.96 WikiText-2 perplexity versus 12.92 for uniform Wanda (mean over 9 runs, $p < 0.001$). The benefit is architecture-dependent: LLaMA-3-8B shows marginal gains and Mistral-7B shows none. We also find that gradient-based allocation -- the seemingly more principled approach -- produces results worse than random, suggesting that gradient magnitude does not predict the impact of discrete weight removal. PALS adds negligible cost to the pruning pipeline and requires no fine-tuning.GIFT: Geometry-Informed Low-precision Gradient Communication for LLM Pretraining
针对梯度通信瓶颈，提出几何感知低精度梯度缩放方法。
Authors: Jieying Wang, Shuyuan Fan, Mingkai Zheng, Zhao Zhang | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07494v1
Abstract
Gradient communication is a primary scaling bottleneck in large language model (LLM) pretraining. Communicating gradients in low-precision formats, such as FP8 and NVFP4, can significantly reduce the communication volume. Existing methods quantize gradients via linear or nonlinear mappings in Euclidean space, often degrading model performance because highly anisotropic gradients incur direction-dependent distortion. We present GIFT, a geometry-informed gradient scaling method that performs low-precision communication in geometry-aware coordinates. By transforming gradients into a near-isotropic space before quantization, GIFT makes low-precision representations substantially more faithful to their high-precision counterparts. GIFT only changes the coordinate system used for low-precision gradient communication and does not change the optimizer, training recipe, communication collective, or low-precision format. We also develop a simplified geometry-aware transformation algorithm with low-rank approximation and selective application to balance the computation overhead and communication reduction. We examine the empirical convergence of GIFT using Llama-300M and Llama-600M models. Our results show that GIFT reduces the end-to-end pretraining time of Llama-600M by 7.6% on 64 NVIDIA GH200 Superchips, while improving the downstream task preservation profile over direct Euclidean FP8 communication under the same optimizer and communication path.FourierQK: Spectral Preprocessing of Query-Key Projections Improves Transformer Attention
通过频谱预处理查询键投影，显著提升字符级语言建模注意力性能。
Authors: Athanasios Zeris | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07478v1
Abstract
FFT-based spectral preprocessing of learned query-key (Q/K) projections substantially improves transformer attention on character-level language modelling. On TinyShakespeare: a fixed random spectral filter achieves val=1.031 (Delta=+0.443); a single learned frequency at paragraph scale achieves val=0.608 (Delta=+0.867); and four learned frequencies spanning paragraph to word scale achieve val=0.309 (Delta=+1.166), a 79% reduction over standard dot-product attention. The single-frequency result is confirmed across three random seeds (mean val=0.236, std=0.019). The four frequencies converge to a near-geometric multi-scale ordering (49, 27, 10, 6 tokens/cycle) corresponding to paragraph, sub-paragraph, phrase, and word scales. The gain is specific to spectral preprocessing: random orthogonal and non-orthogonal projections of Q/K produce no measurable improvement, suggesting the benefit comes from global frequency-domain mixing rather than metric distortion. All results are verified by a shuffled-validation diagnostic against positional leakage. Causal filters (Gaussian, Mexican Hat, Morlet) do not improve over standard attention at character-level tokenisation: the bilateral FFT kernel is structurally non-causal, coupling every position to future tokens. This defines an architectural boundary between bilateral spectral attention (this paper) and genuinely causal spectral attention at word-scale tokenisation (companion paper MorletQK). This work is architecturally distinct from FNet (Lee-Thorp et al., 2021), which replaces attention with Fourier mixing of token embeddings. Here, spectral preprocessing applies only to Q/K projections while the full attention score structure is preserved.DeLS-Spec: Decoupled Long-Short Contexts for Parallel Speculative Drafting
提出解耦长短期上下文并行推测草稿，无需训练草稿模型。
Authors: Hong-Kai Zheng, Piji Li | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07409v1
Abstract
Speculative decoding accelerates LLM inference by drafting multiple tokens and verifying them in parallel. Block-parallel drafters such as DFlash further improve drafting efficiency by predicting an entire block in one pass, but their position-wise predictions lack explicit intra-block causal conditioning. Recent methods such as Domino and DSpark attempt to introduce such causality into block-parallel drafting, but they require training the draft model from scratch, which limits their flexibility and increases training cost. We propose DeLS-Spec, a decoupled long-short context speculative decoding method. DeLS-Spec treats the fixed DFlash model as a long-context expert and introduces a lightweight local head as a short-context expert. The local head can be trained independently with a standard next-token prediction objective, without joint training with the target model or the DFlash backbone, leading to extremely low training cost. At inference time, DeLS-Spec combines long-context and short-context logits, and the local head is not tied to a specific DFlash checkpoint, making the method more modular and flexible. Experiments on Qwen3 models show that DeLS-Spec consistently improves speedup and average acceptance length over DFlash across math, code, and dialogue benchmarks.TF-Engram: A Train-Free Engram with SSD-Backed Memory for Large Language Models
提出免训练记忆模块，使用SSD存储扩展LLM知识，避免语义冲突。
Authors: Yutang Ma, Kecheng Huang, Xikun Jiang, Zili Shao | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07388v1
Abstract
Large Language Models (LLMs) store factual knowledge and domain-specific patterns implicitly in dense Transformer parameters, making knowledge expansion costly through pretraining, fine-tuning, retrieval augmentation, or longer contexts. Engram-style memory offers a compact hidden-state injection pathway, but existing GPU-resident designs often rely on hash-based compression, causing unrelated phrases to collide in shared slots and weakening phrase-level semantic fidelity. We present TF-Engram, a train-free Engram system that constructs phrase-specific semantic memory offline from external corpora, stores large memory tables across a GPU--DRAM--SSD hierarchy, and uses Early-Exit Guided Predictive Prefetching to hide external-memory latency during autoregressive decoding. On Qwen3-0.6B, TF-Engram improves the average downstream score from 57.6 to 59.4, outperforming both the frozen backbone and a parameter-matched LoRA baseline. System evaluation shows that large TF-Engram tables can be built with moderate offline cost, SSD-backed storage substantially reduces GPU memory demand, and predictive prefetching recovers much of the throughput loss caused by external memory access. These results demonstrate that static phrase memory can be integrated into LLM inference as a scalable, train-free, and low-overhead system component.Sparse Delta Memory: Scaling the State of Linear RNNs through Sparsity
提出稀疏增量记忆，通过稀疏性扩展线性RNN状态，提升长上下文召回。
Authors: Loïc Cabannes, Pierre-Emmanuel Mazaré, Gergely Szilvasy, Matthijs Douze, Maria Lomeli, Ilze Amanda Auzina, Justin Carpentier, Gabriel Synnaeve, Hervé Jégou | Date: 2026-07-08
Link: http://arxiv.org/abs/2607.07386v1