每天知道
你的领域
出了什么

Large language models (LLMs) increasingly serve as autonomous agents that invoke external tools. However, this capability introduces tool hallucination, selecting incorrect tools or generating invalid calls. Existing mitigation methods report substantial improvements, yet we identify a previously overlooked failure mode that we term Hallucination Escape. These methods reduce hallucination on the tool configuration they are tuned on but increase it on other configurations, canceling out the gain. We further investigate this phenomenon and find that hallucination rises sharply when a model's intrinsic tool-use tendencies conflict with the current tool configuration, and that existing methods reinforce rather than suppress these tendencies, which in turn contributes to hallucination escape. Building on these findings, we propose EscapeGuard, a training-free inference-time method that combines conflict-aware gating with configuration-derived attention enhancement to mitigate tool hallucination while preventing hallucination escape. Across six benchmarks on various models, EscapeGuard reduces tool-selection hallucination by 9.0 pp and suppresses hallucination escape, lowering the cross-configuration mean by 23.7 pp and achieving an 89.1% net improvement in paired-query evaluation. We hope this work can encourage evaluation beyond a single tool configuration and pave the way for more reliable tool-using LLM agents.

Inflammatory bowel disease (IBD) requires frequent imaging-based assessment, yet interpretation of modalities such as wireless capsule endoscopy (WCE) and intestinal ultrasound remains heavily dependent on specialist expertise. Vision-Language Models (VLMs) demonstrate significant potential in multimodal medical image analysis, but their clinical adoption is hindered by their insufficient domain-specific reasoning, susceptibility to hallucination, scarcity of high quality training data in fine-grained diagnostics and limited interpretability. We introduce ARISE (Adaptive Agentic Reasoning with Image-grounded Self-Evaluation), an autonomous planning framework that models few-shot medical image understanding as a sequential agentic workflow. ARISE structures agent execution into a transparent 5-stage workflow: hypothesis generation, image-grounded evidence summarization, evidence-conditioned refinement, symbolic verification, and final diagnosis. We apply ARISE to IBD assessment across two independent patient cohorts: wireless capsule endoscopy (WCE) images for Crohn's disease and B-mode ultrasound data for ulcerative colitis. ARISE consistently improves diagnostic performance over baseline VLMs while exposing where reasoning succeeds or fails, providing a more interpretable basis for clinical decision support and realistic deployment.

Agents built on large language models (LLMs) build and run software through tool calls. A call reaches its program through several hops, and any hop can change the call without notice. When the changed call fails, the agent retries a correct call, which costs users time and money. Benchmarks and failure analyses do not see the change, because they read the call and its result but not what a hop received. We define intent-execution correspondence (IEC) as the property that the executed action matches the action the emitted call denotes under the tool contract. Our protocol observes what each hop received without executing the call, and names the first hop that changed it by the receiver's own parser. IntAct then delivers the call in a form that this hop cannot alter, or refuses the call. We build IEC-Bench from the changes observed in real-world use, with chains of dependent calls under the execution paths of 4 widely-used harnesses. In 47,828 shell calls within production sessions, Claude Code's Bash tool changes 12.0% of the calls that carry code, escape sequences, or long text. For 80.7% of the calls whose backslashes are changed, the wrong action runs without any reported error. All 10 measured harnesses change a call. Trajectory-based judgment attributes 95.1% of the production failures to the LLM, although the path caused more than half of them. On IEC-Bench, the path raises the token cost per passed task 2.4 times (up to 12.3 times). A hop that changes a call also hides the changes after it, so 55.1% of the failures on one path appear only after its first hop is repaired. IntAct, deployed in a commercial product, recovers 79.2% of the failures with a changed call. Harnesses should therefore be designed and tested hop-by-hop to ensure a correct call executes as intended or is refused.

Recent advances in artificial intelligence-radio access network (AI-RAN) are placing large language model (LLM)-driven agents within the open RAN (O-RAN) control hierarchy. A promising deployment for this agentic AI-RAN co-locates LLM inference with physical-layer (PHY) communication processing over the same accelerated compute pool for infrastructure sharing, data locality, and low control latency. However, this co-location induces bidirectional control-compute coupling, as the agent competes with PHY processing for compute, while its reasoned gNB control actions may alter future PHY workload and hence the compute available to its next inference. To this end, this article proposes the control-compute contract, a coordination protocol linking AI-RAN workload governance with O-RAN control execution. Specifically, we first expose direct compute contention on an over-the-air (OTA) O-RAN testbed, where continuous LLM inference increases the PHY decoding time approximately eightfold. We then formulate the contract as three protocol rules with operator-requirement protection, and map it onto O-RAN and AI-RAN functions as a five-stage workflow. Afterward, an OTA-calibrated case study of simulated uplink gNB control indicates that the contract keeps the proportion of slots violating the PHY time budget below $0.9%$, against $20.7%$ under ungoverned co-location, while achieving faster gNB control than a fixed compute reservation benchmark. Finally, we identify several open issues and outlooks for its practical deployment.

Agentic large language model (LLM) systems are commonly implemented as an LLM in a loop with Planning, Memory, Tools, and Control Flow. This application-focused view connects agentic LLM research with deployable systems and leaves open how such systems should be evaluated beyond end-to-end task success. Building on this view, we define agentic cognitive depth as a trajectory-level profile across five operational criteria. The profile contains context sensitivity ($C$), temporal continuity ($T$), multimodal coordination ($M$), adaptive interaction ($A$), and metacognitive monitoring ($Mc$). The first four criteria measure how well Control Flow, Memory, Tools, and Planning are used across a trajectory. The fifth measures whether the system monitors and regulates the full run. For each criterion, we give operational proxies and a perturbation procedure, then connect the profile to the agent's world model. We provide the structure needed to extend benchmarks such as GAIA, SWE-bench, WebArena, and TRIP-Bench with per-criterion diagnostics. Symbolic verifiers, structured memory, planner coupling, and tool constraints provide practical ways to build and test these capacities.

Large language models (LLMs) have progressively evolved into the core of autonomous agents. Building on this progress, LLM-based multi-agent systems (MAS) coordinate multiple agents into a synergistic team to accomplish complex tasks that exceed the capabilities of individual agents. The effectiveness of such systems depends not only on the agents themselves, but also on how collaboration mechanisms are designed and organized. Note that real-world collaboration is typically partially observable, where each agent can only access partial information about the environment due to physical or privacy-related constraints. However, many existing multi-agent benchmarks assume global observability, and leave limited support for systematically evaluating collaboration mechanisms. To bridge this gap, we introduce MASBench, a multi-agent collaboration benchmark designed under partially observable constraints. It is organized into three progressive task categories: Reasoning, Scheduling, and Game. Through this structure, we progressively evaluate three representative collaboration mechanisms: Protocol, Memory, and Routing. MASBench further provides deterministic evaluation metrics, including performance score, communication cost, and cost effectiveness, to characterize both collaboration outcomes and communication overhead. Experiments across diverse LLM backbones and mechanism configurations offer empirical guidance for effective MAS design. Code is available at: https://github.com/BUPT-GAMMA/MASBench

Efficient long-video understanding with vision-language models (VLMs) is often framed as selecting informative frames or visual tokens at a fixed native resolution. We show that per-frame resolution can instead be traded for denser temporal coverage, while front-end decoding latency depends on the size of the candidate pool rather than the final token budget. An empirical study across multiple VLMs and long-video benchmarks yields three findings: dense low-resolution sampling outperforms sparse native-resolution sampling at matched token budgets; resolution-sensitive tasks benefit from selected high-resolution frames; and front-end decoding dominates wall time for hour-long videos. Motivated by these findings, we introduce LoHi, a training-free, single-pass framework that combines a dense low-resolution video stream with sparse high-resolution image frames through the VLM's native video and image pathways. LoHi-Anchor selects high-resolution frames using codec I-frame metadata, while LoHi-SemDiv uses query relevance and visual diversity over CLIP features. Across three long-video benchmarks, LoHi improves average accuracy by 10.6 percentage points over the native-resolution baseline at a matched token budget and by 5.2 percentage points over the strongest prior efficiency method. It also reduces front-end decoding latency by up to 7x on hour-long videos. Project page: https://sixundong.com/projects/lohi

Ultra-low-latency machine learning and data processing pipelines operating on the sub-microsecond level often contain static dataflow kernels that require fine-grained bitwidth control, arithmetic optimization, and fast hardware performance estimation. This works introduce Alkaid, a free and open source domain specific compiler that translates sub-microsecond latency dataflow kernels into platform agnostic Register Transfer Level (RTL) or High-Level Synthesis (HLS)-ready code in seconds. Alkaid targets ultra-low-latency ML pipelines, especially those addressed by existing ML-to-hardware flows such as hls4ml, while also supporting the surrounding pre-processing, post-processing, and control logic. At its core, Alkaid uses an arithmetic centric intermediate representation, Alkaid Low-level IR (ALIR), to preserve fixed-point semantics, heterogeneous bitwidths, and operation level structure for optimization and code generation. Alkaid further provides comprehensive hardware-aware optimization passes and an interpretable white-box analytical performance model for rapid design space exploration without synthesis in the loop. Across representative ML workloads, including MLPs, GNNs, and transformers, Alkaid achieves 20-30% lower LUT usage than the best prior flows for bit-exact equivalent designs, with up to 40% lower latency in some cases. Beyond typical neural networks, Alkaid achieves comparable performance to the state-of-the-art method for synthesizing boosted decision trees, while also enabling the implementation and integration of non-ML kernels such as sorting networks and histograms.

Transformer models such as BERT and Vision Transformer~(ViT) achieve strong performance via densely parameterized attention backbones. However, the least significant bits~(LSBs) of their 32-bit floating-point weights can be abused as covert channels to conceal malicious payloads, posing a serious threat to the AI model supply chain. We propose \GS (\GSabbr), a lightweight, post-training, zero-data sanitization method that completely replaces the declared mantissa-LSB channel with a Gray-code-guided low-transition sequence. Complete payload-independent overwrite, whether keyed or public, makes the sanitized target bits independent of the embedded payload and gives that declared channel zero capacity. Gray coding supplies overwrite structure, while a keyed per-tensor phase supplies pattern diversity. Benchmarked against seven post-training defenses on four Transformer model presets and two real-world malware payloads, \GSabbr maintains sub-$1%$ accuracy impact and achieves $49.96\pm0.66$ percentage-point Recovery Reduction (RR) under five implemented attacker variants. Because pre-defense recovery is effectively $100%$, RR near 50 percentage points corresponds to post-sanitization bit accuracy at binary chance. Its main empirical advantage is stable near-chance sanitization with substantially smaller weight-distribution shift than the evaluated near-chance baselines PatternMask (PM) and Post-Training Quantization (PTQ).

In worker-equipment proximity monitoring, video transformers are widely used for collision anticipation and have demonstrated strong performance. However, their accuracy comes with substantial computational demands, creating a tension with the need for low-latency inference on mobile robots and the pursuit of lower-carbon computation in construction. To address this, this study investigates where computation within an established video transformer is redundant and whether that redundancy can be removed without materially degrading predictive performance. Using VideoMAEv2-Base on the Nexar Collision Prediction dataset, we first examine how collision-relevant information evolves across network depth and then investigate two complementary forms of redundancy: structured capacity redundancy in multilayer perceptrons (MLPs) and spatiotemporal redundancy in the token stream. Linear probes show that interpretable motion cues, including flow magnitude, looming, and approach versus retreat, are most accessible at intermediate layers, whereas collision-label discrimination strengthens toward the final layer. Token redundancy is axis-specific: adjacent temporal-token similarity reaches 0.970 in later layers, while spatial similarity falls to 0.297, indicating substantially greater redundancy across time than across space. Exploiting this asymmetry, temporal token merging reduces backbone computation from 356.99 to 178.50 GFLOPs and latency from 12.69 to 7.21 ms per clip, a 1.76x speedup, while mean average precision changes only from 0.7478 to 0.7443. Importance-guided retention of 50% of MLP units preserves an AUC of 0.753, compared with 0.529 under matched random retention, and reveals that pruning alters score calibration before discriminative ranking collapses. These findings establish a new pathway for pursuing faster algorithms through targeted temporal token compression and neuron pruning.

Post-training quantization to the GGUF format's mixed-precision K-quants is commonly how open-weight language models reach consumer hardware, yet its effect on fine-grained lexical competence is uncharacterized. We audit 27 quantized artifacts across 13 families and four architecture backbones, 0.35B-14B parameters, evaluated down their published ladder to Q2_K (about 2.6 bits per weight), on 429 frequency-validated rare English words under two probes: surface inclusion of a prompt-supplied word and its one-sentence definition, scored by a tiered multi-synonym matcher, its error measured by a blind LLM-judge census of every definition, with human verification. Three regimes emerge at Q2: total collapse into unusable builds, severe semantic dissociation in sub-2B models, and mostly robust preservation above about 3B. In every sub-2B artifact, definitions fall 20-67% below the artifact's baseline, typically several times the inclusion loss. Two controls separate rarity from task difficulty: within the rare set, loss rises with rarity in six of seven sub-2B artifacts, and on a 100-word common-word set rare words lose more than common words in all eight, significantly in six. Tokenizer vocabulary size does not predict the damage (Spearman rho=0.12); parameter count dominates (rho=0.72), confirmed within five of six same-tokenizer families. Q4_K_M remains lexically clean at >=1B. The damage is frequency-graded, provider-dependent, and not calibrated by WikiText-2 perplexity: across nine artifact-matched ladders, near-identical Q2 penalties (44.7%/47.6%) separate an artifact keeping its definitions (3.6%) from one losing them (43.6%). Aggressively quantized small models can keep generating fluent text while no longer knowing what it means, risking hardware-constrained deployments in domains where semantics carries consequences. Validation must be per artifact.

To mitigate the quadratic complexity bottleneck of the Transformer, sparse attention has emerged as a pivotal technology. Despite the extensive empirical success of sparse Transformers, the theoretical understanding of sparse attention remains fragmented. In particular, two fundamental questions remain unclear: (1) How does sparsification affect the information fidelity of attention mechanisms? (2) How does this information loss interact with the generalization behavior of the model? To bridge this gap, this paper proposes a systematic analysis of the Jensen-Shannon (JS) divergence and of the generalization gap of sparse attention mechanisms. Specifically, we first characterize the approximation error via the JS divergence. Through an order-statistics-based concentration analysis of the truncation mass alpha --- where the attention scores are assumed to be independent and identically distributed sub-Gaussian random variables with parameter sigma --- the JS divergence between the full attention distribution and the sparse attention distribution is shown to admit the closed form log 2 + ((1 - alpha)/2) log(1 - alpha) - ((2 - alpha)/2) log(2 - alpha). Subsequently, we derive a generalization bound through Rademacher complexity, quantified by O(gamma * sqrt(M/n) * (sqrt(log(3eL/M)) + sqrt(pi)/2)). Furthermore, building on a mutual-information-based generalization bound together with an entropy and covering-number analysis of the sparse hypothesis class, we obtain the sparsity-dependent generalization bound O(sqrt((M/(2n)) * (log(eL/M) + log(1 + 2/epsilon)))). Our analysis shows that sparsity reduces the complexity of the considered hypothesis class while introducing approximation error that can be quantified by the JS divergence. These findings provide a theoretical characterization of the trade-off between information fidelity and generalization in sparse Transformer architectures.

On-device assistants run GPU-based LLM inference alongside CPU retrieval on unified-memory systems. Under a saturated local-retrieval workload, four concurrent retrieval workers raise 95th-percentile (p95) decode latency by 60-61% on two M4 systems, whereas prefill latency rises by only 5.7-6.9%. We study LLM phase as an admission signal for independent CPU retrieval under controlled LLM workloads. PHASEGATE calibrates separate concurrency limits for prefill and decode, selecting four and one on our base-M4 configuration. Under a backlogged queue, it achieves 2.0 times the aggregate retrieval throughput of the best tested feasible fixed policy, with both p95 LLM latency metrics within 1.25 times their no-retrieval baselines in all seven held-out runs. A phase-blind control, TimeGate, uses the same two limits on a calibration-derived schedule without observing LLM phase. It achieves similar retrieval throughput but violates the output-token latency limit in every run. M2 and M2 Pro Mac minis reproduce the policy ordering, while output-length sweeps show that the advantage narrows as decode occupies more of each request.

As large language models (LLMs) grow in parameter count, model storage and parameter memory traffic have become major bottlenecks to efficient deployment. Codebook-based weight quantization reduces these costs, but imbalanced nearest-codeword assignments during fitting can leave some codewords insufficiently updated, limiting effective codebook utilization. To address this limitation, we propose Balanced Assignment Refinement for Quantization (BARQ), which improves quantization quality through balanced fitting of existing codebooks. Specifically, we first compute joint soft assignments between weight blocks and codewords through entropically regularized optimal transport with uniform marginals and curvature-weighted reconstruction costs, ensuring equal positive fitting mass for every codeword in the exact solution. We then refine the codewords through an assignment-weighted barycentric update, which we prove minimizes the fitting objective for fixed assignments. For finite Sinkhorn iterations, the implemented update retains this optimality provided all codeword masses exceed the denominator floor. Finally, we discard the soft assignments and use the refined codebook for standard hard nearest-codeword encoding, with our analysis establishing sufficient conditions for reducing hard-quantization distortion and evaluation loss. Across multiple LLMs, BARQ achieves lower perplexity and higher mean zero-shot accuracy than the evaluated baselines at comparable bit budgets. The code is available at https://github.com/chenhangcuisg-code/BARQ.

Serving an LLM means setting dozens of inference-time knobs, from per-layer KV retention to per-layer expert counts. Practice sets them with mechanism-specific heuristics that return a single operating point and do not scale to layer-wise search spaces. We recast inference configuration as constrained multi-objective black-box optimization and build InferOpt, a reusable search framework that requires only variable bounds, a deterministic resource cost, and an evaluation hook. InferOpt searches on a frozen sampled proxy set, rejects over-budget candidates before any model call, tightens the budget adaptively, and re-validates Pareto representatives on full-scale dataset. One pipeline covers a 28-dimensional continuous KV space (Qwen2.5-7B) and a 26-dimensional discrete MoE space (DeepSeek-V2-Lite). On KV, post-prefill pruning cuts the 16K cache by 64.4% and TPOT by 22.9--48.5%, and the searched layer-wise budget by InferOpt beats a matched uniform budget by 7.3% and 14.0% of the Full KV reference points. On MoE, a searched top-k schedule by InferOpt removes 43.0% of routed token--expert pairs while staying within 0.59 points of the default, closer than the matched-budget baselines. Against Random Search, NSGA-II, and MOTPE, InferOpt leads on both spaces, taking the best proxy hypervolume and the lowest retention on KV and staying closest to the uncompressed reference at the lowest experts on MoE.

Diffusion models for structured scientific generation must produce samples satisfying hard geometric constraints imposed by physics, chemistry, or biology, yet inference in these settings is prohibitively slow, demanding hundreds to thousands of neural-function evaluations per sample. We unify eight state-of-the-art models spanning medical volumetrics, molecular conformations, protein backbone design, crystal structure prediction, and multi-view 3D scenes under a single abstraction, Constraint-Manifold Diffusion Models (CMDMs), in which the target distribution is supported on a manifold defined by an externally specified constraint map. All existing acceleration families fail on this class: quantization exhausts memory on high-dimensional volumetric operators; pruning breaks constraint fidelity; fast ODE solvers allow trajectories to drift off the constraint manifold; and feature-caching heuristics are blind to constraint geometry, inducing mode confusion in the high-noise regime. We introduce ManifoldCache, the first training-free, data-free accelerator designed from first principles for CMDMs. The key insight is that the conditional score decomposes orthogonally into a normal component, which enforces constraint satisfaction, and a tangential component, which navigates within the manifold. Exploiting this structure, we prove that the noise-schedule midpoint is a sharp safe-caching boundary: caching before it incurs provably bounded error, while caching after it guarantees a strictly positive fraction of trajectories suffer mode confusion, a gap that persists up to the boundary. We further prove that deeper network blocks admit provably larger certified cache strides within the safe phase, as a consequence of the score decomposition propagating through block Jacobians. The resulting schedule requires no calibration data, along with zero training overhead.

Inference systems determine how fast and how cheaply language models can be served, so making them faster has direct practical value. However, prior work focuses mostly on optimizing certain parts such as kernels or memory within the large system. In this work, we take a holistic approach and apply agentic self-evolution to optimize the whole system end-to-end. Our SEIS (Self-Evolving Inference Systems) autonomously optimizes the entire mini-sglang engine without human intervention through iterative sessions with inherited experiences and code changes. Serving Qwen3-0.6B on H100, the resulting engine reaches 3.27X the throughput of the original mini-sglang implementation and beats SOTA engines like vLLM, TensorRT-LLM, and SGLang in the single-request workload. The correctness of the optimized inference engine by SEIS is tested in terms of numerical difference and downstream accuracy on math and long-context retrieval tasks. The code and session histories show that the speedup comes from redesigning the whole engine and that building on earlier sessions beats independent attempts. These results suggest that agentic self-evolution can optimize a complex system end-to-end. The evaluation also has to evolve with the engine, and letting agents evolve it is a natural next step.

Autoregressive generation in Large Language Models (LLMs) is constrained by the memory and computational demands of attention mechanisms. Sparse attention methods mitigate this cost by selecting only high-probability entries of the attention matrix. We observe that in many such methods, this renders the probability-value multiplication negligible, shifting the bottleneck to the query-key step. Key heads can therefore be reduced to accelerate inference, while retaining more value heads preserves capacity with limited additional decoding cost. We introduce Sparse Asymmetric Group-Query Attention (SAGA), which decouples key and value head counts to exploit this principle, and pair it with approximate top-N (Atop-N) attention, a simple sparse attention method designed to study the interaction between sparsity and head-count asymmetry. We formalize the benefits of this asymmetry theoretically and validate them empirically through latency measurements and quality evaluations on models up to 1.5B parameters. Together, SAGA and Atop-N achieve end-to-end decoding speedups exceeding $2\times$ over our full-attention GQA baseline at long contexts. Models trained from scratch with SAGA nearly match the quality of comparable GQA variants on the evaluated benchmarks. To facilitate adoption, we introduce an efficient fine-tuning method that converts pretrained models to the SAGA architecture, enabling practitioners to benefit from our approach without costly retraining.

Vision Transformers (ViTs) achieve strong performance on image recognition and mobile vision applications, but their high-dimensional linear projections and attention computations still impose substantial storage and inference costs. Extremely low-bit quantization is a promising solution, yet ViTs often suffer severe accuracy degradation because conventional real-valued scalar codebooks are poorly matched to the directional geometry of Transformer projections. We present RPFQ-ViT, a Rotated Phase-Frame Quantization method that quantizes paired channels in two-dimensional phase planes, enabling low-bit codes to better preserve projection directions while recovering magnitude with lightweight scaling. RPFQ-ViT serves as a drop-in QAT replacement for nn.Linear and does not modify the standard real-valued attention, normalization, or activation computation graph. On ImageNet-1K, RPFQ-ViT-B/16 reaches 79.33% Top-1 / 94.48% Top-5 under W2/A4, Swin-T reaches 79.30% Top-1 / 94.79% Top-5 under W2/A8, and DeiT-S reaches 77.41% Top-1 / 93.11% Top-5 under W2/A8. Ablations, phase-geometry analysis, and direction-preservation metrics show that channel pairing, learnable rotation, phase-anchor learning, and residual phase refinement each improve quantization quality. We further deploy RPFQ-ViT image-classification models on native iOS and Android runtime stacks; with 2-bit packed weights, model size shrinks by roughly $5.4$-$7.1\times$ relative to FP32 and end-to-end on-device latency drops by $1.4$-$1.6\times$. All ImageNet results trained in our codebase use a matched 300-epoch recipe and are reported as mean accuracies over three independent runs. These results show that RPFQ-ViT provides a favorable trade-off among accuracy, compression, and practical mobile deployment for extremely low-bit ViTs.

Long-context decoding is limited by memory bandwidth, because every output token reads the KV cache of every layer. Sparse decoding reduces this cost by reading only part of the KV cache. We observe that the number of pages a query needs varies widely across KV heads, layers and steps. Fixed budgets are simple, but they are sized for demanding cases and tuned per workload; adaptive budgets follow this variation more flexibly, but existing designs pay for it with extra selection cost or training. At the kernel level, FlashAttention-3 (FA3) and FlashInfer are designed for rows of similar length: with page lists whose length differs per KV head, they either pad the lists (forfeiting much of the sparse saving), leave thread blocks unbalanced, or rely on a host-side plan that runs outside the CUDA graph. We propose MOIRA, a training-free sparse decode path in vLLM whose budget adapts per KV head and per layer. For every request, layer, KV head and step, a coverage rule keeps the smallest set of pages whose estimated attention mass reaches a fraction $γ$. A new kernel, self-planning attention, lets each thread block derive its own share of the work from the list lengths, so the whole decode step stays inside the CUDA graph. On an H200, at RULER's 128k context, MOIRA with $γ=0.99$ matches dense accuracy while reading about 30% of the pages and reduces the time per output token (TPOT) by 2.2-2.5$\times$ relative to dense FA3; with $γ=0.98$ it reduces TPOT by 2.7$\times$ and stays within the noise of dense. Under high serving load it raises throughput by up to 51%. These results suggest that a budget adapted per head and layer, paired with a kernel that keeps such budgets inside the CUDA graph, makes sparse decoding both flexible and fast.

Subliminal learning is a phenomenon where a student language model acquires a teacher model's behavioral traits by training on semantically unrelated outputs. It is a subtle statistical phenomenon as trait transmission relies on weak statistical patterns in the generated data. To understand trait transmission between teacher-student pairs, we study cross-trait transfer: how data generated under one teacher trait changes the student's preferences of other traits. To this end, we introduce a directed trait-transfer matrix that quantifies these effects using log-probability gains for student answers. We find that the trait-transfer matrix reveals clusters of related traits, with students sometimes developing preferences for traits similar, but not identical, to the teacher's trait. Such cross-trait structure can be partially captured by output distribution metrics and representation-based metrics. Further, we analyze trait development and interaction: learning dynamics shows a progression from broad shared shifts toward more trait-specific transfer, and multi-trait experiments suggest that opposed traits can enhance such differentiation. Together, our findings reveal salient statistical structures over trait transfer and competition, thus providing a broader view of how hidden preferences are transmitted in subliminal learning.

Existing activation steering methods often assume that a high-level concept can be mediated by a single steering direction. To support this, two complementary interventions should be achieved: additive steering should induce the target behavior, while directional ablation should suppress it. Yet behaviors may occupy richer activation geometries beyond a single direction, and semantically similar behaviors may be represented by distinct directions. In this work, we study how many directions can reliably control two different types of refusal behaviors: refusal triggered by the safety alignment and refusal in general contexts. Given the limited expressive capability of a single steering vector, we study the general setting of steering subspaces and introduce the notion of steering dimensionality as the minimum subspace dimensionality required to reliably control a behavior. We characterize sufficient steering subspaces that cover the full extent of the target behavior through both the (monotonic) improvement before the sufficient dimensionality, and the saturation beyond it. Empirically, we find that refusal triggered by safety alignment is 1-dim steerable, while multiple distinct steering directions can achieve comparable control. In contrast, refusal in general contexts exhibits substantially richer activation geometry where even 5-dim steering subspaces fail to reliably capture its full steerable variation. Our results reveal that the activation geometry underlying refusal is highly context-dependent and can be substantially more complex than a single linear steering direction.

Detecting responses that retrieval-augmented generation does not ground in its context trades speed against accuracy: surface checks miss paraphrased fabrication, sampling-based methods cost extra generations. Hidden-state probes sit between the two, but every existing one reads the generating model's own activations, so a change of generator invalidates the detector and a closed-weight generator is out of reach. This paper removes that coupling. The Grounding Probe is logistic regression over the mean-pooled middle-layer hidden states of an observer language model that reads the context, question, and response in one forward pass and generates nothing, with the recipe it needs: pool over response tokens, read a middle layer, and control capacity, which closes the train-test AUROC gap from 0.087-0.202 to 0.009-0.013. Asking the observer outright, rather than reading its hidden state, costs at least +0.166 AUROC in every one of four models. Fitted on 15,090 annotated responses it reaches 0.879-0.894 AUROC on RAGTruth test across four observers, and 0.924 AUROC with 0.820 F1@0.5 averaged with a supervised span detector, 0.060 above that detector alone. One probe holds across six generators, and hold-out controls, including one in which no evaluation prompt appears in training, bound the cost of removing a generator at about 0.02 AUROC. Code, probes, and predictions are released.

Glitch tokens are anomalous vocabulary entries that can cause large language models (LLMs) to produce outputs inconsistent with their inputs. Existing repair methods require access to model internals, making them impractical for frozen checkpoints. We investigate whether glitch tokens can be repaired outside the model by optimizing the input tokenization. An empirical study on BPE merge-rule deletion reveals that (1)deleting a glitch token's merge rule can fix a substantial fraction of failures, yet disrupting normal tokens sharing intermediate merge nodes causes the overall glitch rate to rise, and (2)different decomposition granularities yield non-monotonic fix rates while collateral damage on normal tokens grows monotonically. Motivated by these findings, we propose GlitchPatch, a repair framework for frozen language models based on local retokenization, consisting of two stages: the offline stage uses Behavioral Path Optimization (BPO) to find the behaviorally optimal replacement token sequence for each glitch token and compiles validated replacements into a rule table; the online stage substitutes only the IDs of matched glitch tokens in the canonical token sequence, with no modification to model parameters or internal states. Experiments on ten models spanning six tokenizer families show that GlitchPatch achieves an 85.10% mean fix rate, outperforming the strongest baseline by 14.37 percentage points, and reduces the average glitch rate from 14.88% to 2.27%. GlitchPatch achieves a 0.00% RR in full-vocabulary evaluation and leaves rule-unmatched inputs unchanged by design. We further evaluate the practical impact of repair from the perspectives of time cost, language understanding, and capability, supporting its deployment feasibility. We hope this work provides a practical option for improving tokenizer reliability.

Structured prediction tasks pose unique challenges for in-context learning (ICL): their compositional outputs require modeling fine-grained, token-level patterns that sentence-level approaches fail to capture, and their task-specific annotation conventions are human-defined artifacts that cannot be acquired through pretraining alone. We propose Verb-ICL, a selective annotation framework for ICL-based structured prediction that addresses both challenges. Verb-ICL first selects representative examples using a token-level coverage strategy that captures local semantic patterns critical for structured prediction, then generates actionable error feedback that codifies task-specific annotation guidelines and incorporates this feedback into ICL demonstrations. We evaluate Verb-ICL on six structured prediction datasets spanning information extraction and semantic parsing. Experiments with recent LLMs show that Verb-ICL consistently outperforms strong selective annotation baselines under low-resource settings and continues to provide gains as the annotation budget increases. Extended analyses demonstrate that the generated feedback is predominantly useful across a four-category quality taxonomy, generalizes as task-level guidance beyond instance-specific corrections, and improves performance regardless of the underlying selection strategy.

Structural diagrams are widely used to represent complex systems and relational information across scientific, engineering, procedural, and spatial domains. Recent vision-language models (VLMs) have become increasingly capable of recognizing diagram elements and reasoning about their content, while complete diagram topology extraction remains comparatively underexplored. In this paper, we study diagram-to-graph topology extraction: extracting all diagram entities and the complete relations among them. To enable large-scale supervised training and systematic evaluation of this task, we introduce Knossos, a benchmark of 19,200 diagrams across six diverse domains, with 245,179 nodes and 439,740 edges. Its symbolic generation process provides exact alignment between rendered diagrams and annotations of complete topology, relation types, and connector geometry. To address the modeling challenge of complete topology extraction, we also present Ariadne, a structured framework that decomposes the task into node inventory extraction and source-conditioned edge prediction. Extensive experiments show that training on Knossos substantially improves complete topology extraction in smaller open-source VLMs. Ariadne further improves over one-step extraction under matched supervision, demonstrating the additional benefit of structured decomposition. It achieves the highest average Edge F1 among the evaluated methods on Knossos, while both backbone variants also improve over their unadapted counterparts on the real-world external benchmark. Code and benchmark are available at https://github.com/bangwayne/knossos_Ariadne_Public.

Beyond scaling their parameters and data, large language models currently gain versatility on new problems along a single axis: the tokens they spend on chain-of-thought (CoT). We investigate whether dynamic routing programs, which execute a subset of the model's layers or iterate some of them, can open a second axis of test-time adaptation, complementary to CoT and free of any gradient update. Prior work showed that such programs exist and bring accuracy and efficiency gains on problems similar to those they were trained on; we ask whether they can also be identified rapidly, from a handful of demonstrations (3 or 10), by a strategy that transfers across models and tasks without training. First, we find that strategies that select programs by the probability they assign to the demonstrations' labels, arbitrated by the model's own confidence, bring consistent gains: on average over the 49 tasks of MMLU and substantially on four of seven models, and most of all on far out-of-distribution tasks such as ARC-AGI, where programs double the accuracy of a 7B model whose CoT fails. Second, on MMLU across the seven post-trained models, routing complements CoT in practice: the two succeed on different queries, and their composition exceeds CoT alone. Despite these gains, our analyses show that confidence-based selection leaves much of the potential untapped, in two places in particular: (1) in surfacing the routing potential that is already present early in pre-training but becomes harder to select after post-training, and (2) in making models robust to the refinements routing introduces, since unsuccessful routes tend to drive the residual stream out of the distribution that the following layers expect. Together, our results point to dynamic routing as a paradigm for extending the plasticity of existing and future LLMs in rapid test-time adaptation.

Multi-modal large language models (MLLMs) achieve strong modality understanding by pairing a large language model (LLM) with an encoder for a target modality such as vision, video, or audio. However, improving an MLLM's capability for a given modality typically requires additional training on large modality-specific datasets, incurring substantial data collection and compute costs. Model merging offers an alternative, but it is often infeasible for data-scarce, large per-sample size, or domain-specific modalities (e.g., audio and video), where same-modality model variants are rarely available. In this work, we characterize an intriguing asymmetric phenomenon: merging a well-aligned, data-rich source-modality MLLM into a data-scarce target-modality MLLM substantially improves the target on its own benchmarks. Our theoretical and empirical analyses show that this gain stems from enhanced alignment between modality-specific and textual tokens, induced by the stronger donor modality. Specifically, we derive a mutual-information lower bound that is monotonic in alignment-related quantities and strongly correlated with downstream MLLM performance. Building on this principle, we propose Directional Cross-modal Alignment Transfer (DCAT), a novel framework that transfers textual alignment from a strong, well-aligned source (donor) modality to a weak target (recipient) modality, boosting target-modality performance without further fine-tuning. We further show that the alignment-enhancing objective admits a closed-form weight-space solution computed from only a small calibration set. DCAT outperforms existing model-merging methods, offering an efficient path toward cross-modal alignment transfer. Project page with code is available at \url{https://seohoiki3215.github.io/DCAT_project_page}

Large language models (LLMs) are increasingly used to generate, complete, and transform information in settings where their outputs can shape consequential decisions, raising concerns about their impact on demographic disparities. In this context, causal inference provides a principled basis for assessing fairness, because it attributes observed disparities to the mechanisms that generated them, which a purely statistical approach cannot do even with infinite data. In LLM generation, a query may request several causally related variables, each of which is both an outcome of interest and a possible cause of other outputs, and the information supplied in the prompt need not follow a topological or a temporal order. This calls for methods that can analyze and selectively remove disparities from such a flexible generation process. In this paper we introduce Causally Fair Generation with LLMs (CFG, for short). CFG extracts relevant concepts, grounds generation in a reference population and causal diagram, and removes user-selected causal effects. CFG also allows pathways deemed justifiable for the task's utility to be retained, which is known in legal literature as business necessity. Further, we provide formal guarantees for our method when eliminating all discriminatory causal effects in the adapted population model, under appropriate causal assumptions. We evaluate CFG with four LLMs in three real-world settings based on population data and on a synthetic dataset with a known causal ground truth.

Chain-of-Thought (CoT) faithfulness detectors are widely used to audit reasoning models, yet a detector is itself a predictor whose verdicts are treated as stable properties. We ask whether a detector is faithful to itself under distribution shift. We formalize meta-faithfulness as an invariance principle: a valid detector must return identical verdicts on traces that differ only by transformations preserving ground-truth faithfulness. We prove three results: (i) no detector using only intervention-response profiles can separate faithful from epiphenomenal mechanisms with identical signatures; (ii) any detector relying on shift-sensitive features violates invariance at a rate independent of its in-distribution accuracy; (iii) an asymptotic certified selective-risk guarantee enables confident abstention. We operationalize the principle in FaithShift, a stress-test protocol spanning ten shift axes, and propose SIFT, a hidden-state trajectory detector trained with cross-environment invariance objectives and certified abstention. Across 14,996 traces, four domains, and eight models, three findings emerge. First, transfer collapse is real: all existing detectors show gaps $\geq 0.15$ AUROC. Second, the dominant bottleneck is sampling stochasticity, not shift: over 80% of detector instability stems from random seed variation, falsifying our preregistered prediction that shift-attributable violations exceed 0.25. Third, SIFT cuts invariance violations by 64% over the best single-seed baseline, but a four-seed ensemble of any detector narrows the margin to 0.01 (indistinguishable at matched coverage, $p=0.21$), and SIFT needs a 51% abstention rate. Cross-model transfer degrades from within-family to cross-family to open-weight-to-API, partly closed by multi-model training. We offer a framework for auditing auditors: the real barrier is detector variance, not distribution shift.