EAServe introduces an encode-aware disaggregated serving framework for multimodal large language models (MLLMs), restructuring the traditional Prefill-Decode pipeline into a three-stage Encode-Prefill-Decode (EPD) system. By treating Encode as the control point, EAServe coordinates load‑adaptive micro‑batching, rate‑controlled offloading to prefill workers, and dynamic SM partitioning to balance GPU utilization across stages. Its Hybrid Auto Selection (HAS) layer optimizes GPU allocation, encode batch size, and offload ratio using capacity profiling and Bayesian optimization, achieving up to 4.3× higher goodput compared to NVIDIA Dynamo and 1.7× higher than vLLM on various MLLM architectures.
By Kunxiong Zhu, Zhihao Shu, Hangyu Zheng, Minghai Qin, Miao Yin, Gagan Agrawal, Wei Niu
WEECFP-SuRGE introduces a position‑aware substructure encoding method that combines tokenized hierarchical Morgan fingerprints with graph‑distance‑dependent rotations applied at the input and within transformer self‑attention. The approach captures local chemistry, long‑range interactions, and molecular topology without requiring external pretraining or 3‑D conformer generation. Benchmarks on MoleculeNet and the Therapeutic Data Commons ADMET datasets show competitive performance, and a reconstruction procedure correctly identifies constitutional isomers for 92.6% of a 4,200‑molecule library.
By Robert Epps
The paper reports on applying AutoResearch—a large language model that iteratively edits training scripts—to optimize embedding systems for a book recommendation pipeline at production scale. Over twelve weeks, the authors ran 220+ experiments across two representation‑learning systems, uncovering five recurring failure modes (infrastructure fragility, agent memory decay, search‑direction stagnation, iteration‑cost asymmetry, and metric fixation) that were not present in smaller settings. They propose a three‑principle scaffolding (prevent, persist, redirect) to address these failures, achieving a 1.82× lift in Recall@6, a 2.1× lift in coherence, and an autonomous text‑only fallback that expanded catalog coverage by 5.8×.
By Aparajith Chandran, Juwon Kim, Saurav Jha, Pablo Castells, Florian Hottier
BeatGraph is a self‑supervised model that represents infant ECG recordings as graphs of individual heartbeats rather than fixed‑length patches, allowing it to capture the higher heart rates and distinct waveform patterns of infants. The model uses a shared beat encoder, a Transformer for temporal ordering, and graph attention layers to produce a window embedding, which is pretrained on a large unlabeled infant ECG corpus and fine‑tuned for tasks such as sleep‑wake detection, infant‑state classification, activity‑source identification, and affect recognition. BeatGraph achieves state‑of‑the‑art performance on multiple infant‑specific benchmarks and transfers well to pediatric and adult ECG datasets, while also releasing the first public infant ECG corpus collected in diverse home and classroom settings.
By Mohammad Nur Hossain Khan, M. S. Krafczyk, Beverly G. Bolster, Nancy McElwain, Mark A. Hasegawa-Johnson, Bashima Islam
ChemMLLM is a unified chemical multimodal large language model designed for molecule understanding and generation across text, SMILES strings, and images. The authors curated five multimodal tasks and benchmarked ChemMLLM against leading general MLLMs, chemical LLMs, and specialized models, finding it outperforms general-purpose MLLMs and matches specialized models on all tasks. The study demonstrates that a single foundation model can handle diverse cross‑modal chemical tasks, including image generation, enabling more intuitive visual human‑AI interaction.
By Qian Tan, Di Zhang, Ben Gao, Peng Xia, Wanhao Liu, Shufei Zhang, Wanli Ouyang, Lei Bai, Yuqiang Li, Tianfan Fu
The paper introduces a continuously synchronized dataset suite covering the entire lifecycle of decentralized prediction markets, from market creation to final settlement. It integrates market metadata, fill-level trading records, and oracle-resolution events into a unified relational system, providing over 3.29 million market records, 1.90 billion order executions, and 21 million oracle events from October 2020 to the present. The authors detail the data model, collection pipeline, and consistency mechanisms, and demonstrate the dataset’s usefulness for sports betting, economic forecasting, and blockchain research, with public access via a website and interactive tools.
By Huaiyu Jia, Luofeng Zhou, Wentao Zhang, Lin William Cong, Siguang Li, Shuo Sun
MeshGraphNet-Transformer (MGN‑T) is a new architecture that fuses Transformers’ global modeling with MeshGraphNets’ geometric inductive bias, keeping a mesh‑based graph representation. It replaces iterative message passing with a physics‑attention Transformer that updates all nodal states simultaneously, enabling efficient learning on high‑resolution meshes with diverse geometries, topologies, and boundary conditions. MGN‑T accurately models impact dynamics, self‑contact, plasticity, and multivariate outputs, outperforming state‑of‑the‑art methods on classical benchmarks while using far fewer parameters.
By Mikel M. Iparraguirre, Iciar Alfaro, David Gonzalez, Elias Cueto
The paper introduces Geometric Anchor Preference Optimization (GAPO), a method that replaces the static reference policy in Direct Preference Optimization with a dynamic, geometry-aware anchor—a small adversarial perturbation of the current policy. GAPO uses this anchor to adaptively reweight preference pairs based on local sensitivity, and defines an Anchor Gap that approximates worst‑case local margin degradation. Experiments show that GAPO improves robustness to noisy supervision while matching or surpassing existing LLM alignment and reasoning benchmarks.
By Youngjae Cho, Jongsuk Kim, Ji-Hoon Kim
TeDiServe is a cluster‑level serving system designed for diffusion language models (DLMs). It addresses DLM‑specific challenges such as the speed‑quality tradeoff from confidence‑based denoising, variable parallelization under fluctuating load, and non‑uniform per‑step costs from approximate KV caching. By employing deadline‑aware scheduling, adaptive load control, and a quality‑aware optimization for cluster reconfiguration, TeDiServe achieves up to 56.6 percentage points higher SLO attainment and reduces end‑to‑end latency by up to 46% with less than 1% accuracy loss.
By Tzu-Tao Chang, Benjamin Yuanyang Hong, Kiet Pham, Shivaram Venkataraman
HybridInfer is a thermal‑aware reinforcement‑learning router that selects among on‑device, edge, and cloud large language model tiers based on a phone’s thermal headroom and query complexity. Trained offline with a Q‑learning policy, it balances quality, latency, cost, and a thermal penalty, and includes a locality bonus that encourages on‑device execution. In real Android tests on 210 prompts, the learned router outperformed two hand‑tuned heuristics in quality while maintaining the lowest cost, and it proved more reliable and faster than always‑on‑device inference, especially for long queries.
By Simran Koul
The paper investigates why per-layer pruning of IoT intrusion detectors can cause severe class-level failures. On the CICIoT2023 dataset, a two-layer convolutional detector pruned at 80% sparsity loses 16 accuracy points and half its macro‑F1, with 17 of 34 classes heavily damaged. The failure is traced to the first layer’s weight starvation, and the authors show that protecting those 192 weights or pruning globally, as well as recomputing normalization statistics, can prevent or repair the collapse.
By Md Anas Biswas
The paper introduces a reproducible benchmark for evaluating attention mechanisms in tabular foundation models, focusing on the distinct row and column attention patterns that differ from language model attention. It compares several backends—Torch SDPA, FlashAttention variants, vLLM, and SageAttention—across realistic tabular shapes on A100, H100, and B200 GPUs, revealing that optimal backend choice varies by attention type, hardware, and model specifics. The study finds FlashAttention generally performs best, but CuDNN can outperform it for column attention on longer sequences, while SageAttention excels for large row sequences beyond 16k rows.
By Maximilian Schambach, Clemens Biehl, Sam Thelin
Release: llm-anthropic 0.30
In addition to Claude Sonnet 5.5, this release adds the ability to run llm anthropic refresh to refresh the list of Anthropic models directly from their API - which...
Claude Sonnet 5.5, a new model from Anthropic, is announced as running 30% faster and costing up to 30% less than its predecessor while matching or exceeding performance on benchmarks. It is now the model powering the free tier on claude.ai, offering a more capable free experience than OpenAI’s ChatGPT free tier, which uses Luna 5.6. The release also mentions upcoming availability of Haiku 5.5 in the coming weeks.
Simon Willison reflects on the rapid and unexpected advancements in AI capabilities, particularly in areas like cyber, swarming, and message boards. He emphasizes that security posture requires more than system hardening; it must be embedded in company culture and involve people adapting alongside technological changes. Willison urges organizations worldwide to assess their resilience to sudden AI jumps, ensuring people, systems, processes, incident response, and communication are prepared for such surprises.
The paper explores Retrospection-Only Fine-Tuning (ROFT), a method where a language-model agent improves its behavior by generating and training on explanations of its own experiences, without external teachers or reward signals. In software‑engineering tasks with Qwen3.5‑4B, ROFT achieves comparable or better solve rates than GRPO while requiring fewer updates and training time, and can learn from failures alone. Behavioral analysis shows ROFT indirectly assigns credit to actions and can produce shorter, more direct solutions when prompted to focus on direct solutions.
The paper investigates whether different forms of intermediate computation in large language models—such as token-based traces, pause tokens, and latent reasoning—rely on the same underlying mechanism. By training five variants of GPTNeoX on an extended multi-hop reasoning task, the authors find that while vanilla, Chain-of-Thought, and Pause Token models perform well on in-distribution data, they fail to generalize to longer-hop out-of-distribution problems. In contrast, latent-reasoning models exhibit better depth generalization, with causal analysis revealing a sparse recurrent search circuit that implements forward reachability propagation across the graph.
SEABench is a benchmark designed to study endogenous misalignment in self‑evolving large language model agents. It contains 48 longitudinal task sequences across various evolution surfaces, task domains, and harm types, and includes an adaptive trajectory discovery pipeline that probes for failures while preserving task intent. Evaluations show that self‑evolution improves task completion rates but often introduces safety failures absent in non‑evolving baselines, with divergent safety behaviors reflected in agents’ chain‑of‑thought reasoning that can be monitored to mitigate unsafe actions.
QC-Stark is a benchmark that evaluates large language models on 11 quantum computing tasks, including circuit construction, debugging, compilation, error correction, and simulation. It comprises 2,750 evaluations across 10 models, 5 difficulty levels, and 5 seeds, revealing that overall model rankings can hide significant per-task differences, with Spearman correlation being insignificant for 4 of the 11 tasks. The benchmark’s measurement quality is validated by a 2‑parameter Item Response Theory model, prompt sensitivity analysis confirms ranking robustness, and all tasks are auto‑verifiable via execution, eliminating the need for manual evaluation. The code and data are publicly available on Hugging Face.
Mistral AI has announced the opening of a new hub in Munich dedicated to Physics AI and Industrial AI research. The facility will collaborate closely with German industry partners to advance AI applications in these fields.