Leaderboards, eval harnesses and ablations — the contested business of deciding which model is actually better.
Frequency-domain information is beneficial for semantic segmentation. However, existing frequency-based methods often suffer from insufficient frequency utilization, inefficient fusion, and high compu...
Recent advances in vision-language-action (VLA) and world-action models (WAMs) have significantly improved robotic manipulation, yet deploying robots in the open world demands a higher level of autono...
Text injection trains speech recognition models on unpaired text alongside paired audio. We feed the speech encoder synthetic input derived from text so the whole model learns from it. Previously, we...
Language models for metal-organic frameworks (MOFs) rely on formats that separate atomic geometry from structural topology, forcing models to choose between unstructured coordinate lists and non-spati...
Ask a model "how many settlements did the Normans control?" and it answers that the passage does not say. Tell it "given that the Normans controlled 12 settlements of 400 inhabitants, what is the tota...
Decentralized multi-agent systems (MAS) enable large language model agents to interact without centralized control. Agents need to adapt to other agents at test time and reuse accumulated experience t...
Large reasoning models are costly to serve because they are large and because they generate long chains of thought (CoT). Weight pruning and CoT compression address these two costs separately, but the...
Text-to-motion models generate motion only for the skeletons they were trained on, while a new character arrives as a skeleton file and nothing else. We study text-to-motion for skeletons unseen in tr...
Learning audio representations that effectively bridge holistic global soundscapes with fine-grained acoustic moments is a long-standing goal in self-supervised learning. Recording-level contrastive o...
Autonomous agents are increasingly deployed across sequences of related tasks, yet they often fail to retain and improve their knowledge over time. Existing approaches convert interaction trajectories...
Agents end many steps in a typed decision over a set of candidates: which screen element to click, which product to pick. System One models answer in one forward pass, and Jeva showed that four proper...
Graph Neural Networks are widely used to learn structural representations of relational data. Hierarchical graph pooling learns multi-level representations by progressively coarsening graphs; however,...
Hardware-aware neural architecture search (NAS) requires jointly optimizing predictive accuracy and device-specific latency over discrete search spaces. While multi-objective swarm optimizers show pro...
Simon Willison announces the release of Kimi K3.1 Pro on the Kimi API, noting it is not yet open‑weights but full weights will be available by Nov 4 2026. The model offers a high‑speed variant 2‑3× faster at double the price, with US dollar pricing of $2.00/million input tokens and $10/million output tokens—matching GPT‑6.1 Sol and half the cost of Claude Opus 5.5. Benchmarks from Moonshot show Kimi roughly on par with Opus 5.5 and GPT‑6 Astra, scoring 55 in Artificial Analysis, ahead of several competitors.
Deploying Vision-Language Models for video anomaly detection confronts a fundamental constraint. Their semantic reasoning demands computational budgets incompatible with the scale of real-world video...
MoEZip is a routing‑aware key‑value (KV) cache compression technique designed for sparse Mixture‑of‑Experts (MoE) language models. It introduces a Fisher‑information‑based metric to assess how sensitive expert routing is to cache perturbations, and combines this with attention weights to score contexts for eviction. Experiments demonstrate that, even with only 10% of the KV cache retained, MoEZip outperforms state‑of‑the‑art baselines by significant margins on LongBench and RULER benchmarks.
DiffResearch is an open multi‑agent deep research framework that replaces the traditional autoregressive language model backbone with a diffusion language model (dLLM) across all stages of research, from planning to writing. In experiments on 100 Deep Research Bench tasks, the diffusion backbone achieves roughly six times faster end‑to‑end latency, largely due to significant speedups in planning and long‑form generation. The framework also introduces dFast, a decomposition mode that reads intermediate denoising states to generate subqueries, yielding a median 1.99× speedup without compromising report quality and achieving the best overall score on the RACE rubric. whyItMatters":"The study demonstrates that diffusion language models can dramatically reduce latency in deep research systems, enabling faster interactive use while maintaining or improving output quality."
The paper presents neural‑network (NN) transistor models that predict drain current and charges for a 3 nm nanosheet, and applies them to standard cell library characterization. The NN models achieve sub‑0.11 % error in delay and 0.76 % in total power across a wide range of circuits, while reducing characterization time by up to eight times compared to the industry‑standard Verilog‑A BSIM‑CMG model. The study also explores how NN size affects simulation precision and provides guidance for accurate circuit simulation.
The paper introduces DynaCertBench, a benchmark of 105 diverse systems for certifying stability, safety, and related properties, and presents LLM2Cert, a framework that uses large language models to generate and verify candidate certificates via SMT solvers or Lean. LLM2Cert achieves a 77.1% verification rate on the benchmark, improving by 17.1 percentage points over a single-shot approach, and successfully verifies 22 certificates that All-SMT could not resolve. The authors provide code and data on GitHub for reproducibility.
The paper introduces WaxalNet, a benchmark and adaptation study for spontaneous speech recognition across 19 African languages using the WAXAL corpus. It evaluates zero‑shot and fine‑tuned performance of several multilingual foundation models (Whisper, MMS, Omnilingual ASR, SeamlessM4T‑v2), showing that fine‑tuning can substantially reduce word error rates and that LLM decoders generally outperform CTC decoders when the encoder is shared. The authors also demonstrate that a single multilingual model can serve all languages with only a modest WER penalty and provide a native‑speaker audit of failure modes.