NVIDIA showed that its 'Nemotron Speech' model speeds up clinical speech-recognition (ASR) evaluation. All five sources are NVIDIA official (with some Hugging Face and DeepMind links)—a single-vendor technical presentation, announcement-led but stepping into healthcare as an applied domain. It targets clinical ASR—jargon-heavy, accuracy-critical tasks like transcribing medical records—to speed evaluation and processing; the through-line is a shift from general model performance toward practical accuracy and efficiency in a specific domain. It stands as an application to lighten documentation work in care settings. But this is mainly a technical presentation—clinical accuracy validation, regulatory handling, and adoption breadth are to confirm.
NVIDIA speeds clinical ASR evaluation
NVIDIA speeds clinical ASR evaluation
Evaluate Clinical ASR Models Faster with Agent Skills and NVIDIA Nemotron Speech
NVIDIA speeds clinical ASR evaluation via Agent Skills, Nemotron Speech
AIエージェントもフィッシング詐欺に引っかかる？ 米セキュリティ企業がOpenClawで検証 結果は……
Varonis: AI agents can fall for phishing, tested with OpenClaw
Building a persistent cognitive architecture for LLM agents using Elixir and OTP
Persistent cognitive architecture for LLM agents via Elixir/OTP
Can Voice Agents Handle Bilingual Customers? Benchmarking Frontier ASR on Code-Switched Speech
ServiceNow AI benchmarks frontier ASR on code-switched speech
NVIDIA adds enterprise lifecycle control for AI infra via DGX Spark
NVIDIA converts FP8 checkpoints into fast inference engines via TensorRT
Accelerating Federated Learning Research with AI Agents and NVIDIA FLARE Auto-FL
NVIDIA FLARE Auto-FL and AI agents accelerate FL research
Fluid, natural voice translation with Gemini 3.5 Live Translate
Google debuts Gemini 3.5 Live Translate for real-time speech
Introducing North Mini Code: Cohere’s first model for developers
Cohere open-sources North Mini Code, its first agentic coding model
Ubuntu、サンドボックス化された開発環境をコマンド一発で構築。新機能「Workshop」リリース
Canonical launches Workshop, one-command sandboxed dev environments
Academic (arxiv etc.) 99 ▾
Operadic consistency: a label-free signal for compositional reasoning failures in LLMs
Operadic consistency flags LLM compositional reasoning errors label-free
Valid Inference with Synthetic Data via Task Exchangeability
Task exchangeability enables valid inference from synthetic data with guarantees
Beyond Uniform Tokens: Adaptive Compression for Time Series Language Models
Adaptive token compression streamlines time series language models
Beyond the Commitment Boundary: Probing Epiphenomenal Chain-of-Thought in Large Reasoning Models
Chain-of-thought reasoning crosses a 'commitment boundary,' study shows
SCSB prunes bagging ensembles up to 96% while improving calibration
Timeflies jointly models future observation existence and values in forecasting
A2D2: Fine-Tuning Any-Length Discrete Diffusion for Adaptive Decoding
A2D2 unifies reward-guided fine-tuning for any-length discrete diffusion
Genomic priors solve the cold-start problem in personalized health AI
NetCause: Counterfactual Learning for Root Cause Analysis in Large-Scale Networks
NetCause ranks network incident root causes via counterfactuals
Graphical Causal Reasoning for Root Cause Analysis in Cloud Networks
Causal graph traversal recalls 85.7% of cloud incident root causes
Heterogeneous LiDAR fusion and re-ranking boost place recognition in fields
GF-DiT: Scheduling Parallelism for Diffusion Transformer Serving
GF-DiT makes GPU parallelism schedulable for diffusion transformers
Optical Implementation of Equilibrium Propagation Using Spatial Photonic Ising Machines
Equilibrium propagation realized on spatial photonic Ising machines
Accelerating Speculative Diffusions via Block Verification
Block verification speeds up speculative sampling for diffusion models
PolyFlow embeds polytope constraints into flow matching, projection-free
MiniMax Sparse Attention enables efficient ultra-long-context LLMs
SmartFont: Dynamic Condition Allocation for Few-Shot Font Generation
SmartFont allocates global and local conditions for few-shot font generation
Hölder++: Improving the Quality-Coherence Trade-off in Multimodal VAEs
Hölder++ improves quality-coherence trade-off in multimodal VAEs
VideoMDM: Towards 3D Human Motion Generation From 2D Supervision
VideoMDM learns 3D human motion priors from 2D video supervision
SkillCAT: Contrastive Assessment and Topology-Aware Skill Self-Evolution for LLM Agents
SkillCAT verifies and routes self-evolved skills for LLM agents
SICI: A Semantic-Pragmatic Complexity Index Reveals Regime Shifts in LLM Stance Detection
SICI complexity index reveals regime shifts in LLM stance detection
MiniPIC: Flexible Position-Independent Caching in <100LOC
MiniPIC adds position-independent KV caching to vLLM in <100 LOC
Reroute, Don't Remove: Recoverable Visual Token Routing for Vision-Language Models
Reroute replaces visual-token removal with recoverable routing in VLMs
Context-Driven Incremental Compression for Multi-Turn Dialogue Generation
C-DIC compresses multi-turn dialogue context incrementally for stability
DIRECT: When and Where Should You Allocate Test-Time Compute in Embodied Planners?
DIRECT allocates test-time compute per prompt for embodied planners
Doc-to-Atom: Learning to Compile and Compose Memory Atoms
Doc2Atom decomposes documents into composable micro-LoRA memory atoms
System Report for CCL25-Eval Task 5: New Dataset and LoRA-Fine-Tuned Qwen2.5
PoetryQwen specializes classical Chinese poetry appreciation via CCPoetry-49K
TAHOE: Text-to-SQL with Automated Hint Optimization from Experience
Tahoe learns hints from experience to optimize production Text-to-SQL
ATLAS: Active Theory Learning for Automated Science
ATLAS uses active learning to discover interpretable behavioral models
APPO: Agentic Procedural Policy Optimization
APPO branches and assigns credit at fine-grained decision points
Breaking Entropy Bounds: Accelerating RL Training via MTP with Rejection Sampling
Bebop boosts MTP acceptance via rejection sampling to speed RL training
Latent World Recovery for Multimodal Learning with Missing Modalities
LWR recovers a latent world for multimodal learning with missing modalities
CHORUS: Decentralized Multi-Embodiment Collaboration with One VLA Policy
CHORUS controls multi-robot teams with one decentralized VLA policy
Claw-SWE-Bench: A Benchmark for Evaluating OpenClaw-style Agent Harnesses on Coding Tasks
Claw-SWE-Bench fairly benchmarks OpenClaw-style coding agent harnesses
ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit Mixing
ALIGNBEAM transfers safety logits across model vocabularies
Fourier Features Let Agents Learn High Precision Policies with Imitation Learning
Fourier features give point-cloud policies high-precision control
Measuring Semantic Progress in Multi-turn Dialogue via Information Gain
An information-gain metric measures semantic progress in multi-turn dialogue
PROJECTMEM: A Local-First, Event-Sourced Memory and Judgment Layer for AI Coding Agents
projectmem adds a local-first memory layer for AI coding agents
A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents
A five-plane reference architecture for runtime governance of AI agents
Harness In-Context Operator Learning with Chain of Operators
CHOP chains operators to generalize a frozen ICON to OOD tasks
CCKS: Consensus-based Communication and Knowledge Sharing
CCKS improves cooperative MARL via consensus-based knowledge sharing
INT8 quantization of Ideogram 4.0 holds FP8 quality on consumer GPUs
Mathematical perspective on genetic algorithms with optimization guided operators
A mathematical model of genetic algorithms with optimization-guided operators
The Impossibility of Eliciting Latent Knowledge
Paper formalizes the impossibility of eliciting latent knowledge
VIA-SD: Verification via Intra-Model Routing for Speculative Decoding
VIA-SD speeds speculative decoding via intra-model verifier routing
Re-evaluating Confidence Remasking in Masked Diffusion Language Models
Re-evaluation: WINO remasking adds little in masked diffusion LLMs
Can News Predict the Market? Limits of Zero-Shot Financial NLP and the Role of Explainable AI
Zero-shot financial NLP fails to beat baselines at predicting markets
Survey of automated embodied benchmark construction in five stages
Adaptive Multi-Resolution Procedural Knowledge Compression for Large Language Models
SKIM compresses procedural LLM skills via adaptive multi-resolution tokens
Implicit Neural Representations of Individual Behavior
Behavioral INR learns policy representations from unlabeled multi-policy data
Study finds best speech-text alignment regime at 4.17Hz frame rate
Survey of agentic environment engineering across its lifecycle
A Resource for Enthymeme Detection in Controversial Political Discourse
New dataset enables enthymeme detection in political discourse
Towards Responsibly Non-Compliant Machines
Position paper sketches responsibly non-compliant intelligent machines
FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents
FORT synthesizes shortcut-resistant tasks to train deep search agents
Analysis of 25M comments tracks the surge of 'AI slop' accusations
Toward Generalist Autonomous Research via Hypothesis-Tree Refinement
Arbor runs autonomous research via hypothesis-tree refinement
Language robustness in VLA models is a step-wise control problem
Notes2Skills: From Lab Notebooks to Certainty-Aware Scientific Agent Skills
Notes2Skills turns lab notebooks into certainty-aware agent skills
Beyond representational alignment with brain-guided language models for robust reasoning
Brain-guided language models strengthen robust deductive reasoning
Fine-tuning Multi-modal LLMs with ART: Art-based Reinforcement Training
ART fine-tunes frozen MLLMs by optimizing only the visual input
WorldReasoner: Evaluating Whether Language Model Agents Forecast Events with Valid Reasoning
WorldReasoner evaluates whether agents forecast events with valid reasoning
MultiToP patches visual tokens to cut video-LMM hallucinations
Fast Speech Foundation Model Distillation Using Interleaved Stacking
Interleaved stacking accelerates speech foundation model distillation
EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents
EEVEE: multi-dataset test-time prompt learning for self-improving agents
Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories
Data2Story: a multi-agent framework for verifiable, multimodal data stories
Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models
RL post-training aligns full-duplex speech models across four interactivity axes
ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Models
ReasonAlloc: hierarchical KV cache budget allocation for reasoning models
Ito maps: any-step stochastic flow maps for posterior sampling and control
ABC-Bench: An Agentic Bio-Capabilities Benchmark for Biosecurity
ABC-Bench evaluates LLM agents' biosecurity-relevant capabilities
FADA: a unified vision-language model for fetal ultrasound interpretation
RoboNaldo: motion-guided curriculum RL for powerful humanoid soccer shots
A History-Aware Visually Grounded Critic for Computer Use Agents
HiViG: a history-aware, visually grounded critic for computer-use agents
T1-Bench: Benchmarking Multi-Scenario Agents in Real-World Domains
T1-Bench: high-fidelity evaluation of multi-domain agents across 25 domains
What Fits (Into Few Tokens) Doesn't Overfit: Compression and Generalization in ML Research Agents
Compressibility explains why ML research agents rarely overfit
Workflow-GYM: long-horizon GUI tasks in professional software
AuRA: Internalizing Audio Understanding into LLMs as LoRA
AuRA: internalizing audio understanding into LLMs via LoRA
Diffusion Forcing Planner: history-guided diffusion planning for driving
A plain-language guide to OpenClaw risks for non-technical users
Mind the Gap: Can Frontier LLMs Pass a Standardized Office Proficiency Exam?
Frontier LLMs score at most 36.6% on a standardized Office proficiency exam
CLP: Collocation-Length Prediction for Zero-Loss Adaptive Multi-Token Inference
CLP: backbone-as-architect for zero-loss adaptive multi-token inference
Frontier Coding Agents Use Metaprogramming to Adapt to Unfamiliar Programming Languages
Frontier coding agents use metaprogramming for esoteric languages
Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution
Role-Agent: a single LLM as both agent and environment for co-evolution
Range Penalization: Theoretical Insights with Applications in Federated Learning
Range penalization for quantization-friendly federated learning
What Do Deepfake Speech Detectors Actually Hear?
What deepfake speech detectors actually hear, localized over time
Ethical and Technical Limits of Deepfake Speech Datasets
Auditing 39 deepfake speech datasets reveals fairness gaps and overlap
RAT: Reference-Augmented Training for ASV Anti-Spoofing
RAT: reference-augmented training sets a new ASV anti-spoofing SOTA
Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation
Boosting LLM tool calling via experiential knowledge integration
ConvMemory v2: A Recall-Preserving Top-10 Evidence Reranker for Conversational Memory Retrieval
ConvMemory v2: a recall-preserving Top-10 reranker for conversational memory
Attention-Discounted Adaptive Sampler for Masked Diffusion Language Models
ADAS: attention-discounted reranking for masked diffusion decoding
K-Forcing: Joint Next-K-Token Decoding via Push-Forward Language Modeling
K-Forcing: joint next-k-token decoding via push-forward language modeling
Recovering the Zipfian Distribution in Unsupervised Term Discovery
Graph clustering beats K-means in unsupervised term discovery
Predictor-gated sparsity recipe upcycles dense LLMs to sparse
Attention expansion boosts keyphrase extraction from long docs
REAL: A Reasoning-Enhanced Graph Framework for Long-Term Memory Management of LLMs
REAL manages LLM long-term memory with reasoning-enhanced graphs
Infini Memory: Maintainable Topic Documents for Long-Term LLM Agent Memory
Infini Memory: topic documents for long-term LLM agent memory
Learned DP decoder advances multilingual forced alignment
Speaker Group Encoding in Self-supervised Speech Recognition Models
How self-supervised speech models encode speaker group traits
ParaBridge: Bridging Paralinguistic Perception and Dialogue Behavior in Speech Language Models
ParaBridge links paralinguistic cues to dialogue behavior