https://chiosa.app/papers/beyond-the-transcript-measuring-and-improving-the-use-of-acoustic-information-in-audio-language-models-2611-11573Chiosaen2026-11-10T21:03:32ZA method measures and improves audio models’ use of soundhttps://chiosa.app/papers/bridgekv-large-model-prefill-and-small-model-decoding-for-llm-inference-via-kv-cache-translation-2611-11761Chiosaen2026-11-10T21:03:32ZA larger model can hand prompt work to a smaller onehttps://chiosa.app/papers/cascode-a-closed-loop-agentic-framework-for-analog-sizing-and-layout-co-design-2611-11710Chiosaen2026-11-10T21:03:32ZFeedback-guided design made all 20 amplifier layouts clean and smallerhttps://chiosa.app/papers/chaincheck-near-perfect-evidence-sufficiency-scores-can-conceal-sensitivity-to-construction-artifacts-paired-2611-11965Chiosaen2026-11-10T21:03:32ZA high evidence score can reflect text editshttps://chiosa.app/papers/decoupled-asynchronous-system-for-humanoids-dash-latency-agnostic-scaling-of-vision-language-action-models-fo-2611-11854Chiosaen2026-11-10T21:03:32ZDASH keeps humanoid control fast as models growhttps://chiosa.app/papers/forgesystem-an-architecture-for-demand-specific-artefact-production-with-ai-agents-2611-12288Chiosaen2026-11-10T21:03:32ZAI agents can build specialized software and verify its qualityhttps://chiosa.app/papers/how-much-training-inference-mismatch-can-on-policy-distillation-tolerate-2611-11450Chiosaen2026-11-10T21:03:32ZOne distillation setup tolerated changes that broke reward traininghttps://chiosa.app/papers/infrarsi-harnessing-recursive-self-improvement-for-training-infrastructure-2611-11558Chiosaen2026-11-10T21:03:32ZA harness improves distributed model training speedhttps://chiosa.app/papers/language-model-capabilities-fail-in-a-shared-predictable-order-under-compression-and-perturbation-2611-11695Chiosaen2026-11-10T21:03:32ZA profile predicts which model abilities weaken firsthttps://chiosa.app/papers/longtail-swe-high-quality-swe-tasks-for-underrepresented-languages-2611-10304Chiosaen2026-11-10T21:03:32ZA repair pipeline recovers software issue tasks across 13 languageshttps://chiosa.app/papers/motion-belief-action-from-a-visual-scene-map-to-physical-decisions-2611-12113Chiosaen2026-11-10T21:03:32ZA robot map links movement and force to actionhttps://chiosa.app/papers/nemotron-3-diarization-advancing-the-accuracy-efficiency-frontier-in-streaming-speaker-diarization-2611-12246Chiosaen2026-11-10T21:03:32ZAn eight-speaker model lowers diarization errors across delayshttps://chiosa.app/papers/onetrans-gsr-a-generative-slate-ranking-paradigm-unifying-reranking-and-mix-ranking-with-one-autoregressive-t-2611-04720Chiosaen2026-11-10T21:03:32ZOne learned system selects and orders recommendation slateshttps://chiosa.app/papers/pace-ecg-patient-aware-contrastive-ecg-pretraining-leveraging-longitudinal-structure-for-transferable-represe-2611-12307Chiosaen2026-11-10T21:03:32ZECG learning from one patient’s recordings transfers betterhttps://chiosa.app/papers/pcbrouteagent-how-close-are-llm-agents-to-becoming-pcb-routing-engineers-2611-08175Chiosaen2026-11-10T21:03:32ZPCBRouteAgent coordinates plans with physical PCB routinghttps://chiosa.app/papers/prune-don-t-merge-expert-compression-of-mixture-of-experts-llms-on-multilingual-production-tasks-2611-12240Chiosaen2026-11-10T21:03:32ZPruning preserves production-task quality better than expert merginghttps://chiosa.app/papers/reprobridge-building-reproducibility-into-ml-research-artifacts-2611-08452Chiosaen2026-11-10T21:03:32ZReproBridge turns recovered experiment steps into reusable guideshttps://chiosa.app/papers/reproducible-quantum-clustering-for-single-cell-transcriptome-atlases-2611-11745Chiosaen2026-11-10T21:03:32ZKQE-2 clusters cell atlases reproducibly at large scalehttps://chiosa.app/papers/watching-itself-watch-self-auditing-visual-reliance-for-video-reasoning-2611-11949Chiosaen2026-11-10T21:03:32ZA training method helps video models use visual evidencehttps://chiosa.app/papers/worldfoundry-towards-unified-infrastructure-for-world-intelligence-2611-10771Chiosaen2026-11-10T21:03:32ZWorldFoundry makes world-model testing easier to run and audithttps://chiosa.app/papers/a-local-language-model-layered-with-rule-based-and-neural-detectors-for-clinical-text-de-identification-an-op-2611-12709Chiosaen2026-11-10T21:02:00ZA local detector stack found more PHI in clinical noteshttps://chiosa.app/papers/mitigating-agentic-deception-via-evaluation-side-partially-observable-games-2611-12835Chiosaen2026-11-10T21:02:00ZA verifier helps monitors catch deceptive agent actionshttps://chiosa.app/papers/certify-report-or-refuse-harm-constrained-conformal-prediction-for-clinical-actions-2611-12662Chiosaen2026-11-10T14:40:02ZA medical imaging method checks harmful actions as well as diagnosis coveragehttps://chiosa.app/papers/distilling-instructions-into-weights-closed-form-edits-fix-a-robot-policy-s-misreading-2611-11894Chiosaen2026-11-10T14:40:02ZA robot’s misread instruction can be written into its weightshttps://chiosa.app/papers/geoscope-training-free-program-execution-beyond-the-counting-horizon-for-ultra-high-resolution-remote-sensing-2611-12699Chiosaen2026-11-10T14:40:02ZSplitting crowded views helps remote sensing models counthttps://chiosa.app/papers/learning-heterogeneous-tactile-representation-for-general-dexterous-manipulation-2611-12806Chiosaen2026-11-10T14:40:02ZA shared touch model improves robot task successhttps://chiosa.app/papers/opticover-a-scalable-and-multi-surface-framework-for-cold-start-visual-optimization-under-semantic-constraint-2611-12727Chiosaen2026-11-10T14:40:02ZA system matches Airbnb cover photos to guest intenthttps://chiosa.app/papers/recogem-a-serving-system-for-production-scale-generative-recommendation-2611-12296Chiosaen2026-11-10T14:40:02ZStep-aware serving speeds up generative recommendationhttps://chiosa.app/papers/reframe-observation-aligned-multi-teacher-on-policy-distillation-for-agentic-visual-reasoning-2611-12886Chiosaen2026-11-10T14:40:02ZVisual agents can learn to use evidence returned by toolshttps://chiosa.app/papers/sage-taiyi-a-metacognitive-architecture-for-scientific-discovery-2611-12762Chiosaen2026-11-10T14:40:02ZSage-Taiyi uses evidence to guide scientific searchhttps://chiosa.app/papers/suppression-is-not-erasure-mechanistic-analysis-and-reactivation-boundaries-of-subspace-projection-for-lora-b-2611-12807Chiosaen2026-11-10T14:40:02ZProjection can silence LoRA backdoors without erasing themhttps://chiosa.app/papers/translating-kv-caches-between-hybrid-attention-language-models-mla-latents-linear-attention-states-and-sparse-2611-12062Chiosaen2026-11-10T14:40:02ZCache translation can speed model handoffshttps://chiosa.app/papers/treating-the-model-as-an-untrusted-component-deterministic-policy-enforcement-human-approval-and-taint-tracki-2611-11571Chiosaen2026-11-10T14:40:02ZA task-bound monitor limits injected actions in agentshttps://chiosa.app/papers/why-kubernetes-cannot-contain-a-full-lazy-pulling-cache-two-impossibility-results-and-a-bounded-honest-design-2611-11739Chiosaen2026-11-10T14:40:02ZA lazy-pulling cache keeps full disks from breaking model readshttps://chiosa.app/papers/world-in-context-learning-for-agentic-robot-manipulation-2611-12885Chiosaen2026-11-10T14:40:02ZRobots learn tasks through simulator feedback in contexthttps://chiosa.app/papers/llms-don-t-know-what-they-know-benchmarking-temporal-self-knowledge-for-financial-backtesting-2611-06808Chiosaen2026-11-09T19:24:03ZModels can claim ignorance while recalling past financial factshttps://chiosa.app/papers/confucius4-r2t2-a-stable-prefix-streaming-asr-model-for-real-real-time-transcription-2611-06043Chiosaen2026-11-09T13:50:17ZA speech recognizer learns to keep committed words unchangedhttps://chiosa.app/papers/fast-and-exact-relative-position-attention-via-structure-aware-kernels-2611-01182Chiosaen2026-11-09T13:50:17ZTiled attention cuts costs of relative-position termshttps://chiosa.app/papers/longbanana-an-expert-verified-benchmark-for-long-context-multi-reference-image-synthesis-2611-06856Chiosaen2026-11-09T13:50:17ZA checklist makes multi-reference image errors easier to locatehttps://chiosa.app/papers/p-2-mopd-accelerating-multi-teacher-on-policy-distillation-via-parameter-priors-2611-03953Chiosaen2026-11-09T13:50:17ZA trainable expert blend helps a student recover domain skillshttps://chiosa.app/papers/video-coder-reinforcing-agentic-coding-for-video-generation-2611-09894Chiosaen2026-11-09T13:50:17ZA coding agent guides video generation with 3D sceneshttps://chiosa.app/papers/beyond-truncation-entailment-licensed-elision-and-the-limits-of-semantic-preservation-in-llm-decoding-2611-08975Chiosaen2026-11-09T09:25:23ZShorter answers often stop early instead of removing repetitionhttps://chiosa.app/papers/bitdecoding-fp4-native-fp4-kv-cache-decoding-via-algorithm-system-co-design-2611-09825Chiosaen2026-11-09T09:25:23ZFP4 cache decoding keeps accuracy close while raising serving capacityhttps://chiosa.app/papers/cancellation-is-not-enough-what-retries-leave-behind-in-an-overloaded-llm-engine-2611-08684Chiosaen2026-11-09T09:25:23ZClient retries can keep an overloaded language model stuckhttps://chiosa.app/papers/chainflow-few-step-latent-flow-matching-for-speculative-decoding-2611-08669Chiosaen2026-11-09T09:25:23ZA new drafter speeds up verified language-model generationhttps://chiosa.app/papers/copies-are-not-confirmations-how-news-duplication-fools-llm-judges-and-continual-learners-2611-09203Chiosaen2026-11-09T09:25:23ZCopied news can sway some language-model judgeshttps://chiosa.app/papers/deepradio-a-human-agent-workspace-for-physical-radio-development-2611-09418Chiosaen2026-11-09T09:25:23ZDeepRadio helps people build and verify physical radioshttps://chiosa.app/papers/foscar-accelerating-decode-attention-over-2-and-4-bit-kv-caches-through-affine-folding-2611-09722Chiosaen2026-11-09T09:25:23ZFOSCAR speeds attention by folding quantization into matrix productshttps://chiosa.app/papers/fused-softmax-for-efficient-quantized-attention-2611-09225Chiosaen2026-11-09T09:25:23ZFused softmax makes quantized attention faster on AMD GPUshttps://chiosa.app/papers/global-solar-siting-from-geospatial-embeddings-2611-08831Chiosaen2026-11-09T09:25:23ZA global map learns which land may suit solarhttps://chiosa.app/papers/on-agentic-manipulation-policies-2611-09892Chiosaen2026-11-09T09:25:23ZA generalist model adapts while controlling a robothttps://chiosa.app/papers/photofinish-portable-timerless-probing-of-microarchitectural-latencies-2611-09879Chiosaen2026-11-09T09:25:23ZA shared-memory race measures latency without a clockhttps://chiosa.app/papers/real-world-adoption-use-and-governance-of-an-llm-based-electronic-health-record-search-and-synthesis-platform-2611-08672Chiosaen2026-11-09T09:25:23ZDuke Scout logs show how clinicians used an EHR toolhttps://chiosa.app/papers/s-3-simple-scaling-of-swarms-via-cooperative-multi-agent-rl-2611-08615Chiosaen2026-11-09T09:25:23ZLanguage-model teams learn to coordinate through voluntary messageshttps://chiosa.app/papers/scale-as-a-modality-multi-scale-pixel-space-diffusion-transformers-2611-08802Chiosaen2026-11-09T09:25:23ZA pixel-space image model gains from representing more scaleshttps://chiosa.app/papers/splatanything-gaussian-spacing-normalization-for-feed-forward-3dgs-across-arbitrary-scenes-2611-08264Chiosaen2026-11-09T09:25:23ZGaussian spacing helps 3D scenes render across scaleshttps://chiosa.app/papers/wearablebridge-task-guided-latent-transport-for-free-living-activity-recognition-2611-09067Chiosaen2026-11-09T09:25:23ZA small bridge improves activity recognition from wrist datahttps://chiosa.app/papers/when-to-trust-the-bim-identifiability-aware-lidar-inertial-bim-fusion-under-as-built-deviations-2611-09045Chiosaen2026-11-09T09:25:23ZA robot method waits when scans cannot explain a model conflict