https://blog.schoergendorfer.com/post/anthropic-opens-a-research-preview-of-the-model-hardware-standard-mhs-a-shared-sSecFeedde2026-08-30T06:17:22.138434+00:00Anthropic Opens a Research Preview of the Model Hardware Standard (MHS): A Shared Specification for AI Agents to Safely Operate Physical Deviceshttps://blog.schoergendorfer.com/post/meet-code-as-world-an-agentic-loop-that-rewrites-real-videos-into-executable-mujSecFeedde2026-08-30T01:48:26.242273+00:00Meet ‘Code-as-World’: An Agentic Loop That Rewrites Real Videos Into Executable MuJoCo Physics Programshttps://blog.schoergendorfer.com/post/introducing-hy4-previewSecFeedde2026-08-30T00:17:20.747905+00:00Introducing Hy4 Previewhttps://blog.schoergendorfer.com/post/building-custom-batched-ensemble-weather-forecasting-with-nvidia-earth2studioSecFeedde2026-08-29T19:03:26.463743+00:00Building Custom Batched Ensemble Weather Forecasting with NVIDIA Earth2Studiohttps://blog.schoergendorfer.com/post/deadlock-ransomware-breaking-down-a-rust-based-encryptor-with-decentralized-reco-2SecFeedde2026-08-29T15:48:14.570171+00:00DeadLock ransomware: Breaking down a Rust-based encryptor with decentralized recovery infrastructurehttps://blog.schoergendorfer.com/post/hunting-macsync-stealer-infrastructure-through-behavioral-pivots-2SecFeedde2026-08-29T15:48:14.523306+00:00Hunting MacSync Stealer infrastructure through behavioral pivotshttps://blog.schoergendorfer.com/post/when-ai-infrastructure-becomes-the-target-securing-gateways-and-control-points-2SecFeedde2026-08-29T15:48:14.483411+00:00When AI infrastructure becomes the target: Securing gateways and control pointshttps://blog.schoergendorfer.com/post/your-container-runs-everything-around-it-shouldnt-be-your-problemSecFeedde2026-08-29T15:17:50.257164+00:00Your container runs. Everything around it shouldn’t be your problem.https://blog.schoergendorfer.com/post/google-ai-releases-gemini-omni-11-flash-40-second-scene-extension-firstlast-framSecFeedde2026-08-29T14:47:40.136537+00:00Google AI Releases Gemini Omni 1.1 Flash: 40-Second Scene Extension, First/Last Frame Control, and 4K Upscalinghttps://blog.schoergendorfer.com/post/commits-on-github-have-doubled-in-four-months-verification-capacity-has-notSecFeedde2026-08-29T14:02:59.106180+00:00Commits on GitHub have doubled in four months. Verification capacity has not.https://blog.schoergendorfer.com/post/the-3-roles-ai-agents-play-in-your-developer-platformSecFeedde2026-08-29T13:02:36.906235+00:00The 3 roles AI agents play in your developer platformhttps://blog.schoergendorfer.com/post/hugging-face-unveils-microduck-a-399-open-source-25-cm-biped-you-train-with-reinSecFeedde2026-08-29T05:32:22.979501+00:00Hugging Face Unveils Microduck: A $399 Open-Source 25 cm Biped You Train with Reinforcement Learninghttps://blog.schoergendorfer.com/post/ainews-openai-shuts-off-cursorSecFeedde2026-08-29T05:17:40.209566+00:00[AINews] OpenAI shuts off Cursorhttps://blog.schoergendorfer.com/post/terminalfix-campaign-deploys-a-reverse-tunnel-through-multistage-intrusionSecFeedde2026-08-29T05:17:40.139774+00:00TerminalFix campaign deploys a reverse tunnel through multistage intrusionhttps://blog.schoergendorfer.com/post/kubernetes-v137-pod-certificates-and-cluster-trust-bundlesSecFeedde2026-08-29T01:02:36.312902+00:00Kubernetes v1.37: Pod Certificates and Cluster Trust Bundleshttps://blog.schoergendorfer.com/post/amazon-bedrock-agentcore-memory-now-supports-fine-grained-access-controlSecFeedde2026-08-28T23:17:43.995060+00:00Amazon Bedrock AgentCore Memory now supports fine-grained access controlhttps://blog.schoergendorfer.com/post/amazon-ec2-c8gn-instances-are-now-available-in-aws-europe-paris-regionSecFeedde2026-08-28T23:17:43.945064+00:00Amazon EC2 C8gn instances are now available in AWS Europe (Paris) regionhttps://blog.schoergendorfer.com/post/just-a-rumour-of-a-bug-is-enough-to-find-a-security-exploit-these-daysSecFeedde2026-08-28T22:32:24.227518+00:00Just a rumour of a bug is enough to find a security exploit these dayshttps://blog.schoergendorfer.com/post/jetbrains-told-everyone-to-patch-it-didnt-patch-itselfSecFeedde2026-08-28T21:02:33.810821+00:00JetBrains told everyone to patch. It didn’t patch itself.https://blog.schoergendorfer.com/post/cybersecurity-researchers-uncover-flaw-in-google-ai-coding-toolSecFeedde2026-08-28T19:48:02.410368+00:00Cybersecurity Researchers Uncover Flaw in Google AI Coding Toolhttps://blog.schoergendorfer.com/post/batch-write-and-discover-records-in-amazon-sagemaker-feature-storeSecFeedde2026-08-28T19:48:02.372829+00:00Batch write and discover records in Amazon SageMaker Feature Storehttps://blog.schoergendorfer.com/post/amazon-cloudwatch-agent-adds-support-for-journald-logsSecFeedde2026-08-28T19:19:08.503022+00:00Amazon CloudWatch agent adds support for journald logshttps://blog.schoergendorfer.com/post/amazon-ec2-p6-b300-instances-are-now-available-in-additional-aws-regionsSecFeedde2026-08-28T19:19:08.468360+00:00Amazon EC2 P6-B300 instances are now available in additional AWS Regionshttps://blog.schoergendorfer.com/post/glm-53-flash-vs-qwen38-flash-next-two-chinese-ai-labs-independently-converge-onSecFeedde2026-08-28T19:19:08.442446+00:00GLM-5.3-Flash vs Qwen3.8-Flash-Next: Two Chinese AI Labs Independently Converge on the Same Model Architecturehttps://blog.schoergendorfer.com/post/extend-your-data-perimeter-to-the-aws-management-console-with-private-accessSecFeedde2026-08-28T19:02:50.586723+00:00Extend your data perimeter to the AWS Management Console with Private Accesshttps://blog.schoergendorfer.com/post/lm-studio-built-a-judge-for-ai-commands-then-the-judge-started-agreeing-with-theSecFeedde2026-08-28T18:48:52.729617+00:00LM Studio built a judge for AI commands. Then the judge started agreeing with the defendant.https://blog.schoergendorfer.com/post/alibaba-just-released-qwen38-flash-an-early-preview-of-the-architecture-in-qwen4SecFeedde2026-08-28T18:33:04.061403+00:00Alibaba just released Qwen3.8-Flash: “An early preview of the architecture in Qwen4”https://blog.schoergendorfer.com/post/certificate-renewal-is-a-deployment-workflow-not-a-cron-jobSecFeedde2026-08-28T18:03:51.366829+00:00Certificate Renewal Is a Deployment Workflow, Not a Cron Jobhttps://blog.schoergendorfer.com/post/zais-glm-53-goes-open-weight-but-its-new-license-aims-at-hyperscalersSecFeedde2026-08-28T17:48:03.109793+00:00Z.ai’s GLM-5.3 goes open weight, but its new license aims at hyperscalershttps://blog.schoergendorfer.com/post/vercel-ai-open-sources-vgpu-a-typescript-webgpu-library-for-ai-agent-shadersSecFeedde2026-08-28T17:48:03.071679+00:00Vercel AI Open-Sources vgpu: A TypeScript WebGPU Library for AI Agent Shadershttps://blog.schoergendorfer.com/post/amazon-aurora-mysql-313-compatible-with-mysql-8045-is-generally-availableSecFeedde2026-08-28T17:17:56.365258+00:00Amazon Aurora MySQL 3.13 (compatible with MySQL 8.0.45) is generally availablehttps://blog.schoergendorfer.com/post/spreading-the-load-how-salesforce-met-multi-az-ha-with-sagemaker-inference-compoSecFeedde2026-08-28T16:34:41.037289+00:00Spreading the load: How Salesforce met Multi-AZ HA with SageMaker Inference Componentshttps://blog.schoergendorfer.com/post/how-decathlon-runs-demand-forecasting-at-scale-with-chronos-2SecFeedde2026-08-28T16:34:40.998467+00:00How Decathlon runs demand forecasting at scale with Chronos-2https://blog.schoergendorfer.com/post/what-does-it-cost-to-get-kubernetes-operations-wrongSecFeedde2026-08-28T16:25:20.953277+00:00What Does It Cost to Get Kubernetes Operations Wrong?https://blog.schoergendorfer.com/post/nvidia-is-paying-129-billion-to-keep-open-models-on-its-chipsSecFeedde2026-08-28T16:04:08.432244+00:00Nvidia is paying $12.9 billion to keep open models on its chipshttps://blog.schoergendorfer.com/post/why-a-cryptographic-inventory-is-key-for-addressing-the-quantum-computing-threatSecFeedde2026-08-28T14:18:29.362230+00:00Why a cryptographic inventory is key for addressing the quantum computing threathttps://blog.schoergendorfer.com/post/build-your-own-continuous-modernization-pipeline-with-aws-transform-customSecFeedde2026-08-28T13:36:13.414711+00:00Build your own continuous modernization pipeline with AWS Transform customhttps://blog.schoergendorfer.com/post/botbase-for-operators-a-clearer-path-to-joining-cloudflares-directory-of-bots-anSecFeedde2026-08-28T13:36:13.265550+00:00BotBase for Operators: A clearer path to joining Cloudflare's directory of bots and agentshttps://blog.schoergendorfer.com/post/scale-before-the-spike-predictive-autoscaling-for-gpu-workloads-on-kubernetesSecFeedde2026-08-28T13:21:24.547152+00:00Scale before the spike: Predictive autoscaling for GPU workloads on Kuberneteshttps://blog.schoergendorfer.com/post/ai-doesnt-mean-the-end-of-mathematicsat-least-not-yetSecFeedde2026-08-28T13:21:24.443143+00:00AI Doesn’t Mean the End of Mathematics—at Least Not Yethttps://blog.schoergendorfer.com/post/your-kubernetes-platform-is-ready-for-containers-is-it-ready-for-aiSecFeedde2026-08-28T13:21:24.178443+00:00Your Kubernetes platform is ready for containers. Is it ready for AI?https://blog.schoergendorfer.com/post/papercut-ngmf-critical-zero-day-exploited-in-the-wildSecFeedde2026-08-28T13:21:22.835899+00:00PaperCut NG/MF Critical Zero-Day Exploited in the Wildhttps://blog.schoergendorfer.com/post/ainews-openai-to-reach-agi-bar-by-end-2026SecFeedde2026-08-28T12:49:41.387972+00:00[AINews] OpenAI to reach AGI bar by end-2026https://blog.schoergendorfer.com/post/some-malicious-pe-stats-thu-aug-27thSecFeedde2026-08-28T12:34:32.161476+00:00Some Malicious PE Stats, (Thu, Aug 27th)https://blog.schoergendorfer.com/post/google-ai-releases-gemini-35-transcribe-a-speech-to-text-model-reporting-26-averSecFeedde2026-08-28T11:50:18.907044+00:00Google AI Releases Gemini 3.5 Transcribe: A Speech-to-Text Model Reporting 2.6% Average WER Across 85+ Languageshttps://blog.schoergendorfer.com/post/managed-postgresql-vs-self-hosted-postgresql-key-benefits-and-trade-offsSecFeedde2026-08-28T11:50:18.827480+00:00Managed PostgreSQL vs. self-hosted PostgreSQL: Key benefits and trade-offshttps://blog.schoergendorfer.com/post/et-prune-evidence-aware-dynamic-budgeting-for-visual-token-pruning-in-text-rich-2SecFeedde2026-08-28T11:34:43.528280+00:00ET-Prune: Evidence-Aware Dynamic Budgeting for Visual Token Pruning in Text-Rich MLLMshttps://blog.schoergendorfer.com/post/culturevidbench-benchmarking-cultural-understanding-in-text-to-video-generation-2SecFeedde2026-08-28T11:34:43.494145+00:00CultureVidBench: Benchmarking Cultural Understanding in Text-to-Video Generationhttps://blog.schoergendorfer.com/post/not-truly-multilingual-script-consistency-as-a-missing-dimension-in-vlm-evaluati-2SecFeedde2026-08-28T11:34:43.460433+00:00Not Truly Multilingual: Script Consistency as a Missing Dimension in VLM Evaluationhttps://blog.schoergendorfer.com/post/an-embarrassingly-simple-detector-for-model-extraction-attacks-in-large-languageSecFeedde2026-08-28T11:34:43.427648+00:00An Embarrassingly Simple Detector for Model Extraction Attacks in Large Language Model API Traffichttps://blog.schoergendorfer.com/post/visa-agentic-self-evolving-data-synthesis-for-multimodal-instruction-following-2SecFeedde2026-08-28T11:34:43.394213+00:00VISA: Agentic Self-Evolving Data Synthesis for Multimodal Instruction Followinghttps://blog.schoergendorfer.com/post/clueweaver-reward-guided-dual-agent-evidence-reasoning-for-compact-llms-on-liter-2SecFeedde2026-08-28T11:34:43.360984+00:00ClueWeaver: Reward-Guided Dual-Agent Evidence Reasoning for Compact LLMs on Literary Long Narrativeshttps://blog.schoergendorfer.com/post/unveiling-spectral-mechanisms-in-training-free-llm-text-detection-2SecFeedde2026-08-28T11:34:43.324206+00:00Unveiling Spectral Mechanisms in Training-Free LLM Text Detectionhttps://blog.schoergendorfer.com/post/fireredaudio-a-general-purpose-audio-language-model-with-decoupled-continuous-reSecFeedde2026-08-28T11:34:43.295490+00:00FireRedAudio: A General-Purpose Audio Language Model with Decoupled Continuous Representations for Understanding and Generationhttps://blog.schoergendorfer.com/post/aligned-alone-misaligned-together-forecasting-adversarial-capture-in-llm-agent-p-2SecFeedde2026-08-28T11:34:43.262879+00:00Aligned Alone, Misaligned Together: Forecasting Adversarial Capture in LLM Agent Populationshttps://blog.schoergendorfer.com/post/gui-primitives-diagnosing-spatial-reasoning-failures-in-vision-language-gui-grou-2SecFeedde2026-08-28T11:34:43.232760+00:00GUI-Primitives: Diagnosing Spatial Reasoning Failures in Vision-Language GUI Groundinghttps://blog.schoergendorfer.com/post/training-agents-to-evolve-with-their-harness-taolive-digital-avatar-agent-techni-2SecFeedde2026-08-28T11:34:43.200205+00:00Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical Reporthttps://blog.schoergendorfer.com/post/why-knowing-both-hops-is-not-enough-understanding-two-hop-generalization-in-lang-2SecFeedde2026-08-28T11:34:43.165653+00:00Why Knowing Both Hops Is Not Enough: Understanding Two-Hop Generalization in Language Modelshttps://blog.schoergendorfer.com/post/selective-state-space-adaptation-and-retrieval-for-language-model-reasoning-2SecFeedde2026-08-28T11:34:43.134262+00:00Selective State-Space Adaptation and Retrieval for Language Model Reasoninghttps://blog.schoergendorfer.com/post/the-score-granularity-gap-in-black-box-llm-classification-a-comparative-study-ofSecFeedde2026-08-28T11:34:43.097854+00:00The Score Granularity Gap in Black-Box LLM Classification: A Comparative Study of Confidence Constructionshttps://blog.schoergendorfer.com/post/non-binary-bottom-up-constituency-parsing-without-arity-actions-2SecFeedde2026-08-28T11:34:43.061747+00:00Non-binary bottom-up constituency parsing without arity actionshttps://blog.schoergendorfer.com/post/skillchain-closing-the-loop-on-skill-evolution-for-image-based-e-commerce-ai-assSecFeedde2026-08-28T11:34:43.020008+00:00SkillChain: Closing the Loop on Skill Evolution for Image-Based E-Commerce AI Assistantshttps://blog.schoergendorfer.com/post/forward-free-diffusion-language-models-with-bptt-free-looped-refinementSecFeedde2026-08-28T11:20:22.090181+00:00Forward-Free Diffusion Language Models with BPTT-Free Looped Refinementhttps://blog.schoergendorfer.com/post/the-cold-start-safety-gap-in-llm-agentsSecFeedde2026-08-28T11:20:22.061362+00:00The Cold-Start Safety Gap in LLM Agentshttps://blog.schoergendorfer.com/post/many-circuits-one-mechanism-input-variation-and-evaluation-granularity-in-circuiSecFeedde2026-08-28T11:20:22.037274+00:00Many Circuits, One Mechanism: Input Variation and Evaluation Granularity in Circuit Discoveryhttps://blog.schoergendorfer.com/post/cram-centroid-routing-and-adaptive-moe-for-multimodal-continual-instruction-tuniSecFeedde2026-08-28T11:20:22.004816+00:00CRAM: Centroid-Routing and Adaptive MoE for Multimodal Continual Instruction Tuninghttps://blog.schoergendorfer.com/post/mineexplorer-evaluating-open-world-exploration-of-mllm-agents-in-minecraftSecFeedde2026-08-28T11:20:21.970651+00:00MineExplorer: Evaluating Open-World Exploration of MLLM Agents in Minecrafthttps://blog.schoergendorfer.com/post/nestedkv-nested-memory-routing-for-long-context-kv-cache-compressionSecFeedde2026-08-28T11:20:21.938681+00:00NestedKV: Nested Memory Routing for Long-Context KV Cache Compressionhttps://blog.schoergendorfer.com/post/contextecho-a-benchmark-for-persona-drift-in-long-agentic-coding-sessionsSecFeedde2026-08-28T11:20:21.904247+00:00ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessionshttps://blog.schoergendorfer.com/post/traces-tagging-reasoning-steps-for-adaptive-cost-efficient-early-stoppingSecFeedde2026-08-28T11:20:21.868629+00:00TRACES: Tagging Reasoning Steps for Adaptive Cost-Efficient Early-Stoppinghttps://blog.schoergendorfer.com/post/datastorm-deep-research-on-large-scale-databases-using-exploratory-data-analysis-2SecFeedde2026-08-28T11:20:21.835470+00:00DataSTORM: Deep Research on Large-Scale Databases using Exploratory Data Analysis and Data Storytellinghttps://blog.schoergendorfer.com/post/addressing-the-reasoning-gap-mechanistic-circuit-based-knowledge-editing-in-larg-2SecFeedde2026-08-28T11:20:21.796548+00:00Addressing the Reasoning Gap: Mechanistic Circuit-Based Knowledge Editing in Large Language Modelshttps://blog.schoergendorfer.com/post/many-dialects-many-languages-one-cultural-lens-evaluating-multilingual-vlms-forSecFeedde2026-08-28T11:20:21.727738+00:00Many Dialects, Many Languages, One Cultural Lens: Evaluating Multilingual VLMs for Bengali Culture Understanding Across Historically Linked Languages and Regional Dialectshttps://blog.schoergendorfer.com/post/spar-k-scheduled-periodic-alternating-early-exit-for-spoken-language-modelsSecFeedde2026-08-28T11:20:21.697887+00:00SPAR-K: Scheduled Periodic Alternating Early Exit for Spoken Language Modelshttps://blog.schoergendorfer.com/post/maple-metadata-conditioned-llm-pretraining-for-locale-aware-question-answeringSecFeedde2026-08-28T11:20:21.629905+00:00MAPLE: Metadata Conditioned LLM Pretraining for Locale-Aware Question Answeringhttps://blog.schoergendorfer.com/post/topiclear-adaptive-embedding-clustering-for-interpretable-topic-discovery-from-sSecFeedde2026-08-28T11:20:21.592923+00:00TopiCLEAR: Adaptive embedding clustering for interpretable topic discovery from short textshttps://blog.schoergendorfer.com/post/are-large-language-models-effective-knowledge-graph-constructorsSecFeedde2026-08-28T11:20:21.551851+00:00Are Large Language Models Effective Knowledge Graph Constructors?https://blog.schoergendorfer.com/post/legal-rule-induction-towards-generalizable-principle-discovery-from-analogous-juSecFeedde2026-08-28T11:20:21.518130+00:00Legal Rule Induction: Towards Generalizable Principle Discovery from Analogous Judicial Precedentshttps://blog.schoergendorfer.com/post/arams-28k-the-largest-publicly-released-line-level-dataset-of-historical-arabicSecFeedde2026-08-28T11:20:21.478713+00:00AraMS-28k: The Largest Publicly Released Line-Level Dataset of Historical Arabic Manuscripts with Margin and Insertion-Anchor Annotationshttps://blog.schoergendorfer.com/post/blanc-discovering-patent-white-space-via-changes-in-normalized-pointwise-mutualSecFeedde2026-08-28T11:20:21.430336+00:00BLANC: Discovering Patent White Space via Changes in Normalized Pointwise Mutual Information Between Multi-View Clustershttps://blog.schoergendorfer.com/post/visual-information-guided-parallel-decoding-for-diffusion-multimodal-large-languSecFeedde2026-08-28T11:05:32.027333+00:00Visual Information-Guided Parallel Decoding for Diffusion Multimodal Large Language Modelshttps://blog.schoergendorfer.com/post/assessing-the-downstream-utility-of-evidence-aware-retrieval-in-ragSecFeedde2026-08-28T11:05:31.995341+00:00Assessing the Downstream Utility of Evidence-Aware Retrieval in RAGhttps://blog.schoergendorfer.com/post/mitigating-fabrication-in-multi-stage-llm-pipelines-for-hiring-an-empirical-evalSecFeedde2026-08-28T11:05:31.966218+00:00Mitigating Fabrication in Multi-Stage LLM Pipelines for Hiring: An Empirical Evaluation of Prompt Guardrails and Human-in-the-Loop Checkpointshttps://blog.schoergendorfer.com/post/criticl-inference-time-weak-to-strong-generalization-from-small-language-model-fSecFeedde2026-08-28T11:05:31.937197+00:00CritICL: Inference-Time Weak-to-Strong Generalization from Small Language Model Failure Modeshttps://blog.schoergendorfer.com/post/ttpo-test-time-policy-optimizationSecFeedde2026-08-28T11:05:31.907529+00:00TTPO: Test-Time Policy Optimizationhttps://blog.schoergendorfer.com/post/stochastic-estimation-of-transduced-language-modelsSecFeedde2026-08-28T11:05:31.877248+00:00Stochastic Estimation of Transduced Language Modelshttps://blog.schoergendorfer.com/post/boosting-llm-exploration-via-weak-model-guidance-in-rlvrSecFeedde2026-08-28T11:05:31.846877+00:00Boosting LLM Exploration via Weak-Model Guidance in RLVRhttps://blog.schoergendorfer.com/post/consolidating-rlvr-capabilities-across-domains-a-deep-dive-into-fusion-paradigmsSecFeedde2026-08-28T11:05:31.815721+00:00Consolidating RLVR Capabilities Across Domains: A Deep Dive into Fusion Paradigmshttps://blog.schoergendorfer.com/post/ratio-a-benchmark-for-retrieval-across-typed-ideation-operations-in-scientific-lSecFeedde2026-08-28T11:05:31.783705+00:00RATIO: A Benchmark for Retrieval Across Typed Ideation Operations in Scientific Literaturehttps://blog.schoergendorfer.com/post/intent-as-a-tool-makes-it-easy-to-track-agentic-misalignmentSecFeedde2026-08-28T11:05:31.671483+00:00INTENT-AS-A-TOOL Makes it Easy to Track Agentic Misalignmenthttps://blog.schoergendorfer.com/post/pair-level-essay-scale-republication-and-reuse-from-fragmented-historical-text-rSecFeedde2026-08-28T11:05:31.642079+00:00Pair-Level Essay-Scale Republication and Reuse from Fragmented Historical Text Reuse: A Workflow Study on Eighteenth-Century Books and Newspapershttps://blog.schoergendorfer.com/post/bts-agentbench-a-deterministic-replayable-pipeline-from-read-only-telemetry-logsSecFeedde2026-08-28T11:05:31.604462+00:00BTS-AgentBench: A Deterministic, Replayable Pipeline from Read-Only Telemetry Logs to Agent Benchmarkshttps://blog.schoergendorfer.com/post/scit-testing-causal-cache-carriers-in-latent-chain-of-thought-modelsSecFeedde2026-08-28T11:05:31.577513+00:00SCIT: Testing Causal Cache Carriers in Latent Chain-of-Thought Modelshttps://blog.schoergendorfer.com/post/balms-benchmarking-agentic-llms-for-longitudinal-mental-health-sensingSecFeedde2026-08-28T11:05:31.549529+00:00BALMS: Benchmarking Agentic LLMs for Longitudinal Mental Health Sensinghttps://blog.schoergendorfer.com/post/prediction-of-prediction-pop-inter-layer-activation-fusion-for-single-pass-halluSecFeedde2026-08-28T11:05:31.522805+00:00Prediction of Prediction (PoP): Inter-Layer Activation Fusion for Single-Pass Hallucination Detection in Large Language Modelshttps://blog.schoergendorfer.com/post/twinkv-a-composable-repair-pass-for-kv-cache-eviction-via-pairwise-key-redundancSecFeedde2026-08-28T10:49:29.425225+00:00TwinKV: A Composable Repair Pass for KV Cache Eviction via Pairwise Key Redundancyhttps://blog.schoergendorfer.com/post/cross-lingual-alignment-without-joint-training-do-monolingual-language-models-coSecFeedde2026-08-28T10:49:29.397835+00:00Cross-Lingual Alignment Without Joint Training: Do Monolingual Language Models Converge on Universal Representations?https://blog.schoergendorfer.com/post/doctalkbn-a-novel-dataset-of-expert-telemedicine-conversations-in-bengaliSecFeedde2026-08-28T10:49:29.363919+00:00DocTalkBN: A Novel Dataset of Expert Telemedicine Conversations in Bengalihttps://blog.schoergendorfer.com/post/representing-and-parsing-korean-constituency-structure-at-different-levels-of-grSecFeedde2026-08-28T10:49:29.276243+00:00Representing and Parsing Korean Constituency Structure at Different Levels of Granularityhttps://blog.schoergendorfer.com/post/itl-interpretable-document-alignment-with-structured-reference-frameworksSecFeedde2026-08-28T10:49:29.244669+00:00ITL: Interpretable Document Alignment with Structured Reference Frameworkshttps://blog.schoergendorfer.com/post/judgestealer-extracting-llm-judging-capabilities-across-evaluation-protocolsSecFeedde2026-08-28T10:49:29.218911+00:00JudgeStealer: Extracting LLM Judging Capabilities across Evaluation Protocolshttps://blog.schoergendorfer.com/post/kinyaembed-contrastive-sentence-embeddings-for-kinyarwanda-via-multi-stage-curriSecFeedde2026-08-28T10:49:29.194840+00:00KinyaEmbed: Contrastive Sentence Embeddings for Kinyarwanda via Multi-Stage Curriculum Traininghttps://blog.schoergendorfer.com/post/mapping-written-words-to-spoken-words-in-a-different-language-using-only-visualSecFeedde2026-08-28T10:49:29.167238+00:00Mapping Written Words to Spoken Words in a Different Language Using Only Visual Groundinghttps://blog.schoergendorfer.com/post/tabulm-morphology-aware-tabular-pre-training-for-low-resource-languagesSecFeedde2026-08-28T10:49:29.135754+00:00TabuLM: Morphology-Aware Tabular Pre-training for Low-Resource Languageshttps://blog.schoergendorfer.com/post/planting-a-latent-variable-in-natural-looking-text-a-more-realistic-test-of-beliSecFeedde2026-08-28T10:49:29.104237+00:00Planting a Latent Variable in Natural-Looking Text: a More Realistic Test of Belief States in LLMs and Their Link to Concept Geometryhttps://blog.schoergendorfer.com/post/ruleweaver-benchmarking-rule-centered-scenario-reasoning-for-large-language-modeSecFeedde2026-08-28T10:49:29.077031+00:00RuleWeaver: Benchmarking Rule-Centered Scenario Reasoning for Large Language Modelshttps://blog.schoergendorfer.com/post/instruction-quality-matters-refining-instructions-for-effective-preference-learnSecFeedde2026-08-28T10:49:29.047173+00:00Instruction Quality Matters: Refining Instructions for Effective Preference Learninghttps://blog.schoergendorfer.com/post/preserving-general-capabilities-during-domain-specialization-with-uncertainty-caSecFeedde2026-08-28T10:49:28.990696+00:00Preserving General Capabilities during Domain Specialization with Uncertainty-Calibrated MOPDhttps://blog.schoergendorfer.com/post/towards-expert-financial-qa-via-self-improving-ragSecFeedde2026-08-28T10:49:28.955221+00:00Towards Expert Financial QA via Self-Improving RAGhttps://blog.schoergendorfer.com/post/scaling-phoneme-based-tts-augmentation-for-asr-a-unified-pipeline-and-controlledSecFeedde2026-08-28T10:49:28.895508+00:00Scaling phoneme-based TTS augmentation for ASR: A unified pipeline and controlled study