https://james.trappett.org/2026-08-14daily1.0 https://james.trappett.org/about/2026-08-14monthly0.9 https://james.trappett.org/blog/2026-08-14daily0.9 https://james.trappett.org/research/2026-08-14monthly0.8 https://james.trappett.org/research/ai/2026-08-14monthly0.8 https://james.trappett.org/research/security/2026-08-14monthly0.8 https://james.trappett.org/security/2026-08-14monthly0.7 https://james.trappett.org/contact/2026-08-14yearly0.6 https://james.trappett.org/blog/governing-multi-llm-agents-with-control-theory-eo-framework/2026-08-14weekly0.7 https://james.trappett.org/blog/backtrader-bench-benchmarking-llm-agents-on-algo-trading/2026-08-14weekly0.7 https://james.trappett.org/blog/farsky-generative-latent-space-coupling-for-solar-forecasting/2026-08-14weekly0.7 https://james.trappett.org/blog/when-does-chain-of-thought-actually-help-a-depth-analysis/2026-08-13weekly0.7 https://james.trappett.org/blog/curvefp-closed-product-logarithmic-formats-for-llm-quantization/2026-08-13weekly0.7 https://james.trappett.org/blog/the-multilingual-quantization-tax-in-edge-slms/2026-08-13weekly0.7 https://james.trappett.org/blog/data-centric-parallel-efficient-training-on-variable-sequences/2026-08-12weekly0.7 https://james.trappett.org/blog/fuzzing-mllms-to-expose-hallucination-blind-spots/2026-08-12weekly0.7 https://james.trappett.org/blog/scaling-inherently-interpretable-language-models/2026-08-12weekly0.7 https://james.trappett.org/blog/entropymoe-routing-tokenizer-free-llms-with-patch-entropy/2026-08-11weekly0.7 https://james.trappett.org/blog/texas-smarter-expert-discovery-for-moe-llm-fine-tuning/2026-08-11weekly0.7 https://james.trappett.org/blog/sharding-llm-judges-to-fix-oversight-failures/2026-08-11weekly0.7 https://james.trappett.org/blog/snowflake-extortions-anatomy-of-a-cloud-credential-attack/2026-08-10weekly0.7 https://james.trappett.org/blog/teacher-housing-as-infrastructure-what-school-districts-get-right/2026-08-10weekly0.7 https://james.trappett.org/blog/grounding-llms-in-industrial-simulators-for-causal-qa/2026-08-09weekly0.7 https://james.trappett.org/blog/triple-robustness-analysis-of-graphrag-for-multi-hop-retrieval/2026-08-09weekly0.7 https://james.trappett.org/blog/neutral-collapse-why-sentiment-analysis-fails-political-text/2026-08-09weekly0.7 https://james.trappett.org/blog/woodpecker-distillation-weak-models-fix-strong-model-reasoning/2026-08-08weekly0.7 https://james.trappett.org/blog/mean-field-theory-of-chain-of-thought-reasoning-in-llms/2026-08-08weekly0.7 https://james.trappett.org/blog/rig-rope-gated-rotary-encoding-for-multimodal-llms/2026-08-08weekly0.7 https://james.trappett.org/blog/bounded-aging-in-ai-a-formal-persistence-framework/2026-08-07weekly0.7 https://james.trappett.org/blog/llm-prompting-for-classical-latin-ner-evalatin-2026/2026-08-07weekly0.7 https://james.trappett.org/blog/c-moe-moe-framework-for-incomplete-multimodal-emotion/2026-08-07weekly0.7 https://james.trappett.org/blog/escaping-llm-homogeneity-with-meta-persona-anchoring/2026-08-06weekly0.7 https://james.trappett.org/blog/hyperagent-tool-schema-hypergraphs-for-llm-agent-planning/2026-08-06weekly0.7 https://james.trappett.org/blog/automating-circuit-tracing-annotation-with-llms/2026-08-06weekly0.7 https://james.trappett.org/blog/thinkreset-learning-intermediate-interfaces-for-long-reasoning/2026-08-05weekly0.7 https://james.trappett.org/blog/cheap-llm-judges-match-frontier-models-for-proof-grading/2026-08-05weekly0.7 https://james.trappett.org/blog/progressive-kd-closing-the-teacher-student-gap/2026-08-05weekly0.7 https://james.trappett.org/blog/full-stack-agentic-ai-openclaw-and-ollama-architecture/2026-08-04weekly0.7 https://james.trappett.org/blog/do-llms-understand-item-difficulty-evidence-from-assessment/2026-08-04weekly0.7 https://james.trappett.org/blog/topology-aware-kv-cache-transfer-for-disaggregated-llm-inference/2026-08-04weekly0.7 https://james.trappett.org/blog/h96-tv-boxes-ad-fraud-proxy-networks-ai-deception/2026-08-03weekly0.7 https://james.trappett.org/blog/the-computational-theory-of-mind-a-critical-analysis/2026-08-03weekly0.7 https://james.trappett.org/blog/mu-a-unified-mcp-tool-server-for-ai-agents-analysed/2026-08-03weekly0.7 https://james.trappett.org/blog/rfc-10015-closing-the-last-loopholes-in-tls-1-2-key-exchange/2026-08-02weekly0.7 https://james.trappett.org/blog/why-rl-models-reason-better-representational-evidence/2026-08-01weekly0.7 https://james.trappett.org/blog/prompt-chaining-vs-single-shot-llm-prompting-for-scholarly-digests/2026-08-01weekly0.7 https://james.trappett.org/blog/recursive-transformers-for-semiconductor-package-reliability/2026-08-01weekly0.7 https://james.trappett.org/blog/alignment-faking-persists-without-explicit-consequences/2026-07-31weekly0.7 https://james.trappett.org/blog/merla-meta-learned-reward-shaping-for-better-rlhf/2026-07-31weekly0.7 https://james.trappett.org/blog/transferring-sft-lessons-across-alignment-model-organisms-and-toy-models/2026-07-31weekly0.7 https://james.trappett.org/blog/neuromorphic-diffusion-llms-sparsity-meets-block-decoding/2026-07-30weekly0.7 https://james.trappett.org/blog/kernel-forge-agentic-llm-system-for-cuda-kernel-optimisation/2026-07-30weekly0.7 https://james.trappett.org/blog/llm-reliability-beyond-accuracy-the-paraphrase-consistency-gap/2026-07-29weekly0.7 https://james.trappett.org/blog/causalgate-intervention-based-pruning-for-llm-inference/2026-07-29weekly0.7 https://james.trappett.org/blog/autothinksql-teaching-llms-when-to-reason-for-text-to-sql/2026-07-29weekly0.7 https://james.trappett.org/blog/adversarial-style-optimization-jailbreaking-vlms-via-visual-style/2026-07-28weekly0.7 https://james.trappett.org/blog/consensus-based-relative-preference-evaluation-for-llms/2026-07-28weekly0.7 https://james.trappett.org/blog/probing-latent-colombian-identity-in-llm-residual-streams/2026-07-28weekly0.7 https://james.trappett.org/blog/wattage-token-profiling-and-cost-regression-for-ai-agents/2026-07-27weekly0.7 https://james.trappett.org/blog/llms-as-proof-automation-dependent-types-become-practical/2026-07-27weekly0.7 https://james.trappett.org/blog/lg-s-smart-tv-proxy-ban-what-the-webos-crisis-reveals/2026-07-27weekly0.7 https://james.trappett.org/blog/ai-and-jobs-what-the-data-actually-shows-in-2025/2026-07-26weekly0.7 https://james.trappett.org/blog/moe-routing-as-huffman-coding-the-frequency-diversity-law/2026-07-25weekly0.7 https://james.trappett.org/blog/what-actually-makes-muon-grok-faster-than-adamw/2026-07-25weekly0.7 https://james.trappett.org/blog/varrate-variable-rate-kv-cache-compression-without-training/2026-07-21weekly0.7 https://james.trappett.org/blog/reviewer-precision-vs-critique-uptake-in-multi-agent-math/2026-07-21weekly0.7 https://james.trappett.org/blog/global-workspace-theory-emerges-in-large-language-models/2026-07-21weekly0.7 https://james.trappett.org/blog/cisa-github-credential-leak-six-months-of-preventable-exposure/2026-07-20weekly0.7 https://james.trappett.org/blog/ai-accelerated-vulnerability-discovery-the-570-flaw-patch-problem/2026-07-20weekly0.7 https://james.trappett.org/blog/xai-must-fix-its-foundations-before-building-more-methods/2026-07-19weekly0.7 https://james.trappett.org/blog/class-aware-prompt-reweighting-for-zero-shot-vlms/2026-07-19weekly0.7 https://james.trappett.org/blog/certified-per-domain-contamination-control-for-multi-domain-retrieval/2026-07-19weekly0.7 https://james.trappett.org/blog/just-keep-prompting-vlm-epistemic-stability-under-pressure/2026-07-18weekly0.7 https://james.trappett.org/blog/do-llm-agents-lose-meaning-when-they-talk-to-each-other/2026-07-18weekly0.7 https://james.trappett.org/blog/hg-rag-graph-traversal-rag-for-hierarchical-knowledge/2026-07-18weekly0.7 https://james.trappett.org/blog/interventional-grounding-audits-for-llm-chain-of-thought/2026-07-17weekly0.7 https://james.trappett.org/blog/surveying-self-improvement-in-foundation-model-agents/2026-07-17weekly0.7 https://james.trappett.org/blog/scaling-point-in-time-llms-to-close-the-lookahead-bias-gap/2026-07-16weekly0.7 https://james.trappett.org/blog/grid-grammar-constrained-sql-decoding-for-enterprise-use/2026-07-16weekly0.7 https://james.trappett.org/blog/toulmin-argumentation-for-interpretable-retinal-ai-diagnosis/2026-07-15weekly0.7 https://james.trappett.org/blog/clir-bench-qa-benchmark-for-irregular-clinical-time-series/2026-07-15weekly0.7 https://james.trappett.org/blog/emergent-misalignment-may-be-a-dataset-artifact/2026-07-14weekly0.7 https://james.trappett.org/blog/stickymoe-training-moe-models-for-memory-efficient-inference/2026-07-14weekly0.7 https://james.trappett.org/blog/halo-adaptive-latent-refinement-for-frozen-language-models/2026-07-14weekly0.7 https://james.trappett.org/blog/iris-c2-and-the-zero-day-market-s-credibility-problem/2026-07-13weekly0.7 https://james.trappett.org/blog/building-culturally-specific-stereotype-datasets-with-llms/2026-07-12weekly0.7 https://james.trappett.org/blog/mesh-llm-distributed-inference-over-a-p2p-quic-mesh/2026-07-12weekly0.7 https://james.trappett.org/blog/self-distillation-for-web-search-agents-without-teacher-models/2026-07-11weekly0.7 https://james.trappett.org/blog/recolora-spectral-consolidation-for-continual-llm-fine-tuning/2026-07-11weekly0.7 https://james.trappett.org/blog/from-solvers-to-research-agents-llms-in-formal-mathematics/2026-07-11weekly0.7 https://james.trappett.org/blog/when-does-in-context-search-actually-help-llms/2026-07-10weekly0.7 https://james.trappett.org/blog/agentic-harnesses-for-arc-agi-67-at-under-1-per-task/2026-07-10weekly0.7 https://james.trappett.org/blog/triroute-joint-routing-across-attention-experts-and-kv-cache/2026-07-10weekly0.7 https://james.trappett.org/blog/benchmarking-kv-cache-optimisations-for-long-context-llm-serving/2026-07-09weekly0.7 https://james.trappett.org/blog/narrative-world-model-structured-memory-for-long-form-fiction-ai/2026-07-09weekly0.7 https://james.trappett.org/blog/gemma-4-open-weight-multimodal-models-with-encoder-free-architecture/2026-07-08weekly0.7 https://james.trappett.org/blog/flora-fixing-the-generator-validator-gap-in-llms/2026-07-08weekly0.7 https://james.trappett.org/blog/when-ai-audits-fail-silently-five-benchmark-validity-pitfalls/2026-07-08weekly0.7 https://james.trappett.org/blog/riddle-llm-inference-on-a-remarkable-via-handwriting/2026-07-07weekly0.7 https://james.trappett.org/blog/vessel-can-an-ega-adventure-game-model-machine-grief/2026-07-07weekly0.7 https://james.trappett.org/blog/the-corpus-royalty-who-owns-the-data-that-built-ai/2026-07-06weekly0.7 https://james.trappett.org/blog/the-has-not-been-viewed-much-field-obscurity-in-cultural-apis/2026-07-06weekly0.7 https://james.trappett.org/blog/gpt-5-5-codex-reasoning-token-clustering-a-performance-anomaly/2026-07-05weekly0.7 https://james.trappett.org/blog/fbi-seizes-netnut-and-popa-botnet-what-it-means/2026-07-05weekly0.7 https://james.trappett.org/blog/reversing-brain-aging-with-a-nasal-spray-what-the-science-says/2026-07-05weekly0.7 https://james.trappett.org/blog/kara-sliding-window-kv-cache-compression-for-reasoning-llms/2026-07-04weekly0.7 https://james.trappett.org/blog/bpe-tokenization-creates-exploitable-gaps-in-llm-safety/2026-07-04weekly0.7 https://james.trappett.org/blog/wiola-a-clean-slate-small-language-model-architecture/2026-07-04weekly0.7 https://james.trappett.org/blog/grpo-dr-grpo-and-dapo-are-one-operation-on/2026-07-03weekly0.7 https://james.trappett.org/blog/steering-vectors-and-latent-calibrators-for-llm-control/2026-07-03weekly0.7 https://james.trappett.org/blog/manifestation-units-structured-schema-for-interpretability/2026-07-03weekly0.7 https://james.trappett.org/blog/when-does-learned-early-stopping-beat-simple-thresholds/2026-07-02weekly0.7 https://james.trappett.org/blog/when-does-feedback-actually-help-separating-signal-from-noise/2026-07-02weekly0.7 https://james.trappett.org/blog/why-few-step-diffusion-fails-for-text-but-works-for-images/2026-07-02weekly0.7 https://james.trappett.org/blog/rsea-safer-llm-agent-self-evolution-via-held-out-selection/2026-07-01weekly0.7 https://james.trappett.org/blog/dynasteer-dynamic-representation-editing-for-llm-reasoning/2026-07-01weekly0.7 https://james.trappett.org/blog/hallucinations-in-the-limit-recall-precision-llm-generation/2026-07-01weekly0.7 https://james.trappett.org/blog/teaching-llms-to-simulate-the-future-before-they-act/2026-06-30weekly0.7 https://james.trappett.org/blog/four-axioms-for-evaluating-latent-thought-representations-in-llms/2026-06-30weekly0.7 https://james.trappett.org/blog/why-machine-unlearning-means-too-many-things-in-llm-research/2026-06-30weekly0.7 https://james.trappett.org/blog/black-box-llm-distillation-via-bayesian-distribution-estimation/2026-06-29weekly0.7 https://james.trappett.org/blog/beyond-accuracy-rethinking-ai-benchmark-saturation/2026-06-28weekly0.7 https://james.trappett.org/blog/algoevolve-llm-driven-meta-evolution-of-trading-strategies/2026-06-28weekly0.7 https://james.trappett.org/blog/llm-pipeline-for-comparing-dao-vs-corporate-ai-governance/2026-06-28weekly0.7 https://james.trappett.org/blog/cascading-linear-features-for-detecting-llm-sycophancy/2026-06-27weekly0.7 https://james.trappett.org/blog/refusal-in-chat-models-is-gated-by-persona-representations/2026-06-27weekly0.7 https://james.trappett.org/blog/rl-on-beneficial-traits-generalises-alignment-across-domains/2026-06-26weekly0.7 https://james.trappett.org/blog/wikipedia-edits-shape-llm-values-small-groups-big-impact/2026-06-26weekly0.7 https://james.trappett.org/blog/weight-space-geometry-of-offline-reasoning-training-methods/2026-06-25weekly0.7 https://james.trappett.org/blog/self-recognition-finetuning-as-a-defence-against-emergent-misalignment/2026-06-25weekly0.7 https://james.trappett.org/blog/ncu-metric-quantifying-rag-context-use-vs-parametric-memory/2026-06-25weekly0.7 https://james.trappett.org/blog/scattered-spider-guilty-pleas-anatomy-of-a-cybercrime-network/2026-06-24weekly0.7 https://james.trappett.org/blog/email-validation-via-spam-delivery-a-study-in-absurdity/2026-06-24weekly0.7 https://james.trappett.org/blog/why-vulnerability-reports-may-no-longer-deserve-special-status/2026-06-24weekly0.7 https://james.trappett.org/blog/popa-botnet-how-a-proxy-sdk-became-a-2m-node-threat/2026-06-22weekly0.7 https://james.trappett.org/blog/meta-mci-petition-employee-data-collection-for-ai-training/2026-06-22weekly0.7 https://james.trappett.org/blog/deontic-policies-for-runtime-governance-of-agentic-ai/2026-06-21weekly0.7 https://james.trappett.org/blog/dexposure-claw-agentic-llm-supervision-for-defi-risk/2026-06-21weekly0.7 https://james.trappett.org/blog/diffusion-language-models-a-systematic-benchmark-analysis/2026-06-20weekly0.7 https://james.trappett.org/blog/deepseek-v4-million-token-context-via-hybrid-sparse-attention/2026-06-20weekly0.7 https://james.trappett.org/blog/spsd-edge-prompt-compression-to-cut-cloud-llm-energy-cost/2026-06-20weekly0.7 https://james.trappett.org/blog/jetflow-scaling-speculative-decoding-with-parallel-tree-drafting/2026-06-19weekly0.7 https://james.trappett.org/blog/gaussian-mixture-attention-linear-time-sequence-mixing/2026-06-19weekly0.7 https://james.trappett.org/blog/codeblock-sparse-supervision-for-code-llms-at-block-granularity/2026-06-19weekly0.7 https://james.trappett.org/blog/divinit-diverse-query-init-for-better-agentic-search-scaling/2026-06-18weekly0.7 https://james.trappett.org/blog/repselect-targeting-forget-specific-representations-for-deep-llm-unlearning/2026-06-18weekly0.7 https://james.trappett.org/blog/kv-cache-as-editable-notebook-prefill-writes-conclusions/2026-06-18weekly0.7 https://james.trappett.org/blog/llada-cpp-diffusion-llm-inference-on-mobile-npus/2026-06-16weekly0.7 https://james.trappett.org/blog/fixing-llm-repetition-loops-by-editing-a-single-neuron/2026-06-16weekly0.7 https://james.trappett.org/blog/june-2026-patch-tuesday-ai-driven-vulnerability-discovery-at-scale/2026-06-15weekly0.7 https://james.trappett.org/blog/netherlands-seizes-800-servers-in-stark-industries-takedown/2026-06-15weekly0.7 https://james.trappett.org/blog/anthropic-s-regulatory-boomerang-did-they-ask-for-this/2026-06-15weekly0.7 https://james.trappett.org/blog/who-runs-the-gentlemen-ransomware-an-osint-deep-dive/2026-06-14weekly0.7 https://james.trappett.org/blog/state-ags-investigating-openai-regulatory-reckoning/2026-06-14weekly0.7 https://james.trappett.org/blog/toolsense-auditing-what-llms-actually-know-about-tools/2026-06-13weekly0.7 https://james.trappett.org/blog/arbor-tree-search-as-a-cognition-layer-for-ai-agents/2026-06-13weekly0.7 https://james.trappett.org/blog/pythagoras-prover-efficient-lean-theorem-proving-at-scale/2026-06-13weekly0.7 https://james.trappett.org/blog/the-structural-attention-tax-in-rag-format-hijacks-icl/2026-06-12weekly0.7 https://james.trappett.org/blog/blendin-quality-aware-inference-time-llm-alignment/2026-06-12weekly0.7 https://james.trappett.org/blog/dual-stance-evaluation-exposes-sycophancy-steering-blind-spots/2026-06-12weekly0.7 https://james.trappett.org/blog/opening-the-black-box-mechanistic-analysis-of-llm-alignment/2026-06-11weekly0.7 https://james.trappett.org/blog/training-llms-for-inductive-reasoning-via-probabilistic-programs/2026-06-11weekly0.7 https://james.trappett.org/blog/bi-temporal-memory-for-llm-agents-less-context-more-accuracy/2026-06-11weekly0.7 https://james.trappett.org/blog/cross-lingual-factual-recall-improved-via-consistency-rl/2026-06-10weekly0.7 https://james.trappett.org/blog/unpredictabench-testing-llms-as-statistical-random-generators/2026-06-10weekly0.7 https://james.trappett.org/blog/token-level-signatures-of-llm-reasoning-failures/2026-06-09weekly0.7 https://james.trappett.org/blog/fair-calib-quantizing-diffusion-language-models-safely/2026-06-09weekly0.7 https://james.trappett.org/blog/wav-multi-resolution-residual-routing-for-deep-transformers/2026-06-09weekly0.7 https://james.trappett.org/blog/meta-s-ai-support-bot-exploited-for-instagram-account-takeover/2026-06-08weekly0.7 https://james.trappett.org/blog/perceptron-from-scratch-foundations-of-neural-learning-in-python/2026-06-08weekly0.7 https://james.trappett.org/blog/covert-llm-agents-on-reddit-anatomy-of-ai-persuasion/2026-06-07weekly0.7 https://james.trappett.org/blog/pact-structured-agent-communication-cuts-llm-token-costs/2026-06-07weekly0.7 https://james.trappett.org/blog/gitco-inference-time-context-optimization-for-time-series-models/2026-06-07weekly0.7 https://james.trappett.org/blog/modelling-ai-model-collapse-as-an-epidemic-a-bilayer-sir-approach/2026-06-06weekly0.7 https://james.trappett.org/blog/stereological-theory-reveals-structural-blind-spots-in-llm-benchmarks/2026-06-06weekly0.7 https://james.trappett.org/blog/do-transformers-really-need-three-projections-in-attention/2026-06-05weekly0.7 https://james.trappett.org/blog/rag-in-biomedicine-when-more-context-doesn-t-help/2026-06-05weekly0.7 https://james.trappett.org/blog/polaris-training-small-models-for-long-form-story-writing/2026-06-05weekly0.7 https://james.trappett.org/blog/visual-graph-scaffolds-for-structural-reasoning-in-llms/2026-06-04weekly0.7 https://james.trappett.org/blog/aura-mem-constant-vram-memory-for-long-horizon-robot-policies/2026-06-04weekly0.7 https://james.trappett.org/blog/bank-of-values-context-free-value-vectors-in-deep-attention/2026-06-04weekly0.7 https://james.trappett.org/blog/sense-semantic-retrieval-for-faster-llm-speculative-decoding/2026-06-03weekly0.7 https://james.trappett.org/blog/bitsmoe-spectral-bit-allocation-for-ultra-low-moe-quantization/2026-06-03weekly0.7 https://james.trappett.org/blog/dopa-ood-proxy-guided-demonstration-retrieval-for-icl/2026-06-03weekly0.7 https://james.trappett.org/blog/phydrawgen-neuro-symbolic-physics-diagram-generation/2026-06-02weekly0.7 https://james.trappett.org/blog/rag-based-verification-of-chatgpt-biomedical-associations/2026-06-02weekly0.7 https://james.trappett.org/blog/qasm-eval-benchmarking-llms-on-openqasm-3-hardware-features/2026-06-02weekly0.7 https://james.trappett.org/blog/post-training-not-data-why-llms-write-the-way-they-do/2026-06-01weekly0.7 https://james.trappett.org/blog/chatgpt-for-google-sheets-prompt-injection-data-exfiltration/2026-06-01weekly0.7 https://james.trappett.org/blog/rotary-gpu-running-35b-moe-models-on-8gb-vram/2026-05-31weekly0.7 https://james.trappett.org/blog/simplicial-message-passing-as-a-language-model-inductive-bias/2026-05-30weekly0.7 https://james.trappett.org/blog/can-llm-reviews-be-gamed-alignment-and-vulnerability/2026-05-30weekly0.7 https://james.trappett.org/blog/why-rl-preserves-circuits-better-than-sft-during-fine-tuning/2026-05-30weekly0.7 https://james.trappett.org/blog/modular-llm-architecture-for-human-value-detection-in-text/2026-05-29weekly0.7 https://james.trappett.org/blog/icg-personalised-cover-image-generation-with-mllms/2026-05-29weekly0.7 https://james.trappett.org/blog/llms-as-their-own-data-pipelines-via-self-verified-distillation/2026-05-28weekly0.7 https://james.trappett.org/blog/gem-hyperspherical-clustering-for-llm-data-mixing/2026-05-28weekly0.7 https://james.trappett.org/blog/do-llms-actually-introspect-a-critical-re-examination/2026-05-28weekly0.7 https://james.trappett.org/blog/llms-are-overconfident-especially-on-hard-tasks/2026-05-27weekly0.7 https://james.trappett.org/blog/quantifying-redundancy-in-llm-reasoning-chains/2026-05-27weekly0.7 https://james.trappett.org/blog/water-filling-token-allocation-for-reliable-llm-workflows/2026-05-27weekly0.7 https://james.trappett.org/blog/lcf-faster-llm-agent-communication-via-latent-cache-transfer/2026-05-26weekly0.7 https://james.trappett.org/blog/fura-full-rank-peft-via-spectral-preconditioning/2026-05-26weekly0.7 https://james.trappett.org/blog/the-positional-shortcut-hiding-inside-arithmetic-cot/2026-05-26weekly0.7 https://james.trappett.org/blog/cisa-aws-govcloud-keys-leaked-on-github-a-deep-failure/2026-05-25weekly0.7 https://james.trappett.org/blog/bayesian-gps-with-unknown-coordinates-location-error-modelling/2026-05-25weekly0.7 https://james.trappett.org/blog/kimwolf-botnet-iot-ddos-infrastructure-and-the-dort-arrest/2026-05-25weekly0.7 https://james.trappett.org/blog/cisa-credential-leak-insider-threat-meets-systemic-failure/2026-05-24weekly0.7 https://james.trappett.org/blog/ice-s-25m-iris-scanning-contract-biometrics-without-oversight/2026-05-24weekly0.7 https://james.trappett.org/blog/solar-autonomous-weight-level-meta-learning-for-lifelong-llm-adaptation/2026-05-23weekly0.7 https://james.trappett.org/blog/probabilistic-token-attribution-for-llms-via-stochastic-processes/2026-05-23weekly0.7 https://james.trappett.org/blog/feature-attribution-is-provably-unstable-under-collinearity/2026-05-23weekly0.7 https://james.trappett.org/blog/data-probes-a-principled-framework-for-understanding-llms/2026-05-22weekly0.7 https://james.trappett.org/blog/how-llms-misrepresent-disability-toxic-positivity-in-ai/2026-05-22weekly0.7 https://james.trappett.org/blog/mi-guided-parallel-decoding-in-masked-diffusion-models/2026-05-22weekly0.7 https://james.trappett.org/blog/hellora-smarter-lora-fine-tuning-for-moe-models/2026-05-21weekly0.7 https://james.trappett.org/blog/calibrated-uncertainty-for-cost-optimal-llm-cascade-routing/2026-05-21weekly0.7 https://james.trappett.org/blog/lbw-guard-autonomous-governance-layer-for-stable-llm-training/2026-05-21weekly0.7 https://james.trappett.org/blog/agentwall-runtime-safety-enforcement-for-local-ai-agents/2026-05-20weekly0.7 https://james.trappett.org/blog/scaling-laws-for-skill-libraries-in-llm-agent-systems/2026-05-20weekly0.7 https://james.trappett.org/blog/real-time-diffusion-inference-on-apple-silicon-what-works/2026-05-20weekly0.7 https://james.trappett.org/blog/op-mix-unified-data-mixing-across-the-llm-training-lifecycle/2026-05-19weekly0.7 https://james.trappett.org/blog/quantization-breaks-alignment-bias-emerges-in-compressed-llms/2026-05-19weekly0.7 https://james.trappett.org/blog/teamtr-trust-region-fine-tuning-for-multi-agent-llms/2026-05-19weekly0.7 https://james.trappett.org/blog/project-glasswing-how-ai-is-rewriting-patch-tuesday/2026-05-18weekly0.7 https://james.trappett.org/blog/grid2poster-visualising-electrical-infrastructure-from-osm-data/2026-05-18weekly0.7 https://james.trappett.org/blog/graphbit-deterministic-dag-orchestration-for-llm-agents/2026-05-17weekly0.7 https://james.trappett.org/blog/a-2d-framework-for-classifying-ai-agent-design-patterns/2026-05-17weekly0.7 https://james.trappett.org/blog/preping-building-agent-memory-before-tasks-begin/2026-05-17weekly0.7 https://james.trappett.org/blog/mistletoe-attacking-speculative-decoding-without-changing-outputs/2026-05-16weekly0.7 https://james.trappett.org/blog/hidden-ai-coordinators-create-invisible-safety-risks-in-multi-agent-systems/2026-05-16weekly0.7 https://james.trappett.org/blog/trajectory-balance-post-training-for-diffusion-language-models/2026-05-16weekly0.7 https://james.trappett.org/blog/revelio-uncovering-interpretable-failure-modes-in-vlms/2026-05-15weekly0.7 https://james.trappett.org/blog/differential-privacy-and-social-bias-in-llms-a-complex-trade-off/2026-05-15weekly0.7 https://james.trappett.org/blog/attention-maps-don-t-predict-vlm-reliability-hidden-states-do/2026-05-14weekly0.7 https://james.trappett.org/blog/why-llms-collapse-to-narrow-outputs-a-calibration-view/2026-05-14weekly0.7 https://james.trappett.org/blog/variational-deep-embedding-for-interpretable-eeg-microstates/2026-05-14weekly0.7 https://james.trappett.org/blog/running-local-llms-on-apple-m4-with-24gb-ram/2026-05-11weekly0.7 https://james.trappett.org/blog/canvas-breach-analysis-edtech-security/2026-05-10weekly0.7 https://james.trappett.org/blog/freebsd-execve-privilege-escalation-analysis/2026-05-10weekly0.7 https://james.trappett.org/blog/patch-tuesday-april-2026-167-cves/2026-05-09weekly0.7 https://james.trappett.org/blog/scattered-spider-tylerb-guilty-plea-analysis/2026-05-09weekly0.7 https://james.trappett.org/blog/anti-ddos-firms-enabling-attacks-they-sell-against/2026-05-08weekly0.7 https://james.trappett.org/blog/russia-router-token-harvesting-deep-dive/2026-05-08weekly0.7 https://james.trappett.org/blog/your-brain-processes-language-while-unconscious/2026-05-07weekly0.7 https://james.trappett.org/blog/creativitybench-can-ai-really-think-creatively/2026-05-07weekly0.7 https://james.trappett.org/blog/openai-new-voice-api-features-analysis/2026-05-07weekly0.7 https://james.trappett.org/blog/clojurescript-gets-async-await-analysis/2026-05-07weekly0.7 https://james.trappett.org/blog/ai-annotator-safety-policy-interpretability/2026-05-07weekly0.7 https://james.trappett.org/blog/why-your-doctor-never-calls-back/2026-05-07weekly0.7