Nightly

Archive

Every edition

102 editions and 2,294 stories so far. Press / to search all of them.

October 2026

  1. 23FriLLM Agent Finds 1.4M Fake Docs in Slovak Crawl28
  2. 22ThuExact‑KL clipping reshapes LLM policy28
  3. 21WedGPT-6.1 matches investors 88.6% of time28
  4. 20TueVisual agents get memory boost28
  5. 19MonLinux 7.3 shatters commit record18
  6. 18SunESP32 firmware runs in-browser16Week in review · 2026-W42
  7. 17SatSafe self‑referential structs land in kernel17
  8. 16FriX‑RIC adds agents to telecom controllers28
  9. 15ThuLLMs stay steady when UI flips28
  10. 14WedBLADE pins LLM multi‑agent failures28
  11. 13TueMeta repurposes spare inference for better ads17
  12. 12MonExt4 drops data=journal mode15
  13. 11SunUnderdog‑Saluki hits 27 B parameters13Week in review · 2026-W41
  14. 10SatAsana slashes GPT-6.1 spend18
  15. 9FriLLMs tested for epistemic humility22
  16. 8ThuKV cache shrinks with 2‑bit residuals28
  17. 7WedDeepMind releases lightweight multimodal embedder28
  18. 6TueLLMs get on‑demand memory manager28
  19. 5Mon7.3‑rc6 prepatch lands, no schedule shift14
  20. 4SunAleph Alpha unveils sovereign Kolibri LLM17Week in review · 2026-W40
  21. 3SatOpenAI publishes GPT‑6 selection guide19
  22. 2FriKaliBench measures CLI translation on security tools28
  23. 1ThuGemma-3-12B holds its own vs GPT‑4o28

September 2026

  1. 30WedGPT-6.1 Sol costs a fifth of Astra27
  2. 29TueTokenCast predicts LLM token use on the fly28
  3. 28MonChat templates give LLMs a self‑referential twist15
  4. 27SunModular macro keypad simplifies USB hardware14Week in review · 2026-W39
  5. 26SatTwo new Linux stable kernels land Friday15
  6. 25FriA Living Benchmark for Information Retrieval from Electronic Health Records28
  7. 24ThuAgent-Editing World Model: Rethinking World Modeling for LLM Agents28
  8. 23WedGrow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents28
  9. 22TueHarness-Zero: Harness Distillation via Agent-as-Harness28
  10. 21MonKernel prepatch 7.3-rc416
  11. 20SunSDCC – Small Device C Compiler14
  12. 19SatIntroducing the Australian Youth Safety Blueprint15
  13. 18FriChronicle: Cut-Point Replay for Regression Testing of LLM Agents28
  14. 17ThuCERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents25
  15. 16WedJustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time State Management27
  16. 15TueUpcoming: Codeforces Round (Div. 3)17
  17. 14MonGarry Tan wants US open-weight AI labs to 'distill' frontier models, too13
  18. 13SunSnailSploit/Claude-Red13
  19. 12SatTwo stable kernel updates for Friday16
  20. 11FriRAG-Safety-Bench: Reliable Evaluation of Retrieval-Augmented LLM Safety28
  21. 10ThuOn-Policy Distillation for Vision-Language Model Adaptation, an Effective Paradigm on Low-Quality Multimodal Data26
  22. 9WedAnswer-Distribution Trajectories: A Stochastic-Dynamics View of LLM Reasoning28
  23. 8TueESP32-P4 Powers Video Alarm Clock16
  24. 7MonUpcoming: Codeforces Round (Div. 2)17
  25. 6SunUpcoming: Codeforces Round (Div. 2)15
  26. 5SatUpcoming: Codeforces Round (Div. 1)17
  27. 4FriSENTINEL-RL: Offloading Topological Reasoning from LLM Agents in the Security Operations Center28
  28. 3ThuLoom: Weaving Diagnostic Strands into Free-Text Consensus via Embedding-Space Reweighting28
  29. 2WedTRIAGE: Three-level Routing and Intelligent Agent Guidance for Efficient Execution28
  30. 1TueToken-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data25

August 2026

  1. 31MonHow to build a diffusion language model14
  2. 30SunUpcoming: Educational Codeforces Round 194 (Rated for Div. 2)14
  3. 29Sat[$] The "rnull" Rust block driver16
  4. 28FriCritICL: Inference-Time Weak-to-Strong Generalization from Small Language Model Failure Modes28
  5. 27ThuProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs29
  6. 26WedMoTE: Mixture of Task Experts for Multi-Task Video Understanding26
  7. 25TueChebBooster: A Training-Free Approach for Efficient Diffusion Transformer Inference via Chebyshev-Inspired Extrapolation27
  8. 24MonMalware infects Android-based automotive head unit firmware12
  9. 23SunShow HN: OzBrain, a shared brain for knowledge between agents and your team16
  10. 22SatTumble Forth – from assembly to OS with C compiler (2023)17
  11. 21FriMulti-Agent Orchestration with the Common-Sense Reasoning Capabilities of LLMs for Autonomous Driving26
  12. 20ThuPre-Compiled Pipeline Shards for Distributed LLM Inference on Intel AI PC Fleets29
  13. 19WedWhere A Small Language Model Helps in Invoice Categorisation, Understood Through Embedding Geometry28
  14. 18TueTopological Attribution Distance (TAD): Revealing Segment-Level RAG Influence on LLM Output Geometry for Incident Log Analysis28
  15. 17MonA 3rd World Embedded Engineer Responds to "RISC-V They Should Have Known Better"15
  16. 16SunESP32 Music Sequencer is Clearly Nailing the Y2K Aesthetic16
  17. 15SatRISC-V: They should have known better17
  18. 14FriReduced Matrix Multiplication: Input-Adaptive Matrix-Product Reduction for LLM Inference28
  19. 13ThuOne Frozen Simulator Is Not Enough: Simulator Collapse in Multi-Agent RL28
  20. 12WedV-FiLLM: Verified Financial LLM Reasoning Benchmark28
  21. 11TueModel Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models28
  22. 10MonFour weekend stable kernel updates14
  23. 9SunUpcoming: Codeforces Round (Div. 2)15
  24. 8SatStable kernel releases for Friday with a single bug fix18
  25. 7FriEnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning28
  26. 6ThuArgus: A General-Purpose Agentic Runtime for Long-Horizon Reasoning27
  27. 5WedParVL: Parallel Scaling and Expandable Compute Allocation for Multimodal LLMs29
  28. 4TueAURORA-LM: Autoencoding Unified Representation for Continuous-Latent Diffusion Language Modeling27
  29. 3MonUpcoming: Codeforces Round (Div. 1)15
  30. 2SunLibreTranslate/LibreTranslate16
  31. 1SatEveryone is building LLM routers, we deprecated ours19

July 2026

  1. 31FriUpcoming: Codeforces Round (Div. 3)29
  2. 30ThuInferScale: GPU-Native KV Injection for Personalized LLM Serving29
  3. 29WedA Cost-Effective Multimodal LLM Reasoning Framework for Question Answering over Irregular Clinical Time Series29
  4. 28TueSIREN: Towards End-to-End Extreme-Weather Early Warning with Experience-Grounded LLM Agents29
  5. 27MonAn ESP32 based plane radar for my desk16
  6. 26SunLLM Usage in Debian: Three Proposals18
  7. 25SatHetzner is working on LLM Inference18
  8. 24FriAgentic coding without the cloud: evaluating open-weight large language models on longitudinal data preparation tasks28
  9. 23ThuPyroDash: Cost-Efficient Token-Level Small-Large Language Model Collaborative Inference29
  10. 22WedBeyond Score Prediction: LLM-Based Essay Scoring and Feedback Generation via Reinforcement Learning with Rubric Rewards28
  11. 21TueSWE-Pruner Pro: The Coder LLM Already Knows What to Prune27
  12. 20MonUpcoming: Codeforces Round (Div. 2)16
  13. 19SunUpcoming: Codeforces Round (Div. 1)16
  14. 18SatFine-tune video and image models at scale with NVIDIA NeMo Automodel and 🤗 Diffusers17
  15. 17FriDigital Pantheon: Simulating and Auditing Coalition Formation with LLM Agents28
  16. 16ThuTraffic-Aware Randomized Smoothing for LLM-Based Network Intrusion Detection28
  17. 15WedVerifier-Based Reinforcement Fine-Tuning of Reasoning Models for Thermal Energy Storage Control29
  18. 14TueJobHop v2: A Large-Scale Career Trajectory Dataset from Unstructured Resumes29