[BLG]arXiv cs.CL/arxiv.org21h
FineWeb-CLaR: Culture, Language, and Region Annotations for Benchmark-Aligned Corpus AuditingYusser Al Ghussin, Eva Gavaller, Cristina Espa\~na-Bonet, Josef van Genabith, Simon Ostermann
>trending topics
- no.01
Claude Opus 5.5 release
→Big model release today - I wrote about Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna - plus comparison grids …
- no.02
Meta Muse personal AI agent for Mac
→I'm skeptical about Muse www.platformer.news/meta-muse-co...
- no.03
Trump renames AI to super intelligence
→Trump says US will henceforth call AI 'super intelligence'
- no.04
GPT-6 Sol and Luna launch
→"I know I am GPT-6. I cannot confirm whether this session uses GPT-6-Sol."
- no.05
Jev structured output decision models launch
→The Jagged Frontier of Jev 1.13
- no.06
ZCode code theft controversy and open source
→Z.ai uploading local code repos to overseas cloud without consent
- no.07
arXiv computer science papers September 23
→"As a Language Model...": Chat Template Switches LLM Self-Referential Voice and Activation Steering Reprodu…
- no.08
arXiv computational linguistics papers September 23
→What Does 99% Accuracy Measure? A Reproducible Audit of Shortcut Learning in a Widely Used Fake News Corpus
- no.09
arXiv machine learning papers September 23
→Beyond Scalar Sensitivity: Activation-Aware Mixed-Precision LLM Quantization with Cross-Layer Refinement
- no.10
Claude Opus 5.5 release and benchmarks
→Claude Opus 5.5 (High Effort) Intelligence, Performance and Price Analysis
>raw feed
[HN]HN: GPT/github.com1d
Show HN: I generated 21k astrology pages with Swiss Ephemeris, open sourcedtianji-astrolog·▲ 2
[BSKY]@emollick/bsky.app1d
Opus 5.5 was a good model in my early use tests, it is the first non-Fable/Astra model to feel like a Fable-class model, and much cheaper, but still hasn't fully solved the dense/weird language issue of the recent Claudes.@emollick.bsky.social·▲ 119
[BLG]arXiv cs.CL/arxiv.org21h
Conduct Under Pressure: What Sixty Language Models Do When a User PushesTapan Parikh
[HN]HN: AI/reuters.com1d
Z.ai uploading local code repos to overseas cloud without consenttxscdn·▲ 4
[BSKY]@caseynewton/bsky.app1d
Getting my associate’s degree in regulatory arbitrage from Horowitz Andreessen Academy@caseynewton.bsky.social·▲ 91
[BLG]arXiv cs.CL/arxiv.org21h
FinFIRST: Benchmarking Search Agents for Financial Information Retrieval, Sourcing and TraceabilityWenqing Wang, Haitao Xiang, Xinyi Zhao, Mingming Yin, Ying Zhong, Zhaoxin Huan, Qiheng Zhou, Jin Zhu, Xiaolu Zhang, Shi Chang, Jun Zhou
[HN]HN: Gemini/github.com1d
Show HN: Birthed, a daily board of what mattered that locks at midnightjpage2·▲ 2
[BSKY]@emollick/bsky.app1d
For all the tension between AI and the arts, I have received enthusiastic receptions from the William Carlos Williams Society, the TS Eliot Society, and others about the AI interpretations of their work that I have posted here. There are o…@emollick.bsky.social·▲ 71
[HN]HN: AI/rnz.co.nz23h
The 'AI confidence trap' many New Zealanders are caught inbillybuckwheat·▲ 8
[BSKY]@natolambert/bsky.app1d
Debating RSI, the US-China Gap, and Jaggedness with JS Denain of Epoch AI@natolambert.bsky.social·▲ 11
[BLG]arXiv cs.CL/arxiv.org21h
Matryoshka attribution: Learning to attribute language model outputs to representations and weightsAryaman Arora, Kirill Acharya, Nathan Hu, Yanzhe Zhang, Noah Goodman, Dan Jurafsky, Christopher Potts
[BSKY]@emollick/bsky.app22h
I had Fable/Opus build a hard science fiction starship combat game, with orbital mechanics, delta-v, heat, and realistic tactics, but simplified for 2D space with the hard math done by the game.@emollick.bsky.social·▲ 35
[BLG]arXiv cs.CL/arxiv.org21h
CoVeR: Coverage-Based Routing of Verifier Calls in Agentic RetrievalDaeyoung Roh, Donghee Han
[HN]HN: AI/theregister.com1d
Naming AI agents after Seinfeld characters helps bots join the team – reportmikelgan·▲ 3
[HN]HN: GPT/github.com1d
RepoDocs – turn your repo's Markdown/*.md into a Figma-like board you can editalecerca·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
TelecomGPT-R1: Unified Post-Training for Reasoning Across Heterogeneous Telecom TasksBohao Wang, Chenwei Wu, Hang Zou, Yu Tian, Lina Bariah, Li Wei, Chongwen Huang, Yongliang Shen, Zhaoyang Zhang, Merouane Debbah
[BSKY]@simonw/bsky.app1d
Big model release today - I wrote about Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna - plus comparison grids of pelicans by the different model families at different reasoning levels simonwillison.net/2026/Sep/22/...@simonwillison.net·▲ 57
[BSKY]@caseynewton/bsky.app1d
I'm skeptical about Muse www.platformer.news/meta-muse-co...@caseynewton.bsky.social·▲ 77
[HN]HN: RAG/journals.sagepub.com1d
We Are Speaking 300 Daily Words Fewer Every Yeargeox·▲ 2
[BLG]GitHub Trending — Python/github.com21h
TNT-Likely/PanWatch[BLG]arXiv cs.CL/arxiv.org21h
Rethinking Length-Based Training: Batch Composition and Loss Normalization in Speech Token Language ModelsHongjin Song, Runwu Shi, Weiqiao Shan, Jiale Luo, Yujin Wang, Yifei Wu, Chunxiang Jin
[BSKY]@yoshuabengio/bsky.app1d
J’étais de passage à l’émission Tout le monde en parle dimanche dernier pour parler de l'urgence d'encadrer l'IA face aux comportements d’agents autonomes et désalignés observés récemment et du travail @law-zero.bsky.social pour développer…@yoshuabengio.bsky.social·▲ 8
[HN]HN: AI/notateslaapp.com1d
Tesla Closes Texas Airspace for Roadster Launch Eventm463·▲ 4
[HN]HN: RAG/agekey.org22h
AgeKey – Privacy First Age Verificationpickledish·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
ICDAR2026 Competition on Multimodal Reasoning over Documents in Multiple DomainsArtemis Llabr\'es, Marc Serra Ortega, Tom\`as Ockier, Samuel Ortega Cuadra, Amritpal Singh, Christos Georgakilas, Andrey Barsky, Ernest Valveny, Dimosthenis Karatzas
[HN]HN: AI/arxiv.org22h
Specs cut defects in AI-generated code from 148 to 23 across five modelssandeepdhuri·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Truth for Believable AI: Expressed Doubt, Provenance, and Belief Revision as an Engineerable StanceSebastian Cochinescu
[HN]HN: AI/enworld.org1d
EN World Is Being DDoS'd by AI Botsschrectacular·▲ 3
[BLG]arXiv cs.CL/arxiv.org21h
From Utterances to Networks: Modelling Slang Adoption and Diffusion Across SubredditsXiaoning Wang, Ted Underwood, Zhewei Sun
[BLG]arXiv cs.CL/arxiv.org21h
MemoryAthena: Adaptive Routing over Latent and Generated MemoriesMingyuan Li, Guangsheng Yu, Juyuan Zhang, Xu Wang, Zhibo Man, Haonan Zhang, Shaoxiong Ji
[HN]HN: AI/vercel.com1d
Open-weight models take 56% of token volumemgdo·▲ 2
[HN]HN: GPT/abc.net.au1d
Trump administration attacks Australia optout algorithm law in rare interventionjyhrow·▲ 4
[BLG]arXiv cs.CL/arxiv.org21h
Passes Alone, Fails Together: Benchmarking Semantic Coordination in Parallel LLM-Agent DevelopmentHaocheng Xia, Eugene Wu, Yongjoo Park
[HN]HN: GPT/twitter.com23h
"I know I am GPT-6. I cannot confirm whether this session uses GPT-6-Sol."benwills·▲ 5
[BLG]arXiv cs.CL/arxiv.org21h
ClusterFewshot: Improving Few-shot Optimization for LLMs workflowOmri Bar Haim, Shahar Katz, Lior Wolf
[HN]HN: RAG/godaddy.com23h
Don't Trust. Verify. Offline, Sub-Millisecond Agent Verification with ANStmuhlestein·▲ 2
[BLG]arXiv cs.CL/arxiv.org21h
ARAFA: An LLM-Generated Arabic Fact-Checking DatasetChristophe Khalil, Shady Elbassuoni, Rida Assaf
[HN]HN: GPT/github.com1d
Web-based IBM 1620 emulator and IPL-V from 1963abrax3141·▲ 4
[BLG]arXiv cs.CL/arxiv.org21h
Differentiable Fuzzy Inference Layer: A Monotone, Compositional Ordinal Reasoning Head for Large Language ModelsZhen Zhang, Amr Alanwar
[HN]HN: RAG/developer.mozilla.org23h
HTTP Range Requestrickcarlino·▲ 2
[BLG]arXiv cs.CL/arxiv.org21h
Qwen3.8-Omni: Towards Native Omni-Modal Agents Qwen Team
[BLG]GitHub Trending — Daily/github.com21h
mvt-project/mvt[HN]HN: LLM/arxiv.org22h
Emergent Collusion in Long-Horizon LLM Agent Interactionsbulaev·▲ 1
[BLG]GitHub Trending — Python/github.com21h
cv-cat/DouYin_Spider[HN]HN: RAG/codeberg.org1d
AQAB: Anti-fascist blocklist QAnon, conspiracy, fake news, far-right, hate sitesBaljhin·▲ 3
[BLG]arXiv cs.CL/arxiv.org21h
Magnitude Profile Pruning: Calibration-Free Structured Attention Head Removal for Transformer CompressionKasun Dewage, Marianna Pensky, Heranga K. Rathnasekara, Suranadi De Silva
[HN]HN: AI/yovico.ai1d
Formal methods can start smallsiberiankot·▲ 1
[HN]HN: GPT/bbc.com23h
US criticises Australia's proposed algorithm opt-out laws as 'censorship'1659447091·▲ 84
[BLG]arXiv cs.CL/arxiv.org21h
Syndrome, Synergy, and Safety: Structured Reasoning and Knowledge-Driven Alignment for TCM Prescription GenerationZheng Chen, ZhiCheng Du, Haoxuan Li, Peiwu Qin
[HN]HN: AI/thenextmove.org21h
A Machine Beat Me at Chess. I'm Still Rooting for AIakbarnama·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Understanding Reliability in LLM-based Human Behavior SimulationPei Wang, Lei Wang, Yuanzi Li, Xu Chen
[BLG]arXiv cs.CL/arxiv.org21h
ChainDoRA: Tensor-Train Factorized Weight-Decomposed Low-Rank Adaptation for Parameter-Efficient LLM Fine-TuningAshfak Yeafi, Mehedi Hasan, Md Khairul Islam
[HN]HN: GPT/github.com23h
Tsetlin Machine MNIST classifier running on a Tang Nano 9K FPGA boardUncleOxidant·▲ 3
[HN]HN: AI/techcrunch.com22h
Snorkel AI triples valuation to $3.5B as demand for AI training data boomsnlpnerd·▲ 2
[BLG]GitHub Trending — TypeScript/github.com21h
bendlang/bend[BLG]arXiv cs.CL/arxiv.org21h
Graph-Based Inference for Feedback-Driven Word Deduction: A Scalable Framework for the Jotto ProblemDakshi Arora, Prakhar Kumar Srivastava, Ranjib Banerjee
[HN]HN: GPT/rgpu.dev22h
RGPU – Use a remote Nvidia GPU from PyTorch on a machine that has noneboxstream·▲ 1
[HN]HN: RAG/federalreserve.gov23h
In 2025, 49 percent of adults under age 30 lived with a parent [pdf]gscott·▲ 8
[BLG]arXiv cs.CL/arxiv.org21h
ufakzeka-1: Building and Evaluating a 151M-Parameter Turkish Language Model from ScratchSait Furkan Teke (ufak AI)
[HN]HN: AI/reuters.com1d
Trump says US will henceforth call AI 'super intelligence'OuterVale·▲ 20
[BLG]arXiv cs.CL/arxiv.org21h
Mining Legal Arguments in U.S. Corporate Case LawLuis Brena, William Jurayj, Gregory Deyesu, Zaid Al-Huneidi, Andrew Blair-Stanek, Benjamin Van Durme
[HN]HN: AI/fiit.ai1d
Weak models write tool calls as prose. We started executing themairbai·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Reply to comments arXiv:2512.07881 and arXiv:2601.06104 on quantum structure in human and AI-generated languageMassimiliano Sassoli de Bianchi, Roberto Leporini
[BLG]arXiv cs.CL/arxiv.org21h
TSS: Target-Side Sparsification for Speculative Decoding in Domain-Specific Large Language ModelsHaibo Hu, Lianming Huang, Qiao Li, Nan Guan, Chun Jason Xue
[HN]HN: RAG/agent-device.dev22h
Agent-browser equivalent for desktop and mobile appsnovoreorx·▲ 1
[HN]HN: AI/academy.dair.ai23h
Wiki Foundation Model for Complex Agentic Reasoningomarsar·▲ 4
[BLG]arXiv cs.CL/arxiv.org21h
BELXTR: Biomedical Entity Linking via Contextualized Token RetrievalSamuele Garda, Ulf Leser
[BLG]arXiv cs.CL/arxiv.org21h
SpecialEduBench: Benchmarking Vision-Language Models on Knowledge, Skill, and Attitude in Language Intervention for Autistic ChildrenJihoi Na, Taeyeong Kim, Sungjune Kong, Jaemin Jung, Min Joung Park, Kyungtae Joo, Ahhyun Kim, Shim Jaechang, Sooyoung Joo, Dongjin Ka, SeJoong Kim, Jimin Kim, HyunJin Jung, Unggi Lee
[HN]HN: GPT/github.com22h
Porting Bhyve to Rustnwilkens·▲ 3
[HN]HN: AI/polso.info1d
I'm sick of Claudisms, & what will happen next in AI-boosted software devrizsyed1·▲ 4
[BLG]arXiv cs.CL/arxiv.org21h
One Domain, Many Tongues: Composing Domain and Language LoRAs for Cross-Lingual Remote-Sensing MLLMs without Paired DataXuechen Li
[HN]HN: AI/vece.ai23h
Get to know your finance brokoliev·▲ 2
[BLG]arXiv cs.CL/arxiv.org21h
Optimizing Denoising Trajectories in dLLMs: A Lightweight Evolutionary Heuristic ApproachZijian Zhao, Dian Jin, Xialiang Tong, Sen Li, Mingxuan Yuan
[HN]HN: LLM/github.com21h
Loment: A self-hosted systems language with multi-syntax front endsFujoLoment·▲ 2
[BLG]arXiv cs.CL/arxiv.org21h
Compressing Long Context into Answer-Aligned Memory Embeddings for LLM InferenceMd Mostafizer Rahman, Md Faizul Ibne Amin, Md Shahajada Mia, Yutaka Watanobe, Fang Liu
[HN]HN: GPT/github.com1d
Conway's Game of Life Running on a Tang Nano 9k FPGA Board with SPI LCDUncleOxidant·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Modality-Gated Deep Adapters: Adding a Modality to a Frozen Embedding Model with Exact PreservationAbdul Basit Tonmoy, Kazi Fardinul Hoque, Md. Shahrier Islam Arham, Arman Luthra
[HN]HN: LLM/arxiv.org22h
Measuring behavioral signals of LLM through psychometric profilinganigbrowl·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Isolated Sign Language Recognition for Icelandic Sign Language: Experiments in a Low-resource SettingFinnur \'Ag\'ust Ingimundarson, Gu{\dh}n\'y Bj\"ork {\TH}orvaldsd\'ottir, Mathias M\"uller, Sarah Ebling
[HN]HN: AI/every.to22h
How to Create Your Own Personal AI BenchmarkGarbage·▲ 2
[BLG]arXiv cs.CL/arxiv.org21h
Domain-Adaptive Pretraining Enhances Water Treatment Semantic Representation for Large-Scale Structured Literature MiningMudi Zhai (UNSW Water Research Centre, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia), Ruihong Qiu (School of Electrical Engineering and Computer Science, The University of Queensland, Brisbane, QLD 4072, Australia), Qingyun Zeng (Microsoft Copilot Studio AI, Redmond, WA 98052, United States, Departments of Mathematics & Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA 19104, United States), T. David Waite (UNSW Water Research Centre, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia), Bing-Jie Ni (UNSW Water Research Centre, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia), Haoran Duan (UNSW Water Research Centre, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia, Department of Civil Engineering, The University of Hong Kong, Pokfulam, Hong Kong SAR, China)
[HN]HN: RAG/vgchartz.com1d
Nearly Half of Age of Empires Dev Edge Laid OffJimmc414·▲ 5
[BLG]GitHub Trending — TypeScript/github.com21h
Jakubantalik/thinking-orbs[HN]HN: LLM/nytimes.com22h
Sadness and Suicidal Behavior Among Teenagers Is Declining, CDC Findslxm·▲ 1
[BLG]arXiv cs.CL/arxiv.org21h
Informed Masking: Structure-Aware Perturbation for Reinforcement Learning in Diffusion Large Language ModelsXiaoyi Yu, Enver Sangineto, Pei Fu, Fiorenzo Parascandolo, Wenhui Tan, Ruikang Zhang, Rita Cucchiara, Ruihua Song, Jian Luan
[HN]HN: RAG/waymo.com22h
Transit rewards (Waymo pays you to take the train)raybb·▲ 12
[BLG]arXiv cs.CL/arxiv.org21h
Impact Is Not Invalidation: Ask About the Claim, Not the DiffAtul Anand