source&pool
A daily wire of long-form journalism, video, and discourse — filed, tagged, and laid out flat.
VOL. I·NO. 01
SUNDAY, SEPTEMBER 27, 2026
X 主题热门3541Hacker News3513CNBC66YahooFinance58aihot56Verge52IGN469to5Mac41MacRumors37Kotaku30Engadget299to5Google27TechCrunch24AndroidAuthority21Eurogamer21NintendoLife20Guardian19PushSquare18ArsTechnica16TechPowerUp15Wccftech15FoxBusiness14Investor'sBusinessDaily13Polygon13BusinessInsider12Mashable12USAToday12Fortune11Gematsu11CBS10Gizmodo10AndroidPolice9bgr9CNN9NPR9VideoGamesChronicle9CNET8MotleyFool8NBC8NewYorkPost8PureXbox8SeekingAlpha8BleepingComputer7Fox7GSMArena7Notebookcheck7Tom'sGuide7VideoCardz7ABC6NintendoEverything6WarhammerCommunity6AndroidCentral5AppleInsider5CoinDesk5GameGPU5GamesIndustry.biz5XBOXWire5PokeBeach5TechSpot5WIRED5AlJazeera4DroidLife4GameInformer4GAMINGbible4HollywoodReporter4InsiderGaming4PetaPixel4PlayStationLifeStyle4RockPaperShotgun4SlashGear4Conversation4Hacker4Yahoo4Aftermath3ChromeUnboxed3Draftsim3Electrek3EventHubs3Futurism3Hackaday3Jalopnik3Lifehacker3Motor13NYT3SamMobile3SouthChinaMorningPost3UploadVR3WhatHi-Fi?3404Media26abcPhiladelphia2AndroidHeadlines2AZFamily2Benzinga2DCRainmaker2Deadline2GameRant2GearPatrol2MP1st2MyNintendo2Nature2CrudeOilPricesToday2PCMag2Pokemon2RoadtoVR2RPGSite2SFGATE2SimsCommunity2Slate2DailyBeast2Register2TimeExtension2TODAY2Variety2WindowsCentral224/7WallSt.180Level1ABC7LosAngeles1ageofempires1Anthropic1Apple1ArizonaSports1Autonocion1AVClub1BellofLostSouls1BostonGlobe1BusinessTimes1BuzzFeed1Yahoo!FinanceCanada1YahooLifestyleCanada1CarandDriver1CarBuzz1CineD1ClaimDepot1Skin.ClubCommunity1consequence1CreativeBloq1YahooCreators1Currently1DailyKos1DaringFireball1DarkHorizons1Decrypt1Deseret1Designboom1Dezeen1DigitalCameraWorld1DirtonDirt1DSOGaming1ForexFactory1FOX13Seattle1franchisetimes1Futurity1GameWorldObserver1garymarcus.substack1GeekWire1GeekyGadgets1Global1GosuGamers1Gothamist1Hackster.io1HouseDigest1HoustonChronicle1iLovetheUpperWestSide1InsideEVs1InterestingEngineering1investor.costco1Invezz1I/OFund1iPhoneinCanada1KCRA1MacObserver1Magic:Gathering1MakeUseOf1Mashed1MLive1MortgageDaily1Motorsport1mtgrocks1NBCBayArea1NBC5Chicago1BloombergLaw1SemiAnalysis1Newsweek1NintendoWire1OregonPublicBroadcasting1OregonLive1PersonaCentral1Phoronix1PickupTruck+SUVTalk1politico.eu1Psyche1qz1Realtor1Richmonder1Road&Track1ScienceDaily1SeattleRed1SeattleTimes1SanFranciscoChronicle1YahooFinanceSingapore1Yahoo1SimpleFlying1GhostHowls1SlippedDisc1SlowBoring1SoraNews241SpaceNews1statnews1svg1TampaBayTimes1YahooTech1the5krunner1DailyMeal1Drive1Hindu1Intercept1TimesofIndia1TMZ1TopGear1TweakTown1YahooFinanceUK1PCMagUK1Vulture1WCVB1WFMZ1WHYY1WKYT1YGOrganization1
  1. 021Hacker NewsSEP · 21English

    94% on AIME with 1B Params

    Limite is a 1-billion parameter language model developed by Paradigma that achieves 94% on AIME and 74.25% on BeyondAIME through efficient training on curated data and mathematical reasoning optimizations. The model is designed for high-throughput mathematical problem-solving with minimal instruction-tuning, though it has limitations in general instruction-following tasks outside mathematics.

    By Mario Prignano; Gabriele Cirillo; Alessio Morosini; Luca Cerovaz; Alessandro Bartolocci; Emanuele Rodolà; Giulio Starace; Francesco Pappone
  2. 022Hacker NewsSEP · 21English

    Show HN: MicroLLM lab – try small LMs in browser, see tok/s and accuracy

    MicroLLM lab lets users run 135M-class language models directly in the browser using WebGPU, with no API keys or external dependencies required. The tool measures model quality and speed tradeoffs for on-device features, supporting models like PetitGPT, SmolLM2, and GPT-2 with local privacy, zero latency, and no ongoing costs.

    By Robss
  3. 023Hacker NewsSEP · 21English

    I made an LLM using 521 Jev models

    Alex built ChatJEVs, a text generation system using 521 Jev decision models that produces language through probabilistic choices rather than traditional neural network training. The system samples words by querying a frozen decision engine across multiple architectural versions, with each word generation requiring multiple batched requests to answer questions about grammar, context, and appropriateness.

    By Alex
  4. 024Hacker NewsSEP · 21English

    Heretic removes restrictions from language models

    Heretic is a tool that removes restrictions from language models to ensure they follow user instructions without limitations.

    By Bluestein
  5. 025Hacker NewsSEP · 20English

    Ternary-Bonsai-8B-Gguf

    Ternary-Bonsai-8B is a 1.58-bit quantized language model in GGUF Q2_0 format, using ternary weights ({-1, 0, +1}) with shared FP16 scales. The model ranks 2nd among compared 6B-9B parameter models despite being 1/8th their size, with implementation support in a custom llama.cpp fork.

    By Bluestein
  6. 026Hacker NewsSEP · 20English

    Jev vs. classical ML. Strong on sentiment: Mixed across tasks

    This analysis compares Jev (a pretrained language model API) against classical machine learning approaches across multiple tasks. Jev shows strong performance on sentiment analysis (96.3% on IMDb vs 88.4% for logistic regression) but struggles with tabular business datasets and produces mixed results on text classification when few-shot examples are added. The comparison reveals that threshold adjustment significantly impacts relative performance and that Jev's strengths are task-dependent rather than universal.

    By theanonymousone