HTML 77.2%
TypeScript 10.5%
Python 9.6%
JavaScript 2.5%
1<!doctype html>2<html class="">3 <head>4 <meta charset="utf-8" />56 <meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=no" />78 <meta name="description" content="Your daily dose of AI research from AK" />910 <meta property="fb:app_id" content="1321688464574422" />1112 <meta name="twitter:card" content="summary_large_image" />1314 <meta name="twitter:site" content="@huggingface" />1516 <meta name="twitter:image" content="https://huggingface.co/front/thumbnails/papers.png" />1718 <meta property="og:title" content="Daily Papers - Hugging Face" />1920 <meta property="og:description" content="Your daily dose of AI research from AK" />2122 <meta property="og:type" content="website" />2324 <meta property="og:url" content="https://huggingface.co/papers" />2526 <meta property="og:image" content="https://huggingface.co/front/thumbnails/papers.png" />2728 <link29 rel="preload"30 href="/front/assets/fonts/source-sans-pro-latin-400-normal.woff2"31 as="font"32 type="font/woff2"33 crossorigin34 />3536 <link37 rel="preload"38 href="/front/assets/fonts/source-sans-pro-latin-600-normal.woff2"39 as="font"40 type="font/woff2"41 crossorigin42 />4344 <link45 rel="preload"46 href="/front/assets/fonts/source-sans-pro-latin-700-normal.woff2"47 as="font"48 type="font/woff2"49 crossorigin50 />5152 <link rel="stylesheet" href="/front/build/kube-8a47f95/style.css" />5354 <link55 rel="preload"56 href="https://cdnjs.cloudflare.com/ajax/libs/KaTeX/0.12.0/katex.min.css"57 as="style"58 onload="59 this.onload = null;60 this.rel = 'stylesheet';61 "62 />6364 <noscript>65 <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/KaTeX/0.12.0/katex.min.css" />66 </noscript>67 <script>const guestTheme = document.cookie.match(/theme=(\w+)/)?.[1]; document.documentElement.classList.toggle('dark', guestTheme === 'dark' || ( (!guestTheme || guestTheme === 'system') && window.matchMedia('(prefers-color-scheme: dark)').matches));</script> 68 <title>Daily Papers - Hugging Face</title>6970 <script defer src="/js/script.js"></script>7172 <script>73 ((window.plausible =74 window.plausible75 || function () {76 (plausible.q = plausible.q || []).push(arguments);77 }),78 (plausible.init =79 plausible.init80 || function (i) {81 plausible.o = i || {};82 }));83 plausible.init({84 customProperties: {85 loggedIn: "false",86 },87 endpoint: "/api/event",88 });89 </script>9091 <script>92 window.hubConfig = {"features":{"signupDisabled":false},"sshGitUrl":"git@hf.co","moonHttpUrl":"https:\/\/huggingface.co","captchaApiKey":"5bd005a4-6ac8-4a86-8e60-53083832ed22","datasetViewerPublicUrl":"https:\/\/datasets-server.huggingface.co","stripePublicKey":"pk_live_x2tdjFXBCvXo2FFmMybezpeM00J6gPCAAc","environment":"production","userAgent":"HuggingFace (production)","spacesIframeDomain":"hf.space","spacesApiUrl":"https:\/\/api.hf.space","logoDev":{"apiUrl":"https:\/\/img.logo.dev\/","apiKey":"pk_UHS2HZOeRnaSOdDp7jbd5w"}};93 window.requestId = "Root=1-6aa45939-0a6eb60c58b2aac8173dace1";94 window.featureFlags = {};95 </script>9697 <script>98 /// Start offset for looping animations that should carry over across page loads instead of99 /// restarting: `animation-delay: var(--animation-phase)`. Only correct for durations dividing 12s.100 document.documentElement.style.setProperty("--animation-phase", -(Date.now() % 12000) + "ms");101 </script>102 <script type="text/javascript" src="https://de5282c3ca0c.edge.sdk.awswaf.com/de5282c3ca0c/526cf06acb0d/challenge.js" defer></script> 103 </head>104 <body class="flex flex-col min-h-dvh bg-white dark:bg-gray-950 text-black DailyPapersPage">105 <!--[--><div class="flex min-h-dvh flex-col"><div class="SVELTE_HYDRATER contents" data-target="ClientErrorCatcher" data-props="{}"><!----></div> <div class="SVELTE_HYDRATER contents" data-target="DeviceProvider" data-props="{}"><!----></div> <div class="SVELTE_HYDRATER contents" data-target="SystemThemeMonitor" data-props="{"isLoggedIn":false}"><!----></div> <!--[0--><div class="SVELTE_HYDRATER contents" data-target="MainHeader" data-props="{"classNames":"","isWide":false,"isZh":false,"canCreateKernels":false,"isPro":false}"><header class="border-b border-gray-100 "><div class="w-full px-4 container flex h-16 items-center"><div class="flex flex-1 items-center"><a class="mr-5 flex flex-none items-center lg:mr-6" href="/"><img alt="Hugging Face's logo" class="w-7 md:mr-2" src="/front/assets/huggingface_logo-noborder.svg"/> <span class="hidden whitespace-nowrap text-lg font-bold md:block">Hugging Face</span></a> <div class="relative flex-1 lg:max-w-sm mr-2 sm:mr-4 md:mr-3 xl:mr-6"><input autocomplete="off" value="" class="w-full dark:bg-gray-950 pl-8 form-input-alt h-9 pr-3 focus:shadow-xl " name="" placeholder="Search models, datasets, users..." spellcheck="false" type="text"/> <!--[0--><svg class="absolute left-2.5 text-gray-400 top-1/2 transform -translate-y-1/2" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M30 28.59L22.45 21A11 11 0 1 0 21 22.45L28.59 30zM5 14a9 9 0 1 1 9 9a9 9 0 0 1-9-9z" fill="currentColor"></path></svg><!--]--> <!--[-1--><!--]--></div><!----> <div class="flex flex-none items-center justify-center p-0.5 place-self-stretch lg:hidden"><button class="relative z-40 flex h-6 w-8 items-center justify-center" type="button"><!--[--><svg width="1em" height="1em" viewBox="0 0 10 10" class="text-xl" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" preserveAspectRatio="xMidYMid meet" fill="currentColor"><path fill-rule="evenodd" clip-rule="evenodd" d="M1.65039 2.9999C1.65039 2.8066 1.80709 2.6499 2.00039 2.6499H8.00039C8.19369 2.6499 8.35039 2.8066 8.35039 2.9999C8.35039 3.1932 8.19369 3.3499 8.00039 3.3499H2.00039C1.80709 3.3499 1.65039 3.1932 1.65039 2.9999ZM1.65039 4.9999C1.65039 4.8066 1.80709 4.6499 2.00039 4.6499H8.00039C8.19369 4.6499 8.35039 4.8066 8.35039 4.9999C8.35039 5.1932 8.19369 5.3499 8.00039 5.3499H2.00039C1.80709 5.3499 1.65039 5.1932 1.65039 4.9999ZM2.00039 6.6499C1.80709 6.6499 1.65039 6.8066 1.65039 6.9999C1.65039 7.1932 1.80709 7.3499 2.00039 7.3499H8.00039C8.19369 7.3499 8.35039 7.1932 8.35039 6.9999C8.35039 6.8066 8.19369 6.6499 8.00039 6.6499H2.00039Z"></path></svg><!--]--> <!--[-1--><!--]--></button> <!--[-1--><!--]--></div><!----></div> <nav aria-label="Main" class="ml-auto hidden lg:block"><ul class="flex items-center gap-x-1 2xl:gap-x-2"><!--[--><li class="hover:text-indigo-700"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/models"><!--[--><svg class="mr-1.5 text-gray-400 group-hover:text-indigo-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path class="uim-quaternary" d="M20.23 7.24L12 12L3.77 7.24a1.98 1.98 0 0 1 .7-.71L11 2.76c.62-.35 1.38-.35 2 0l6.53 3.77c.29.173.531.418.7.71z" opacity=".25" fill="currentColor"></path><path class="uim-tertiary" d="M12 12v9.5a2.09 2.09 0 0 1-.91-.21L4.5 17.48a2.003 2.003 0 0 1-1-1.73v-7.5a2.06 2.06 0 0 1 .27-1.01L12 12z" opacity=".5" fill="currentColor"></path><path class="uim-primary" d="M20.5 8.25v7.5a2.003 2.003 0 0 1-1 1.73l-6.62 3.82c-.275.13-.576.198-.88.2V12l8.23-4.76c.175.308.268.656.27 1.01z" fill="currentColor"></path></svg><!--]--> Models <!--[-1--><!--]--></a></li><li class="hover:text-red-700"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/datasets"><!--[--><svg class="mr-1.5 text-gray-400 group-hover:text-red-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 25 25"><ellipse cx="12.5" cy="5" fill="currentColor" fill-opacity="0.25" rx="7.5" ry="2"></ellipse><path d="M12.5 15C16.6421 15 20 14.1046 20 13V20C20 21.1046 16.6421 22 12.5 22C8.35786 22 5 21.1046 5 20V13C5 14.1046 8.35786 15 12.5 15Z" fill="currentColor" opacity="0.5"></path><path d="M12.5 7C16.6421 7 20 6.10457 20 5V11.5C20 12.6046 16.6421 13.5 12.5 13.5C8.35786 13.5 5 12.6046 5 11.5V5C5 6.10457 8.35786 7 12.5 7Z" fill="currentColor" opacity="0.5"></path><path d="M5.23628 12C5.08204 12.1598 5 12.8273 5 13C5 14.1046 8.35786 15 12.5 15C16.6421 15 20 14.1046 20 13C20 12.8273 19.918 12.1598 19.7637 12C18.9311 12.8626 15.9947 13.5 12.5 13.5C9.0053 13.5 6.06886 12.8626 5.23628 12Z" fill="currentColor"></path></svg><!--]--> Datasets <!--[-1--><!--]--></a></li><li class="hover:text-blue-700"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/spaces"><!--[--><svg class="mr-1.5 text-gray-400 group-hover:text-blue-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 25 25"><path opacity=".5" d="M6.016 14.674v4.31h4.31v-4.31h-4.31ZM14.674 14.674v4.31h4.31v-4.31h-4.31ZM6.016 6.016v4.31h4.31v-4.31h-4.31Z" fill="currentColor"></path><path opacity=".75" fill-rule="evenodd" clip-rule="evenodd" d="M3 4.914C3 3.857 3.857 3 4.914 3h6.514c.884 0 1.628.6 1.848 1.414a5.171 5.171 0 0 1 7.31 7.31c.815.22 1.414.964 1.414 1.848v6.514A1.914 1.914 0 0 1 20.086 22H4.914A1.914 1.914 0 0 1 3 20.086V4.914Zm3.016 1.102v4.31h4.31v-4.31h-4.31Zm0 12.968v-4.31h4.31v4.31h-4.31Zm8.658 0v-4.31h4.31v4.31h-4.31Zm0-10.813a2.155 2.155 0 1 1 4.31 0 2.155 2.155 0 0 1-4.31 0Z" fill="currentColor"></path><path opacity=".25" d="M16.829 6.016a2.155 2.155 0 1 0 0 4.31 2.155 2.155 0 0 0 0-4.31Z" fill="currentColor"></path></svg><!--]--> Spaces <!--[-1--><!--]--></a></li><li class="hover:text-blue-800 dark:text-blue-400 dark:hover:text-blue-300 max-xl:hidden"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/storage"><!--[--><svg class=" mr-1.5 text-gray-400 text-blue-600! dark:text-blue-500! svelte-11u2e10" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 48 48" fill="none"><!--[-1--><!--]--><path opacity="0.25" d="M36 4H12C7.58172 4 4 7.58172 4 12V36C4 40.4183 7.58172 44 12 44H36C40.4183 44 44 40.4183 44 36V12C44 7.58172 40.4183 4 36 4Z" fill="currentColor" class="svelte-11u2e10"></path><path opacity="0.5" d="M31 11H17C13.6863 11 11 13.6863 11 17V31C11 34.3137 13.6863 37 17 37H31C34.3137 37 37 34.3137 37 31V17C37 13.6863 34.3137 11 31 11Z" fill="currentColor" class="svelte-11u2e10"></path><path d="M27 18H21C19.3431 18 18 19.3431 18 21V27C18 28.6569 19.3431 30 21 30H27C28.6569 30 30 28.6569 30 27V21C30 19.3431 28.6569 18 27 18Z" fill="currentColor" class="svelte-11u2e10"></path></svg><!--]--> Buckets <!--[0--><span class="ml-1.5 translate-y-px rounded bg-blue-600/15 px-1 py-px text-[.65rem] font-bold uppercase leading-tight text-blue-600 dark:bg-blue-500/20 dark:text-blue-300">new</span><!--]--></a></li><li class="hover:text-yellow-700"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/docs"><!--[--><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" class="mr-1.5 text-gray-400 group-hover:text-yellow-500" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 16 16"><path d="m2.28 3.7-.3.16a.67.67 0 0 0-.34.58v8.73l.01.04.02.07.01.04.03.06.02.04.02.03.04.06.05.05.04.04.06.04.06.04.08.04.08.02h.05l.07.02h.11l.04-.01.07-.02.03-.01.07-.03.22-.12a5.33 5.33 0 0 1 5.15.1.67.67 0 0 0 .66 0 5.33 5.33 0 0 1 5.33 0 .67.67 0 0 0 1-.58V4.36a.67.67 0 0 0-.34-.5l-.3-.17v7.78a.63.63 0 0 1-.87.59 4.9 4.9 0 0 0-4.35.35l-.65.39a.29.29 0 0 1-.15.04.29.29 0 0 1-.16-.04l-.65-.4a4.9 4.9 0 0 0-4.34-.34.63.63 0 0 1-.87-.59V3.7Z" fill="currentColor" class="dark:opacity-40"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M8 3.1a5.99 5.99 0 0 0-5.3-.43.66.66 0 0 0-.42.62v8.18c0 .45.46.76.87.59a4.9 4.9 0 0 1 4.34.35l.65.39c.05.03.1.04.16.04.05 0 .1-.01.15-.04l.65-.4a4.9 4.9 0 0 1 4.35-.34.63.63 0 0 0 .86-.59V3.3a.67.67 0 0 0-.41-.62 5.99 5.99 0 0 0-5.3.43l-.3.17L8 3.1Zm.73 1.87a.43.43 0 1 0-.86 0v5.48a.43.43 0 0 0 .86 0V4.97Z" fill="currentColor" class="opacity-40 dark:opacity-100"></path><path d="M8.73 4.97a.43.43 0 1 0-.86 0v5.48a.43.43 0 1 0 .86 0V4.96Z" fill="currentColor" class="dark:opacity-40"></path></svg><!--]--> Docs <!--[-1--><!--]--></a></li><li class="hover:text-black dark:hover:text-white max-2xl:hidden"><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/enterprise"><!--[--><svg class="mr-1.5 text-gray-400 group-hover:text-black dark:group-hover:text-white" xmlns="http://www.w3.org/2000/svg" fill="none" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill-rule="evenodd" clip-rule="evenodd" d="M4.9 1.35a3.16 3.16 0 0 0-2.8 2.07L.37 8.58C0 9.71.7 10.65 1.86 10.65H7.3a3.2 3.2 0 0 0 2.84-2.07l1.67-5.16c.36-1.13-.3-2.07-1.46-2.07H4.91Zm.4 2.07L3.57 8.47h3.57l.36-1.12H5.4l.28-.91h1.75l.4-1.1H6.07l.3-.83h2l.36-1.1H5.27h.04Z" fill="currentColor"></path></svg><!--]--> Enterprise <!--[-1--><!--]--></a></li><!--]--> <li><a class="group flex items-center px-2 py-0.5 dark:text-gray-300 dark:hover:text-gray-100" href="/pricing">Pricing</a></li> <li><div class="relative group"><button class="px-2 py-0.5 hover:text-gray-500 dark:hover:text-gray-600 flex items-center " type="button"><!--[-1--><!--[0--><!--[--><svg class=" text-gray-500 w-5 group-hover:text-gray-400 dark:text-gray-300 dark:group-hover:text-gray-100" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 32 18" preserveAspectRatio="xMidYMid meet"><path fill-rule="evenodd" clip-rule="evenodd" d="M14.4504 3.30221C14.4504 2.836 14.8284 2.45807 15.2946 2.45807H28.4933C28.9595 2.45807 29.3374 2.836 29.3374 3.30221C29.3374 3.76842 28.9595 4.14635 28.4933 4.14635H15.2946C14.8284 4.14635 14.4504 3.76842 14.4504 3.30221Z" fill="currentColor"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M14.4504 9.00002C14.4504 8.53382 14.8284 8.15588 15.2946 8.15588H28.4933C28.9595 8.15588 29.3374 8.53382 29.3374 9.00002C29.3374 9.46623 28.9595 9.84417 28.4933 9.84417H15.2946C14.8284 9.84417 14.4504 9.46623 14.4504 9.00002Z" fill="currentColor"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M14.4504 14.6978C14.4504 14.2316 14.8284 13.8537 15.2946 13.8537H28.4933C28.9595 13.8537 29.3374 14.2316 29.3374 14.6978C29.3374 15.164 28.9595 15.542 28.4933 15.542H15.2946C14.8284 15.542 14.4504 15.164 14.4504 14.6978Z" fill="currentColor"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M1.94549 6.87377C2.27514 6.54411 2.80962 6.54411 3.13928 6.87377L6.23458 9.96907L9.32988 6.87377C9.65954 6.54411 10.194 6.54411 10.5237 6.87377C10.8533 7.20343 10.8533 7.73791 10.5237 8.06756L6.23458 12.3567L1.94549 8.06756C1.61583 7.73791 1.61583 7.20343 1.94549 6.87377Z" fill="currentColor"></path></svg><!--]--><!--]--> <!--]--> <!--[-1--><!--]--></button> <!--[0--><nav class="sr-only" inert=""><!--[-1--><!--]--> <ul><!--[0--><li><div class="bg-linear-to-r col-span-full flex items-center justify-between whitespace-nowrap to-white font-semibold dark:to-gray-925 px-4 py-0.5 text-blue-800 from-blue-50 dark:text-blue-100 dark:from-blue-900"><!--[--><!---->Website<!--]--></div><!----> <ul><!--[--><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/tasks"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path class="uim-tertiary" d="M15.273 18.728A6.728 6.728 0 1 1 22 11.999V12a6.735 6.735 0 0 1-6.727 6.728z" opacity=".5" fill="currentColor"></path><path class="uim-primary" d="M8.727 18.728A6.728 6.728 0 1 1 15.455 12a6.735 6.735 0 0 1-6.728 6.728z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Tasks</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/chat"><!--[-1--><!--[0--><!--[--><svg width=".8em" height=".8em" class="mr-1.5 flex-none text-gray-400 dark:invert" viewBox="0 0 12 12" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M6 12a6 6 0 1 1 5.374-3.33.976.976 0 0 0-.081.692l.36 1.337a.78.78 0 0 1-.955.954l-1.336-.36a.977.977 0 0 0-.69.081 6.001 6.001 0 0 1-2.673.63V12Z" fill="#000"></path><path d="M2 6a.675.675 0 1 1 1.35 0 2.476 2.476 0 0 0 4.952 0 .675.675 0 1 1 1.35 0A3.826 3.826 0 1 1 2 6Z" fill="#fff"></path></svg><!--]--><!--]--> <span class="truncate">HuggingChat</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/collections"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" width="1em" height="1em" aria-hidden="true" focusable="false" role="img" viewBox="0 0 12 13" fill="none" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><rect x="2" y="2.49902" width="8" height="3.76425" rx="1.16774" fill="currentColor" fill-opacity="0.4"></rect><rect x="6.21875" y="6.7334" width="3.78055" height="3.76425" rx="1.16774" fill="currentColor" fill-opacity="0.7"></rect><rect x="2" y="6.73438" width="3.78055" height="3.76425" rx="1.16774" fill="currentColor" fill-opacity="0.5"></rect></svg><!--]--><!--]--> <span class="truncate">Collections</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/languages"><!--[-1--><!--[0--><!--[--><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" class="mr-1.5 flex-none text-gray-400" preserveAspectRatio="xMidYMid meet" width="1em" height="1em" viewBox="0 0 10 10"><path fill-rule="evenodd" clip-rule="evenodd" d="M0.625 5C0.625 6.16032 1.08594 7.27312 1.90641 8.09359C2.72688 8.91406 3.83968 9.375 5 9.375C6.16032 9.375 7.27312 8.91406 8.09359 8.09359C8.91406 7.27312 9.375 6.16032 9.375 5C9.375 3.83968 8.91406 2.72688 8.09359 1.90641C7.27312 1.08594 6.16032 0.625 5 0.625C3.83968 0.625 2.72688 1.08594 1.90641 1.90641C1.08594 2.72688 0.625 3.83968 0.625 5ZM7.64365 7.48027C7.61734 7.50832 7.59054 7.53598 7.56326 7.56326C7.13828 7.98824 6.61864 8.2968 6.0539 8.46842C6.29802 8.11949 6.49498 7.64804 6.63475 7.09483C7.00845 7.18834 7.35014 7.3187 7.64365 7.48027ZM8.10076 6.87776C8.37677 6.42196 8.55005 5.90894 8.60556 5.37499H6.86808C6.85542 5.71597 6.82551 6.04557 6.77971 6.35841C7.25309 6.47355 7.68808 6.6414 8.062 6.85549C8.07497 6.86283 8.08789 6.87025 8.10076 6.87776ZM6.03795 6.22536C6.07708 5.95737 6.1044 5.67232 6.11705 5.37499H3.88295C3.89666 5.69742 3.92764 6.00542 3.9722 6.29287C4.37075 6.21726 4.79213 6.17749 5.224 6.17749C5.50054 6.17749 5.77294 6.19376 6.03795 6.22536ZM4.1261 7.02673C4.34894 7.84835 4.68681 8.375 5 8.375C5.32122 8.375 5.66839 7.82101 5.8908 6.963C5.67389 6.93928 5.45082 6.92699 5.224 6.92699C4.84316 6.92699 4.47332 6.96176 4.1261 7.02673ZM3.39783 7.21853C3.53498 7.71842 3.72038 8.14579 3.9461 8.46842C3.42141 8.30898 2.93566 8.03132 2.52857 7.65192C2.77253 7.48017 3.06711 7.33382 3.39783 7.21853ZM3.23916 6.48077C3.18263 6.13193 3.14625 5.76074 3.13192 5.37499H1.39444C1.4585 5.99112 1.67936 6.57938 2.03393 7.08403C2.3706 6.83531 2.78055 6.63162 3.23916 6.48077ZM1.39444 4.62499H3.13192C3.14615 4.24204 3.18211 3.87344 3.23794 3.52681C2.77814 3.37545 2.36731 3.17096 2.03024 2.92123C1.67783 3.42469 1.45828 4.011 1.39444 4.62499ZM2.5237 2.35262C2.76812 2.52552 3.06373 2.67281 3.39584 2.78875C3.53318 2.28573 3.71928 1.85578 3.9461 1.53158C3.41932 1.69166 2.93178 1.97089 2.5237 2.35262ZM3.97101 3.71489C3.92709 4.00012 3.89654 4.30547 3.88295 4.62499H6.11705C6.10453 4.33057 6.07761 4.04818 6.03909 3.78248C5.77372 3.81417 5.50093 3.83049 5.224 3.83049C4.79169 3.83049 4.3699 3.79065 3.97101 3.71489ZM5.8928 3.04476C5.67527 3.06863 5.45151 3.08099 5.224 3.08099C4.84241 3.08099 4.47186 3.04609 4.12405 2.98086C4.34686 2.1549 4.68584 1.625 5 1.625C5.32218 1.625 5.67048 2.18233 5.8928 3.04476ZM6.78083 3.6493C6.826 3.95984 6.85552 4.28682 6.86808 4.62499H8.60556C8.55029 4.09337 8.37827 3.58251 8.10436 3.1282C8.0903 3.1364 8.07618 3.14449 8.062 3.15249C7.68838 3.36641 7.25378 3.53417 6.78083 3.6493ZM7.64858 2.52499C7.35446 2.68754 7.0117 2.81868 6.63664 2.91268C6.49676 2.35623 6.29913 1.88209 6.0539 1.53158C6.61864 1.7032 7.13828 2.01176 7.56326 2.43674C7.59224 2.46572 7.62068 2.49514 7.64858 2.52499Z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Languages</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/organizations"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path class="uim-tertiary" d="M12 18a3.5 3.5 0 1 1 3.5-3.5A3.504 3.504 0 0 1 12 18z" opacity=".5" fill="currentColor"></path><path class="uim-primary" d="M14.64 16.772a3.452 3.452 0 0 1-5.28 0A4.988 4.988 0 0 0 7 21a1 1 0 0 0 1 1h8a1 1 0 0 0 1-1a4.988 4.988 0 0 0-2.36-4.228z" fill="currentColor"></path><path class="uim-tertiary" d="M21 12a.996.996 0 0 1-.664-.252L12 4.338l-8.336 7.41a1 1 0 0 1-1.328-1.496l9-8a.999.999 0 0 1 1.328 0l9 8A1 1 0 0 1 21 12z" opacity=".5" fill="currentColor"></path><path class="uim-quaternary" d="M12 4.338l-8 7.111V21a1 1 0 0 0 1 1h3a1 1 0 0 1-1-1a4.988 4.988 0 0 1 2.36-4.228A3.469 3.469 0 0 1 8.5 14.5a3.5 3.5 0 0 1 7 0a3.469 3.469 0 0 1-.86 2.272A4.988 4.988 0 0 1 17 21a1 1 0 0 1-1 1h3a1 1 0 0 0 1-1v-9.551z" opacity=".25" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Organizations</span><!--]--><!----></a><!----> <!----></li><!--]--></ul></li> <li><div class="bg-linear-to-r col-span-full flex items-center justify-between whitespace-nowrap to-white font-semibold dark:to-gray-925 px-4 py-0.5 text-yellow-800 from-yellow-50 dark:text-yellow-100 dark:from-yellow-900"><!--[--><!---->Community<!--]--></div><!----> <ul><!--[--><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/blog"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 15 15"><path fill="currentColor" d="M3.417 12.75c-.32 0-.596-.114-.824-.342a1.126 1.126 0 0 1-.343-.825V3.417c0-.321.114-.597.343-.824.23-.228.503-.343.824-.343h8.166c.321 0 .596.114.825.343.228.229.342.503.342.824v8.166c0 .32-.114.596-.342.825a1.12 1.12 0 0 1-.825.342zm1.166-2.333h4.084V9.25H4.583v1.167m0-2.334h5.834V6.917H4.583v1.166m0-2.333h5.834V4.583H4.583V5.75"></path></svg><!--]--><!--]--> <span class="truncate">Blog</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/posts"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 12 12" preserveAspectRatio="xMidYMid meet"><path fill="currentColor" fill-rule="evenodd" d="M3.73 2.4A4.25 4.25 0 1 1 6 10.26H2.17l-.13-.02a.43.43 0 0 1-.3-.43l.01-.06a.43.43 0 0 1 .12-.22l.84-.84A4.26 4.26 0 0 1 3.73 2.4Z" clip-rule="evenodd"></path></svg><!--]--><!--]--> <span class="truncate">Posts</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/papers"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 12 12" preserveAspectRatio="xMidYMid meet" fill="none"><path fill="currentColor" fill-rule="evenodd" d="M8.007 1.814a1.176 1.176 0 0 0-.732-.266H3.088c-.64 0-1.153.512-1.153 1.152v6.803c0 .64.513 1.152 1.153 1.152h5.54c.632 0 1.144-.511 1.144-1.152V3.816c0-.338-.137-.658-.412-.887L8.007 1.814Zm-1.875 1.81c0 .695.55 1.253 1.244 1.253h.983a.567.567 0 0 1 .553.585v4.041c0 .165-.119.302-.283.302h-5.55c-.156 0-.275-.137-.275-.302V2.7a.284.284 0 0 1 .284-.301h2.468a.574.574 0 0 1 .434.19.567.567 0 0 1 .142.395v.64Z" clip-rule="evenodd" fill-opacity=".8"></path><path fill="currentColor" fill-opacity=".2" fill-rule="evenodd" d="M6.132 3.624c0 .695.55 1.253 1.244 1.253h.97a.567.567 0 0 1 .566.585v4.041c0 .165-.119.302-.283.302h-5.55c-.156 0-.275-.137-.275-.302V2.7a.284.284 0 0 1 .284-.301h2.468a.567.567 0 0 1 .576.585v.64Z" clip-rule="evenodd"></path></svg><!--]--><!--]--> <span class="truncate">Daily Papers</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/hardware"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill-rule="evenodd" clip-rule="evenodd" d="M9.92865 7.42863H10.643C10.7377 7.42863 10.8285 7.39098 10.8955 7.32405C10.9625 7.25705 11.0001 7.16619 11.0001 7.07148C11.0001 6.97676 10.9625 6.88591 10.8955 6.81891C10.8285 6.75198 10.7377 6.71433 10.643 6.71433H9.92865V5.28575H10.643C10.7377 5.28575 10.8285 5.24812 10.8955 5.18114C10.9625 5.11417 11.0001 5.02333 11.0001 4.9286C11.0001 4.83388 10.9625 4.74304 10.8955 4.67607C10.8285 4.60909 10.7377 4.57146 10.643 4.57146H9.92865V2.78573C9.92865 2.59628 9.85336 2.4146 9.71943 2.28065C9.5855 2.1467 9.40378 2.07144 9.21435 2.07144H7.42862V1.35715C7.42862 1.26242 7.39098 1.17158 7.32405 1.10461C7.25705 1.03763 7.16619 1 7.07148 1C6.97676 1 6.88591 1.03763 6.81891 1.10461C6.75198 1.17158 6.71433 1.26242 6.71433 1.35715V2.07144H5.28575V1.35715C5.28575 1.26242 5.24812 1.17158 5.18114 1.10461C5.11417 1.03763 5.02333 1 4.9286 1C4.83388 1 4.74304 1.03763 4.67607 1.10461C4.60909 1.17158 4.57146 1.26242 4.57146 1.35715V2.07144H2.78573C2.59628 2.07144 2.4146 2.1467 2.28065 2.28065C2.14669 2.4146 2.07144 2.59628 2.07144 2.78573V4.57146H1.35714C1.26242 4.57146 1.17158 4.60909 1.10461 4.67607C1.03763 4.74304 1 4.83388 1 4.9286C1 5.02333 1.03763 5.11417 1.10461 5.18114C1.17158 5.24812 1.26242 5.28575 1.35714 5.28575H2.07144V6.71433H1.35714C1.26242 6.71433 1.17158 6.75198 1.10461 6.81891C1.03763 6.88591 1 6.97676 1 7.07148C1 7.16619 1.03763 7.25705 1.10461 7.32405C1.17158 7.39098 1.26242 7.42863 1.35714 7.42863H2.07144V9.21435C2.07144 9.40378 2.14669 9.5855 2.28065 9.71943C2.4146 9.85336 2.59628 9.92865 2.78573 9.92865H4.57146V10.6429C4.57146 10.7377 4.60909 10.8285 4.67607 10.8955C4.74304 10.9624 4.83388 11.0001 4.9286 11.0001C5.02333 11.0001 5.11417 10.9624 5.18114 10.8955C5.24812 10.8285 5.28575 10.7377 5.28575 10.6429V9.92865H6.71433V10.6429C6.71433 10.7377 6.75198 10.8285 6.81891 10.8955C6.88591 10.9624 6.97676 11.0001 7.07148 11.0001C7.16619 11.0001 7.25705 10.9624 7.32405 10.8955C7.39098 10.8285 7.42862 10.7377 7.42862 10.6429V9.92865H9.21435C9.40378 9.92865 9.5855 9.85336 9.71943 9.71943C9.85336 9.5855 9.92865 9.40378 9.92865 9.21435V7.42863Z" fill="currentColor" fill-opacity="0.4"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M5 4C4.44771 4 4 4.44771 4 5V7C4 7.55227 4.44771 8 5 8H7C7.55227 8 8 7.55227 8 7V5C8 4.44771 7.55227 4 7 4H5Z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Hardware</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/learn"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 32 32"><path fill-rule="evenodd" clip-rule="evenodd" d="M5.44971 14.5964L5.44977 19.5221L5.44971 19.5321L5.44977 19.5422V19.6062H5.4533C5.59676 21.0844 10.0242 22.2706 15.465 22.2706C20.9057 22.2706 25.3332 21.0844 25.4767 19.6062H25.4803L25.4802 14.8524L15.7373 17.5754L5.44971 14.5964Z" fill="currentColor" fill-opacity="0.8"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M1.16846 11.7412L15.7374 15.9487L30.832 11.7412V13.3494H30.8069L30.832 13.3566L15.7374 17.5754L1.16846 13.3566L1.1927 13.3494H1.16846V11.7412ZM1.16846 11.7412V11.7412L1.16849 11.7412L1.16846 11.7412ZM1.16846 11.7412V11.6674H1.41519L1.16846 11.7412ZM30.5764 11.6674H30.832V11.7412L30.5764 11.6674Z" fill="currentColor" fill-opacity="0.5"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M3.02827 18.7976C3.398 17.0129 4.40214 15.3593 5.21357 14.3777L5.68497 14.7674C4.91639 15.6971 3.97099 17.2621 3.62717 18.9216C3.28615 20.5676 3.53867 22.2742 5.1103 23.5887C5.88204 24.2342 6.92784 24.5392 8.1691 24.6236C9.40985 24.7079 10.8132 24.5697 12.2706 24.3583C12.9462 24.2603 13.6292 24.1473 14.3104 24.0346C14.3626 24.0259 14.4148 24.0173 14.467 24.0087C15.1987 23.8877 15.9275 23.7684 16.6346 23.6714C18.0439 23.4782 19.3924 23.3696 20.5369 23.5201L20.4572 24.1265C19.3982 23.9872 18.1177 24.0854 16.7176 24.2774C16.0201 24.373 15.2991 24.491 14.5667 24.6121C14.5143 24.6208 14.4618 24.6295 14.4093 24.6382C13.7292 24.7507 13.0404 24.8647 12.3583 24.9636C10.8913 25.1764 9.43547 25.3227 8.12763 25.2338C6.8203 25.145 5.62813 24.8191 4.71791 24.0579C2.92563 22.5589 2.65573 20.5957 3.02827 18.7976Z" fill="currentColor"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M20.4304 24.1128L20.5291 23.5358L28.0463 24.2759L27.9888 24.8599L27.966 24.8577L27.8807 25.3377L27.8992 25.3422L27.7593 25.9121L20.4304 24.1128Z" fill="currentColor" fill-opacity="0.25"></path><path d="M15.7374 15.9487L1.16846 11.7412L15.7374 7.38605L30.832 11.7412L15.7374 15.9487Z" fill="currentColor" fill-opacity="0.25"></path></svg><!--]--><!--]--> <span class="truncate">Learn</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/join/discord" target="_blank"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 32 32" preserveAspectRatio="xMidYMid meet"><path d="M25.6933 7.10665C23.92 6.27998 22 5.67998 20 5.33331C19.9825 5.33275 19.965 5.33604 19.9489 5.34295C19.9328 5.34987 19.9183 5.36023 19.9067 5.37331C19.6667 5.81331 19.3867 6.38665 19.2 6.82665C17.0787 6.50665 14.9213 6.50665 12.8 6.82665C12.6133 6.37331 12.3333 5.81331 12.08 5.37331C12.0667 5.34665 12.0267 5.33331 11.9867 5.33331C9.98666 5.67998 8.07999 6.27998 6.29333 7.10665C6.27999 7.10665 6.26666 7.11998 6.25333 7.13331C2.62666 12.56 1.62666 17.84 2.11999 23.0666C2.11999 23.0933 2.13333 23.12 2.15999 23.1333C4.55999 24.8933 6.86666 25.96 9.14666 26.6666C9.18666 26.68 9.22666 26.6666 9.23999 26.64C9.77333 25.9066 10.2533 25.1333 10.6667 24.32C10.6933 24.2666 10.6667 24.2133 10.6133 24.2C9.85333 23.9066 9.13333 23.56 8.42666 23.16C8.37333 23.1333 8.37333 23.0533 8.41333 23.0133C8.55999 22.9066 8.70666 22.7866 8.85333 22.68C8.87999 22.6533 8.91999 22.6533 8.94666 22.6666C13.5333 24.76 18.48 24.76 23.0133 22.6666C23.04 22.6533 23.08 22.6533 23.1067 22.68C23.2533 22.8 23.4 22.9066 23.5467 23.0266C23.6 23.0666 23.6 23.1466 23.5333 23.1733C22.84 23.5866 22.1067 23.92 21.3467 24.2133C21.2933 24.2266 21.28 24.2933 21.2933 24.3333C21.72 25.1466 22.2 25.92 22.72 26.6533C22.76 26.6666 22.8 26.68 22.84 26.6666C25.1333 25.96 27.44 24.8933 29.84 23.1333C29.8667 23.12 29.88 23.0933 29.88 23.0666C30.4667 17.0266 28.9067 11.7866 25.7467 7.13331C25.7333 7.11998 25.72 7.10665 25.6933 7.10665V7.10665ZM11.36 19.88C9.98666 19.88 8.83999 18.6133 8.83999 17.0533C8.83999 15.4933 9.95999 14.2266 11.36 14.2266C12.7733 14.2266 13.8933 15.5066 13.88 17.0533C13.88 18.6133 12.76 19.88 11.36 19.88ZM20.6533 19.88C19.28 19.88 18.1333 18.6133 18.1333 17.0533C18.1333 15.4933 19.2533 14.2266 20.6533 14.2266C22.0667 14.2266 23.1867 15.5066 23.1733 17.0533C23.1733 18.6133 22.0667 19.88 20.6533 19.88Z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Discord</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="https://discuss.huggingface.co/" target="_blank"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 " xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" focusable="false" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path d="M12.077 3C7.149 3 3 6.96 3 11.843V21l9.075-.01c4.928 0 8.925-4.11 8.925-8.993C21 7.113 17 3 12.077 3zm3.92 12.859a5.568 5.568 0 0 1-6.102 1.043l-3.595.805l1.001-3.192a5.435 5.435 0 0 1 .11-5.415a5.55 5.55 0 0 1 4.753-2.678v.001h.006a5.533 5.533 0 0 1 5.131 3.438a5.442 5.442 0 0 1-1.304 5.998z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Forum</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="https://github.com/huggingface" target="_blank"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 text-sm" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">GitHub</span><!--]--><!----></a><!----> <!----></li><!--]--></ul></li> <li><div class="bg-linear-to-r col-span-full flex items-center justify-between whitespace-nowrap to-white font-semibold dark:to-gray-925 px-4 py-0.5 text-pink-800 from-pink-50 dark:text-gray-200 dark:from-pink-900"><!--[--><!---->Solutions<!--]--></div><!----> <ul><!--[--><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/enterprise"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" fill="none" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill-rule="evenodd" clip-rule="evenodd" d="M4.9 1.35a3.16 3.16 0 0 0-2.8 2.07L.37 8.58C0 9.71.7 10.65 1.86 10.65H7.3a3.2 3.2 0 0 0 2.84-2.07l1.67-5.16c.36-1.13-.3-2.07-1.46-2.07H4.91Zm.4 2.07L3.57 8.47h3.57l.36-1.12H5.4l.28-.91h1.75l.4-1.1H6.07l.3-.83h2l.36-1.1H5.27h.04Z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Team & Enterprise</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/pro"><!--[-1--><!--[0--><!--[--><!--$s1--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" viewBox="0 0 12 12" fill="none"><path d="M6.48 1.26c0 1.55.67 2.58 1.5 3.24.86.68 1.9 1 2.58 1.07v.86A5.3 5.3 0 0 0 7.99 7.5a3.95 3.95 0 0 0-1.51 3.24h-.96c0-1.55-.67-2.58-1.5-3.24a5.3 5.3 0 0 0-2.58-1.07v-.86A5.3 5.3 0 0 0 4.01 4.5a3.95 3.95 0 0 0 1.51-3.24h.96Z" fill="url(#pro-sparkle-gradient-s1)"></path><defs><linearGradient id="pro-sparkle-gradient-s1" x1="3.37" y1="3.43" x2="8.14" y2="8.9" gradientUnits="userSpaceOnUse"><stop stop-color="#FF0789"></stop><stop offset=".63" stop-color="#21DE75"></stop><stop offset="1" stop-color="#FF8D00"></stop></linearGradient></defs></svg><!--]--><!--]--> <span class="truncate">Hugging Face PRO</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/support"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 25 24"><path d="M12.6213 22.4475C12.7096 22.4825 12.8038 22.5003 12.8988 22.5C13.068 22.4991 13.2318 22.4409 13.3638 22.335L17.1138 19.335C17.2022 19.2652 17.2737 19.1763 17.323 19.0751C17.3724 18.9738 17.3983 18.8627 17.3988 18.75V13.0575L20.2338 10.23C21.0025 9.46567 21.6117 8.5563 22.0263 7.55467C22.4409 6.55304 22.6524 5.47907 22.6488 4.39505V3.00005C22.6488 2.60222 22.4908 2.22069 22.2095 1.93939C21.9281 1.65808 21.5466 1.50005 21.1488 1.50005H19.7538C18.6698 1.49639 17.5958 1.70798 16.5942 2.12254C15.5925 2.5371 14.6832 3.14638 13.9188 3.91505L11.0913 6.75005H5.39879C5.28552 6.75146 5.17405 6.77851 5.07273 6.82917C4.97141 6.87983 4.88288 6.95278 4.81379 7.04255L1.81379 10.7925C1.73008 10.8963 1.67553 11.0205 1.65576 11.1523C1.63599 11.2841 1.6517 11.4188 1.70129 11.5425C1.75027 11.6654 1.83088 11.7732 1.93494 11.8548C2.03899 11.9365 2.1628 11.9892 2.29379 12.0075L7.54379 12.7575L11.4063 16.605L12.1563 21.855C12.1747 21.986 12.2274 22.1098 12.309 22.2139C12.3907 22.318 12.4984 22.3986 12.6213 22.4475Z" opacity=".5" fill="currentColor"></path><path d="M11.0837 10.9416L5.08569 16.9396L7.20363 19.0576L13.2017 13.0595L11.0837 10.9416Z" opacity="1" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Enterprise Support</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/inference/models"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill-rule="evenodd" clip-rule="evenodd" d="M7.587 8.498a.613.613 0 0 0 .812-.915l-.004-.004a.61.61 0 0 0-.27-.157l-.538 1.076Zm.935-1.871a.612.612 0 0 0 .211.037h.006a.613.613 0 0 0 .33-1.13l-.547 1.093Zm.057-2.531a.613.613 0 0 0-1.046-.444l-.005.005a.61.61 0 0 0-.178.402h.969a.93.93 0 0 1 .26.037ZM3.57 5.518l-.549 1.097a.613.613 0 0 1 .24-1.177h.007a.61.61 0 0 1 .302.08Zm.921-1.842L3.979 4.7a.61.61 0 0 1-.373-.177l-.004-.004a.613.613 0 0 1 .866-.867l.005.005a.619.619 0 0 1 .018.019Zm-1.067 4.29c.112.049.237.077.37.077h.857a.61.61 0 0 1-.178.403l-.005.004a.613.613 0 0 1-1.044-.484ZM8.121 6.06V6.05c0-.204.099-.384.252-.496l-.252.505ZM3.849 7.43h.381a.615.615 0 0 0-.381 0Zm4.305-2.758H7.77c.124.041.26.041.384 0ZM3.162 2.346a.613.613 0 0 0-.867.866l.002.002a.613.613 0 1 0 .867-.866l-.002-.002ZM5.777 4.96a.613.613 0 0 0-1.029.578.612.612 0 0 0 0 1.024.612.612 0 0 0 1.03.578l.004-.004a.613.613 0 0 0-.094-.944.617.617 0 0 0 0-.284.613.613 0 0 0 .094-.944l-.005-.004Zm.536 1.232a.613.613 0 0 0-.094.944l.005.004a.613.613 0 0 0 1.029-.578.612.612 0 0 0 0-1.024.613.613 0 0 0-1.03-.578l-.003.004a.613.613 0 0 0 .093.944.615.615 0 0 0 0 .284Zm3.391 2.695a.613.613 0 0 0-.867.867l.003.002a.613.613 0 0 0 .866-.867l-.002-.002Zm-8.263-3.45a.613.613 0 0 0 0 1.226h.003a.613.613 0 1 0 0-1.226h-.003Zm9.116 0a.613.613 0 1 0 0 1.226h.003a.613.613 0 1 0 0-1.226h-.003Zm-.85-3.092c.239.239.239.627 0 .866l-.003.002a.613.613 0 0 1-.867-.866l.003-.002c.239-.24.627-.24.866 0ZM3.163 8.888c.24.24.24.627 0 .867l-.002.002a.613.613 0 0 1-.867-.867l.002-.002c.24-.24.627-.24.867 0Z" fill="currentColor" class="text-gray-400 dark:text-gray-500"></path><path d="M5.597 7.736a.306.306 0 0 0-.306-.306H3.795a.306.306 0 0 1-.274-.444l2.415-4.83c.145-.289.58-.186.58.137v2.073c0 .169.138.306.307.306h1.496c.228 0 .376.24.274.444l-2.415 4.83c-.145.289-.58.186-.58-.137V7.736Z" fill="currentColor" class="text-yellow-500"></path></svg><!--]--><!--]--> <span class="truncate">Inference Providers</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/inference-endpoints" target="_blank"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 flex-none text-gray-400 text-gray-400" xmlns="http://www.w3.org/2000/svg" width="1em" height="1em" viewBox="0 0 32 32" fill="none"><path d="M12.3938 16.0389L21.3611 29.9562H11.3396L2.37699 16.0467H2.3692L2.37232 16.0389L2.3692 16.0343H2.37699L11.1901 2.04378H21.2116L12.3938 16.0389Z" fill="currentColor"></path><path d="M24.9907 21.291C27.5533 21.291 29.6308 19.2135 29.6308 16.6509C29.6308 14.0882 27.5533 12.0107 24.9907 12.0107C22.428 12.0107 20.3505 14.0882 20.3505 16.6509C20.3505 19.2135 22.428 21.291 24.9907 21.291Z" fill="currentColor"></path></svg><!--]--><!--]--> <span class="truncate">Inference Endpoints</span><!--]--><!----></a><!----> <!----></li><li><!----><a class="flex w-full cursor-pointer items-center whitespace-nowrap px-3 py-1.5 text-left leading-tight hover:bg-gray-50 dark:hover:bg-gray-800 hover:underline " href="/storage"><!--[-1--><!--[0--><!--[--><svg class=" mr-1.5 flex-none text-gray-400 svelte-11u2e10" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 48 48" fill="none"><!--[-1--><!--]--><path opacity="0.25" d="M36 4H12C7.58172 4 4 7.58172 4 12V36C4 40.4183 7.58172 44 12 44H36C40.4183 44 44 40.4183 44 36V12C44 7.58172 40.4183 4 36 4Z" fill="currentColor" class="svelte-11u2e10"></path><path opacity="0.5" d="M31 11H17C13.6863 11 11 13.6863 11 17V31C11 34.3137 13.6863 37 17 37H31C34.3137 37 37 34.3137 37 31V17C37 13.6863 34.3137 11 31 11Z" fill="currentColor" class="svelte-11u2e10"></path><path d="M27 18H21C19.3431 18 18 19.3431 18 21V27C18 28.6569 19.3431 30 21 30H27C28.6569 30 30 28.6569 30 27V21C30 19.3431 28.6569 18 27 18Z" fill="currentColor" class="svelte-11u2e10"></path></svg><!--]--><!--]--> <span class="truncate">Storage Buckets</span><!--]--><!----></a><!----> <!----></li><!--]--></ul></li><!----><!--]--></ul></nav><!--]--> <!--[-1--><!--]--></div><!----></li> <li><hr class="h-5 w-0.5 border-none bg-gray-100 dark:bg-gray-800"/></li> <!--[-1--><li><a class="block cursor-pointer whitespace-nowrap px-2 py-0.5 hover:text-gray-500 dark:text-gray-300 dark:hover:text-gray-100" href="/login">Log In</a></li> <!--[0--><li><a class="whitespace-nowrap rounded-full border border-transparent bg-gray-900 px-3 py-1 leading-none text-white hover:border-black hover:bg-white hover:text-black" href="/join">Sign Up</a></li><!--]--><!--]--></ul></nav><!----></div></header><!----></div><!--]--> <!--[-1--><!--]--> <!--[-1--><!--]--><!----> <!--[-1--><!--]--><!----> <!--[0--><div class="SVELTE_HYDRATER contents" data-target="SSOBanner" data-props="{}"><!--[-1--><!--]--><!----></div><!--]--> <!--[-1--><!--]--><!----> <main class="flex flex-1 flex-col"><!--[0--><div class="SVELTE_HYDRATER contents" data-target="DailyPapersBannerSubscribe" data-props="{"isLoggedIn":false}"><div class="-mt-px flex h-9 w-full justify-center text-gray-600"><svg class="hidden h-9 flex-none text-gray-100/80 dark:text-gray-800/40 sm:block" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 110 41"><path fill="currentColor" d="M110 0H0c39.1 0 44 9.6 49 19.5C54.6 30 60 41 108 41h2V0Z"></path></svg> <div class="flex items-center justify-center gap-3 bg-gray-100/80 text-sm dark:bg-gray-800/40 max-sm:flex-1"><!--[0--><div class="rounded-md bg-blue-500/20 px-1 text-xs font-semibold uppercase text-blue-600">new</div> <p class="hidden sm:inline">Get trending papers in your email inbox once a day!</p> <p class="inline sm:hidden">Get trending papers in your email inbox!</p> <!--[0--><a href="/login?next=%2Fpapers" class="btn px-2! text-sm leading-none">Subscribe</a><!--]--><!--]--></div> <svg class="hidden h-9 flex-none text-gray-100/80 dark:text-gray-800/40 sm:block" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 110 41"><path fill="currentColor" d="M0 0h110C70.9 0 66 9.6 61 19.5 55.4 30 50 41 2 41H0V0Z"></path></svg></div><!----></div><!--]--> <div class="SVELTE_HYDRATER contents" data-target="DailyPapers" data-props="{"canSubmit":false,"dateString":"2026-09-11","dailyPapers":[{"paper":{"id":"2609.10715","authors":[{"_id":"6aa36c8547a406da7901e71d","name":"NCP Team","hidden":false},{"_id":"6aa36c8547a406da7901e71e","user":{"_id":"6507c84ec53e1a7f17d996fd","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/U2qzEGvhPk7RjINeaIYid.jpeg","isPro":false,"fullname":"Maximus Cao","user":"Clover-Hill","type":"user","name":"Clover-Hill"},"name":"Jiaqi Cao","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:24.093Z","hidden":false},{"_id":"6aa36c8547a406da7901e71f","name":"Chiyu Chen","hidden":false},{"_id":"6aa36c8547a406da7901e720","name":"Shuang Cheng","hidden":false},{"_id":"6aa36c8547a406da7901e721","name":"Xu Cheng","hidden":false},{"_id":"6aa36c8547a406da7901e722","name":"Beiya Dai","hidden":false},{"_id":"6aa36c8547a406da7901e723","name":"Yufan Feng","hidden":false},{"_id":"6aa36c8547a406da7901e724","name":"Kewen Ge","hidden":false},{"_id":"6aa36c8547a406da7901e725","name":"Ruijun Ge","hidden":false},{"_id":"6aa36c8547a406da7901e726","name":"Jiayi Huang","hidden":false},{"_id":"6aa36c8547a406da7901e727","name":"Yang Jiao","hidden":false},{"_id":"6aa36c8547a406da7901e728","name":"Dahua Lin","hidden":false},{"_id":"6aa36c8547a406da7901e729","name":"Zhouhan Lin","hidden":false},{"_id":"6aa36c8547a406da7901e72a","name":"Yifan Liu","hidden":false},{"_id":"6aa36c8547a406da7901e72b","name":"Yuliang Liu","hidden":false},{"_id":"6aa36c8547a406da7901e72c","name":"Biqing Qi","hidden":false},{"_id":"6aa36c8547a406da7901e72d","name":"Mowen Ruan","hidden":false},{"_id":"6aa36c8547a406da7901e72e","name":"Junzhe Shen","hidden":false},{"_id":"6aa36c8547a406da7901e72f","name":"Yunchong Song","hidden":false},{"_id":"6aa36c8547a406da7901e730","name":"Hao Sun","hidden":false},{"_id":"6aa36c8547a406da7901e731","name":"Zhongbo Tian","hidden":false},{"_id":"6aa36c8547a406da7901e732","name":"Yixuan Wang","hidden":false},{"_id":"6aa36c8547a406da7901e733","user":{"_id":"6609a53bd81d611249ef5266","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6609a53bd81d611249ef5266/h31hdQFl-jhRnO8R6Gr4C.png","isPro":false,"fullname":"Rubin Wei","user":"Rubin-Wei","type":"user","name":"Rubin-Wei"},"name":"Rubin Wei","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.740Z","hidden":false},{"_id":"6aa36c8547a406da7901e734","name":"Jiaxin Xiong","hidden":false},{"_id":"6aa36c8547a406da7901e735","name":"Kangyu Yang","hidden":false},{"_id":"6aa36c8547a406da7901e736","name":"Qian Yao","hidden":false},{"_id":"6aa36c8547a406da7901e737","name":"Qi Zhang","hidden":false},{"_id":"6aa36c8547a406da7901e738","name":"Bowen Zhou","hidden":false}],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction","submittedOnDailyBy":{"_id":"6529f79e802e3d1a4f8ec662","avatarUrl":"/avatars/d05320c370a6497d8792ef5acb563dd5.svg","isPro":false,"fullname":"Yuliang Liu","user":"yuliang03181","type":"user","name":"yuliang03181"},"summary":"We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective while preserving standard token-level autoregressive generation. NCP-ArchPreview builds a latent space by constructing a product-quantized concept vocabulary directly from its hidden states, and subsequently learns to predict future concepts via a dedicated Concept Module. These predicted concepts are then fed back to the token level to guide subsequent generation, with NTP and NCP trained jointly end-to-end. We scale this architecture to 8.9B parameters and train it on 5.73T tokens from the Dolma-3 dataset, marking the largest demonstration of a latent-space language model to date. Remarkably, by consuming only 51.3% of the total training tokens, NCP-ArchPreview achieves the final pretraining loss of OLMo-3-7B. Following full pretraining, it outperforms OLMo-3-7B by 2.45 points on the downstream macro-average, including a notable 5.99-point gain on GSM8K. Controlled experiments isolate a clear progression of performance gains stemming from both the latent architecture and the NCP objective. Furthermore, utilizing only 85% of the standard computation, NCP-ArchPreview approaches the training loss of a strictly parameter-aligned 8.9B baseline. The learned latent space remains highly valuable after the pretraining stage: updating just the 17M-parameter VQ module yields a novel, lightweight interface for domain adaptation, while a simple injection of concept representations into a DFlash2 drafter improves the mean accepted length by 4.17% with negligible overhead.","upvotes":157,"discussionId":"6aa36c8647a406da7901e739","ai_summary":"NCP-ArchPreview is a large latent-space language model that jointly trains next-token and next-concept prediction to improve pretraining efficiency and downstream performance.","ai_keywords":["latent-space language model","next-token prediction","Next Concept Prediction","product-quantized concept vocabulary","Concept Module","autoregressive generation","domain adaptation","DFlash2 drafter"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"publishedAt":"2026-09-08T20:00:00.000Z","title":"NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction","summary":"We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective while preserving standard token-level autoregressive generation. NCP-ArchPreview builds a latent space by constructing a product-quantized concept vocabulary directly from its hidden states, and subsequently learns to predict future concepts via a dedicated Concept Module. These predicted concepts are then fed back to the token level to guide subsequent generation, with NTP and NCP trained jointly end-to-end. We scale this architecture to 8.9B parameters and train it on 5.73T tokens from the Dolma-3 dataset, marking the largest demonstration of a latent-space language model to date. Remarkably, by consuming only 51.3% of the total training tokens, NCP-ArchPreview achieves the final pretraining loss of OLMo-3-7B. Following full pretraining, it outperforms OLMo-3-7B by 2.45 points on the downstream macro-average, including a notable 5.99-point gain on GSM8K. Controlled experiments isolate a clear progression of performance gains stemming from both the latent architecture and the NCP objective. Furthermore, utilizing only 85% of the standard computation, NCP-ArchPreview approaches the training loss of a strictly parameter-aligned 8.9B baseline. The learned latent space remains highly valuable after the pretraining stage: updating just the 17M-parameter VQ module yields a novel, lightweight interface for domain adaptation, while a simple injection of concept representations into a DFlash2 drafter improves the mean accepted length by 4.17% with negligible overhead.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10715.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6529f79e802e3d1a4f8ec662","avatarUrl":"/avatars/d05320c370a6497d8792ef5acb563dd5.svg","fullname":"Yuliang Liu","name":"yuliang03181","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":4,"isUserFollowing":false},"isAuthorParticipating":false},{"paper":{"id":"2609.11929","authors":[{"_id":"6aa369c447a406da7901e6cd","user":{"_id":"64b4a717aa03b6520839e9b8","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64b4a717aa03b6520839e9b8/Rt3ERG-6BVEA4hAwOz0_I.jpeg","isPro":false,"fullname":"Haiwen Diao","user":"Paranioar","type":"user","name":"Paranioar"},"name":"Haiwen Diao","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.718Z","hidden":false},{"_id":"6aa369c447a406da7901e6ce","user":{"_id":"66bc7862aa7cdcb1c31a1efb","avatarUrl":"/avatars/4c2ab907247fe071ff5cdd71c404ca7c.svg","isPro":false,"fullname":"wang jiahao","user":"TokenWang","type":"user","name":"TokenWang"},"name":"Jiahao Wang","status":"claimed_verified","statusLastChangedAt":"2026-09-11T12:32:35.269Z","hidden":false},{"_id":"6aa369c447a406da7901e6cf","name":"Chenjing Ding","hidden":false},{"_id":"6aa369c447a406da7901e6d0","name":"Hanming Deng","hidden":false},{"_id":"6aa369c447a406da7901e6d1","name":"Jiangnan Chen","hidden":false},{"_id":"6aa369c447a406da7901e6d2","name":"Ruixi Zhang","hidden":false},{"_id":"6aa369c447a406da7901e6d3","user":{"_id":"6583b88d1e065c276dcbc975","avatarUrl":"/avatars/53c3092905a4d0f677b09f410459845e.svg","isPro":false,"fullname":"RW","user":"WRHC","type":"user","name":"WRHC"},"name":"Ruohui Wang","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:26.636Z","hidden":false},{"_id":"6aa369c447a406da7901e6d4","user":{"_id":"647d4f1236e109abce409c3b","avatarUrl":"/avatars/d166f5f8be666e96b522a0a0effd21c4.svg","isPro":false,"fullname":"Wenwen Tong","user":"tongww","type":"user","name":"tongww"},"name":"Wenwen Tong","status":"claimed_verified","statusLastChangedAt":"2026-09-11T16:45:04.547Z","hidden":false},{"_id":"6aa369c447a406da7901e6d5","name":"Xiangyu Fan","hidden":false},{"_id":"6aa369c447a406da7901e6d6","name":"Yubo Wang","hidden":false},{"_id":"6aa369c447a406da7901e6d7","name":"Yue Zhu","hidden":false},{"_id":"6aa369c447a406da7901e6d8","name":"Yuwei Niu","hidden":false},{"_id":"6aa369c447a406da7901e6d9","name":"Zhengqi Bai","hidden":false},{"_id":"6aa369c447a406da7901e6da","name":"Zhiqian Lin","hidden":false},{"_id":"6aa369c447a406da7901e6db","name":"Zhitao Yang","hidden":false},{"_id":"6aa369c447a406da7901e6dc","user":{"_id":"652d06833b5997ed71ce5c46","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/652d06833b5997ed71ce5c46/O_D6bpa5mGxLA7uCjmVCG.jpeg","isPro":false,"fullname":"Zhongang Cai","user":"caizhongang","type":"user","name":"caizhongang"},"name":"Zhongang Cai","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.725Z","hidden":false},{"_id":"6aa369c447a406da7901e6dd","name":"Bo Yang","hidden":false},{"_id":"6aa369c447a406da7901e6de","name":"Chen Feng","hidden":false},{"_id":"6aa369c447a406da7901e6df","name":"Chengguang Lv","hidden":false},{"_id":"6aa369c447a406da7901e6e0","name":"Guangjia Liu","hidden":false},{"_id":"6aa369c447a406da7901e6e1","name":"Guanlin Wang","hidden":false},{"_id":"6aa369c447a406da7901e6e2","name":"Hanyu Zhang","hidden":false},{"_id":"6aa369c447a406da7901e6e3","name":"Haojia Yu","hidden":false},{"_id":"6aa369c447a406da7901e6e4","name":"Hongcan Xiao","hidden":false},{"_id":"6aa369c447a406da7901e6e5","name":"Hongli Wang","hidden":false},{"_id":"6aa369c447a406da7901e6e6","name":"Huan Wu","hidden":false},{"_id":"6aa369c447a406da7901e6e7","name":"Huaping Zhong","hidden":false},{"_id":"6aa369c447a406da7901e6e8","name":"Jian Fang","hidden":false},{"_id":"6aa369c447a406da7901e6e9","name":"Jianan Fan","hidden":false},{"_id":"6aa369c447a406da7901e6ea","name":"Jiaqi Li","hidden":false},{"_id":"6aa369c447a406da7901e6eb","name":"Jiefan Lu","hidden":false},{"_id":"6aa369c447a406da7901e6ec","name":"Jing Zuo","hidden":false},{"_id":"6aa369c447a406da7901e6ed","name":"Jingcheng Ni","hidden":false},{"_id":"6aa369c447a406da7901e6ee","name":"Junxiang Xu","hidden":false},{"_id":"6aa369c447a406da7901e6ef","name":"Linjun Dai","hidden":false},{"_id":"6aa369c447a406da7901e6f0","name":"Mutian Xu","hidden":false},{"_id":"6aa369c447a406da7901e6f1","user":{"_id":"66af2ce2685fc4c1c2c38f8a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/iKBNo2T19s37ZiX-8yxQ_.png","isPro":false,"fullname":"Peishen Yan","user":"Zillionkkk","type":"user","name":"Zillionkkk"},"name":"Peishen Yan","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.732Z","hidden":false},{"_id":"6aa369c447a406da7901e6f2","name":"Penghao Wu","hidden":false},{"_id":"6aa369c447a406da7901e6f3","name":"Ruijie Mao","hidden":false},{"_id":"6aa369c447a406da7901e6f4","name":"Ruisi Wang","hidden":false},{"_id":"6aa369c447a406da7901e6f5","name":"Shihao Bai","hidden":false},{"_id":"6aa369c447a406da7901e6f6","name":"Shuang Yang","hidden":false},{"_id":"6aa369c447a406da7901e6f7","name":"Shuya Yang","hidden":false},{"_id":"6aa369c447a406da7901e6f8","name":"Shuyan Zheng","hidden":false},{"_id":"6aa369c447a406da7901e6f9","name":"Silei Wu","hidden":false},{"_id":"6aa369c447a406da7901e6fa","name":"Siying Li","hidden":false},{"_id":"6aa369c447a406da7901e6fb","name":"Tao Chu","hidden":false},{"_id":"6aa369c447a406da7901e6fc","name":"Tianbo Zhong","hidden":false},{"_id":"6aa369c447a406da7901e6fd","name":"Tongxi Zhou","hidden":false},{"_id":"6aa369c447a406da7901e6fe","name":"Weichao Luo","hidden":false},{"_id":"6aa369c447a406da7901e6ff","name":"Weichen Fan","hidden":false},{"_id":"6aa369c447a406da7901e700","name":"Wenhao Jia","hidden":false},{"_id":"6aa369c447a406da7901e701","name":"Wenjie Gao","hidden":false},{"_id":"6aa369c447a406da7901e702","name":"Xiangli Kong","hidden":false},{"_id":"6aa369c447a406da7901e703","name":"Yan Li","hidden":false},{"_id":"6aa369c447a406da7901e704","name":"Yang Yong","hidden":false},{"_id":"6aa369c447a406da7901e705","name":"Zimo Wen","hidden":false},{"_id":"6aa369c447a406da7901e706","name":"Zixuan Qian","hidden":false},{"_id":"6aa369c447a406da7901e707","name":"Wenxiu Sun","hidden":false},{"_id":"6aa369c447a406da7901e708","name":"Ruihao Gong","hidden":false},{"_id":"6aa369c447a406da7901e709","name":"Quan Wang","hidden":false},{"_id":"6aa369c447a406da7901e70a","name":"Lewei Lu","hidden":false},{"_id":"6aa369c447a406da7901e70b","name":"Lei Yang","hidden":false},{"_id":"6aa369c447a406da7901e70c","name":"Ziwei Liu","hidden":false},{"_id":"6aa369c447a406da7901e70d","name":"Dahua Lin","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"SenseNova-U1.5: Towards Native Unified Visual Intelligence","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"We launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture. We strengthen its visual interface through spatially coherent patch reconstruction and scale its training with carefully curated generation and editing data, improved task formulation, structural prompt enhancement, and native resolutions of up to 4K. For post-training, we optimize specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing, and consolidate their capabilities through multi-expert on-policy distillation. Across extensive evaluations, SenseNova-U1.5 largely advances image fidelity, text rendering, complex composition, multi-reference editing, and interleaved generation, while improving instruction following and preserving subject identity, geometry, and unmodified regions. Despite limited exposure to structured formats in its generation data, SenseNova-U1.5 generalizes effectively to long, complex, and structured visual instructions, further proving that multimodal understanding can transfer to visual planning and creation. Together, these findings position native unified modelling as a promising path towards systems that perceive, reason and create within a fully end-to-end framework. We will open-source training code, including supervised fine-tuning, reinforcement learning, and on-policy distillation.","upvotes":124,"discussionId":"6aa369c547a406da7901e70e","ai_summary":"SenseNova-U1.5 is an 8B native unified multimodal model that performs visual understanding, reasoning, and generation without encoders or VAEs, achieving high fidelity and instruction following through patch reconstruction, curated data, expert optimization, and on-policy distillation.","ai_keywords":["8B-MoT","native unified multimodal model","encoder-free","VAE-free","spatially coherent patch reconstruction","multi-expert on-policy distillation","visual aesthetics","bilingual text rendering","infographic generation","interleaved generation","reinforcement learning"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"publishedAt":"2026-09-09T20:00:00.000Z","title":"SenseNova-U1.5: Towards Native Unified Visual Intelligence","summary":"We launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture. We strengthen its visual interface through spatially coherent patch reconstruction and scale its training with carefully curated generation and editing data, improved task formulation, structural prompt enhancement, and native resolutions of up to 4K. For post-training, we optimize specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing, and consolidate their capabilities through multi-expert on-policy distillation. Across extensive evaluations, SenseNova-U1.5 largely advances image fidelity, text rendering, complex composition, multi-reference editing, and interleaved generation, while improving instruction following and preserving subject identity, geometry, and unmodified regions. Despite limited exposure to structured formats in its generation data, SenseNova-U1.5 generalizes effectively to long, complex, and structured visual instructions, further proving that multimodal understanding can transfer to visual planning and creation. Together, these findings position native unified modelling as a promising path towards systems that perceive, reason and create within a fully end-to-end framework. We will open-source training code, including supervised fine-tuning, reinforcement learning, and on-policy distillation.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11929.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","fullname":"taesiri","name":"taesiri","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":373,"isUserFollowing":false},"isAuthorParticipating":false},{"paper":{"id":"2609.07064","authors":[{"_id":"6aa21457a2aeb74440b1dd87","user":{"_id":"660beaac9b5015f91d6b4308","avatarUrl":"/avatars/38459c92f2bc1b95a3ee43ce0234c9c8.svg","isPro":false,"fullname":"Soohyun Ryu","user":"rsoohyun","type":"user","name":"rsoohyun"},"name":"Soohyun Ryu","status":"claimed_verified","statusLastChangedAt":"2026-09-10T08:45:05.222Z","hidden":false},{"_id":"6aa21457a2aeb74440b1dd88","user":{"_id":"6690a286181e2af45c742dd8","avatarUrl":"/avatars/511d0f86386e3b29a17b445d855b3aef.svg","isPro":false,"fullname":"Sohee Kim","user":"joyhee","type":"user","name":"joyhee"},"name":"Sohee Kim","status":"claimed_verified","statusLastChangedAt":"2026-09-10T08:45:05.229Z","hidden":false},{"_id":"6aa21457a2aeb74440b1dd89","name":"Eunho Yang","hidden":false}],"publishedAt":"2026-09-07T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem","submittedOnDailyBy":{"_id":"660beaac9b5015f91d6b4308","avatarUrl":"/avatars/38459c92f2bc1b95a3ee43ce0234c9c8.svg","isPro":false,"fullname":"Soohyun Ryu","user":"rsoohyun","type":"user","name":"rsoohyun"},"summary":"Large Vision-Language Models (LVLMs) have achieved strong performance on diverse visual tasks, yet their ability to reconstruct and reason about the 3D structure of the scene depicted in 2D images -- referred to as spatial intelligence -- remains limited. Existing approaches attempt to address this gap by using real-scene spatial question answering datasets that require dense geometric annotations. However, constructing such labels is costly, time-consuming, and often noisy due to reliance on external perception modules. In this work, we propose a novel paradigm inspired by human cognitive development: learning foundational spatial skills through structured block-manipulation tasks. We introduce SpatialBlock-15k, a synthetic dataset of 15,000 block-stacking problems covering 3D-to-2D projection, viewpoint transformation, and structural combination. The dataset further incorporates controlled color modulation as visual cues to encourage anchor-based reasoning in visually complex conditions. Experiments demonstrate that LVLMs trained on our dataset through either direct answering or reasoning-based prediction significantly outperform baselines and generalize to real-world spatial tasks, despite the dataset's synthetic and compact nature. Code and data are available at https://github.com/rsoohyun/SpatialBlock.","upvotes":55,"discussionId":"6aa21458a2aeb74440b1dd8a","githubRepo":"https://github.com/rsoohyun/SpatialBlock","githubRepoAddedBy":"user","ai_summary":"Large vision-language models trained on synthetic block-manipulation tasks improve 3D spatial reasoning and generalize to real-world visual tasks.","ai_keywords":["Large Vision-Language Models","spatial intelligence","3D-to-2D projection","viewpoint transformation","structural combination","anchor-based reasoning","reasoning-based prediction"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0,"organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"}},"publishedAt":"2026-09-06T20:00:00.000Z","title":"SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem","summary":"Large Vision-Language Models (LVLMs) have achieved strong performance on diverse visual tasks, yet their ability to reconstruct and reason about the 3D structure of the scene depicted in 2D images -- referred to as spatial intelligence -- remains limited. Existing approaches attempt to address this gap by using real-scene spatial question answering datasets that require dense geometric annotations. However, constructing such labels is costly, time-consuming, and often noisy due to reliance on external perception modules. In this work, we propose a novel paradigm inspired by human cognitive development: learning foundational spatial skills through structured block-manipulation tasks. We introduce SpatialBlock-15k, a synthetic dataset of 15,000 block-stacking problems covering 3D-to-2D projection, viewpoint transformation, and structural combination. The dataset further incorporates controlled color modulation as visual cues to encourage anchor-based reasoning in visually complex conditions. Experiments demonstrate that LVLMs trained on our dataset through either direct answering or reasoning-based prediction significantly outperform baselines and generalize to real-world spatial tasks, despite the dataset's synthetic and compact nature. Code and data are available at https://github.com/rsoohyun/SpatialBlock.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.07064.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"660beaac9b5015f91d6b4308","avatarUrl":"/avatars/38459c92f2bc1b95a3ee43ce0234c9c8.svg","fullname":"Soohyun Ryu","name":"rsoohyun","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.05903","authors":[{"_id":"6aa27ee9a2aeb74440b1e028","user":{"_id":"67f5c7f215e30a165ee73334","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/6bmu8qdzZtvQK37I4RiXH.png","isPro":false,"fullname":"Nanxi Li","user":"andyc03","type":"user","name":"andyc03"},"name":"Nanxi Li","status":"claimed_verified","statusLastChangedAt":"2026-09-10T16:45:04.690Z","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e029","name":"Yingzi Ma","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e02a","name":"Yulong Cao","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e02b","name":"Edward Suh","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e02c","name":"Bo Li","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e02d","name":"Dawn Song","hidden":false},{"_id":"6aa27ee9a2aeb74440b1e02e","name":"Chaowei Xiao","hidden":false}],"publishedAt":"2026-09-05T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents","submittedOnDailyBy":{"_id":"67f5c7f215e30a165ee73334","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/6bmu8qdzZtvQK37I4RiXH.png","isPro":false,"fullname":"Nanxi Li","user":"andyc03","type":"user","name":"andyc03"},"summary":"Large Language Model (LLM) agents are turning language into real-world effects, making safety necessary against both indirect prompt injections and direct harmful requests. System-level safety harnesses add an enforcement layer beyond model-level defenses, but existing harnesses are usually designed once by experts and applied across heterogeneous models and domains. Effective protection is deployment-dependent: models differ in how much enforcement they need before utility declines, while domains differ in the effects, state, and action sequences that must be governed. A harness that is strict enough for one model may over-block another, and a policy that transfers across domains may miss application-specific safety relations.\n We present EvoSafeHarness, a safety-specific optimization framework that synthesizes a deployable harness for a frozen model in a target domain. It jointly searches a natural-language policy and executable code logic, guided by model behavior, domain specifications, and fresh-context adversarial review to reject benchmark-specific rules. Across four agent benchmark families, EvoSafeHarness achieves a stronger safety-utility frontier than fixed expert-designed defenses. On DecodingTrust-Agent, it reduces average attack success rate from 45.6% to 10.0% at a 3.3-point utility cost and achieves the best score in 14 of 15 cells. On AgentDojo, it reaches 82.8% utility at 0.0% ASR, twice CaMeL's utility at the same operating point, and transfers unchanged to unseen AgentDyn suites. It also achieves the best score on Agent-SafetyBench for every victim and keeps mean ASR below 20% under adaptive PAIR attacks with a refinement budget of 16. Analysis shows that domain semantics determine which safety relations and trajectory state are needed, while model and runtime behavior determine how and where those relations should be enforced.","upvotes":36,"discussionId":"6aa27ee9a2aeb74440b1e02f","projectPage":"https://andylinx.github.io/EvoSafeHarness/","githubRepo":"https://github.com/SaFo-Lab/EvoSafeHarness","githubRepoAddedBy":"user","ai_summary":"EvoSafeHarness optimizes deployable safety harnesses by jointly searching natural-language policies and executable logic tailored to a frozen model and target domain, improving safety-utility trade-offs across agent benchmarks.","ai_keywords":["LLM agents","indirect prompt injection","system-level safety harness","EvoSafeHarness","natural-language policy","executable code logic","adversarial review","safety-utility frontier","attack success rate","DecodingTrust-Agent","AgentDojo","PAIR attacks"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":1,"organization":{"_id":"653945b47ba797097a7b4eab","name":"JohnsHopkins","fullname":"Johns Hopkins University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/653944e58e687a41625a4694/qqHzBOarppVrUuZbbjqwh.png"}},"publishedAt":"2026-09-04T20:00:00.000Z","title":"EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents","summary":"Large Language Model (LLM) agents are turning language into real-world effects, making safety necessary against both indirect prompt injections and direct harmful requests. System-level safety harnesses add an enforcement layer beyond model-level defenses, but existing harnesses are usually designed once by experts and applied across heterogeneous models and domains. Effective protection is deployment-dependent: models differ in how much enforcement they need before utility declines, while domains differ in the effects, state, and action sequences that must be governed. A harness that is strict enough for one model may over-block another, and a policy that transfers across domains may miss application-specific safety relations.\n We present EvoSafeHarness, a safety-specific optimization framework that synthesizes a deployable harness for a frozen model in a target domain. It jointly searches a natural-language policy and executable code logic, guided by model behavior, domain specifications, and fresh-context adversarial review to reject benchmark-specific rules. Across four agent benchmark families, EvoSafeHarness achieves a stronger safety-utility frontier than fixed expert-designed defenses. On DecodingTrust-Agent, it reduces average attack success rate from 45.6% to 10.0% at a 3.3-point utility cost and achieves the best score in 14 of 15 cells. On AgentDojo, it reaches 82.8% utility at 0.0% ASR, twice CaMeL's utility at the same operating point, and transfers unchanged to unseen AgentDyn suites. It also achieves the best score on Agent-SafetyBench for every victim and keeps mean ASR below 20% under adaptive PAIR attacks with a refinement budget of 16. Analysis shows that domain semantics determine which safety relations and trajectory state are needed, while model and runtime behavior determine how and where those relations should be enforced.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.05903.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"67f5c7f215e30a165ee73334","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/6bmu8qdzZtvQK37I4RiXH.png","fullname":"Nanxi Li","name":"andyc03","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":2,"isUserFollowing":false},"organization":{"_id":"653945b47ba797097a7b4eab","name":"JohnsHopkins","fullname":"Johns Hopkins University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/653944e58e687a41625a4694/qqHzBOarppVrUuZbbjqwh.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.11317","authors":[{"_id":"6aa3967047a406da7901e7ff","user":{"_id":"69e1eb933ceb9605d4db2007","avatarUrl":"/avatars/d2899498627e36d573965d713ec87c89.svg","isPro":false,"fullname":"Jiayin Chen","user":"cnbird","type":"user","name":"cnbird"},"name":"Jiayin Chen","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.759Z","hidden":false},{"_id":"6aa3967047a406da7901e800","name":"Yicheng Xu","hidden":false},{"_id":"6aa3967047a406da7901e801","user":{"_id":"6a1ecd3fdc5908e8dd1d2fba","avatarUrl":"/avatars/c38bad66752804d41dace2df11b575d8.svg","isPro":false,"fullname":"muwing wang","user":"muwing","type":"user","name":"muwing"},"name":"Muting Wang","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:18.773Z","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Mi-Ripple: Restoring Images Degraded by Iterative AI Editing","submittedOnDailyBy":{"_id":"69e1eb933ceb9605d4db2007","avatarUrl":"/avatars/d2899498627e36d573965d713ec87c89.svg","isPro":false,"fullname":"Jiayin Chen","user":"cnbird","type":"user","name":"cnbird"},"summary":"Iterative reference-conditioned image editing can introduce grid-like and granular textures, commonly described as digital ripple. We present Mi-Ripple, a diagnosis-guided restoration workflow that suppresses this digital ripple while protecting image structure. Mi-Ripple separates periodic lattice artifacts from content-entangled granular texture, then combines selective spectral notching, structure-aware smoothing, and cleaned-reference regeneration. This separation enables low-distortion filtering when artifacts are spectrally isolated and visual reconstruction when filtering would erase legitimate detail. Across fourteen notch-only executions, whole-image residual standard deviation is 0.08--0.44 in CIELAB lightness units. In a paired regeneration example, reference cleaning reduces output debris density by 45\\%. Mi-Ripple links measurable artifact reduction to visibly cleaner generated images, rather than optimizing a spectral score alone.","upvotes":20,"discussionId":"6aa3967147a406da7901e802","projectPage":"https://lab.miyang.cn/ripple/","githubRepo":"https://github.com/miyang-ai/Mi-Ripple","githubRepoAddedBy":"user","ai_summary":"Mi-Ripple reduces digital ripple artifacts in edited images by separating lattice artifacts from texture and applying targeted spectral filtering and reference cleaning.","ai_keywords":["Mi-Ripple","digital ripple","periodic lattice artifacts","spectral notching","structure-aware smoothing","cleaned-reference regeneration","CIELAB lightness"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":32,"organization":{"_id":"6aa3c33df87217ec43aa3ecc","name":"Miyang-AI","fullname":"Miyang Technology (Shanghai) Co., Ltd.","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/69e1eb933ceb9605d4db2007/-cr0jWbvkbFP9qIHCwfXX.png"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Mi-Ripple: Restoring Images Degraded by Iterative AI Editing","summary":"Iterative reference-conditioned image editing can introduce grid-like and granular textures, commonly described as digital ripple. We present Mi-Ripple, a diagnosis-guided restoration workflow that suppresses this digital ripple while protecting image structure. Mi-Ripple separates periodic lattice artifacts from content-entangled granular texture, then combines selective spectral notching, structure-aware smoothing, and cleaned-reference regeneration. This separation enables low-distortion filtering when artifacts are spectrally isolated and visual reconstruction when filtering would erase legitimate detail. Across fourteen notch-only executions, whole-image residual standard deviation is 0.08--0.44 in CIELAB lightness units. In a paired regeneration example, reference cleaning reduces output debris density by 45\\%. Mi-Ripple links measurable artifact reduction to visibly cleaner generated images, rather than optimizing a spectral score alone.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11317.png","numComments":2,"upvoted":false,"submittedBy":{"_id":"69e1eb933ceb9605d4db2007","avatarUrl":"/avatars/d2899498627e36d573965d713ec87c89.svg","fullname":"Jiayin Chen","name":"cnbird","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"6aa3c33df87217ec43aa3ecc","name":"Miyang-AI","fullname":"Miyang Technology (Shanghai) Co., Ltd.","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/69e1eb933ceb9605d4db2007/-cr0jWbvkbFP9qIHCwfXX.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.11412","authors":[{"_id":"6aa397d547a406da7901e805","name":"Haojun Zhang","hidden":false},{"_id":"6aa397d547a406da7901e806","name":"Yi Zou","hidden":false},{"_id":"6aa397d547a406da7901e807","name":"Min Chen","hidden":false},{"_id":"6aa397d547a406da7901e808","user":{"_id":"692fa5d17ff1da99eb783dfb","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/692fa5d17ff1da99eb783dfb/GZ2TD9ua-pigxz5kEHRYX.jpeg","isPro":false,"fullname":"Qize Yu","user":"Skywalker0410","type":"user","name":"Skywalker0410"},"name":"Qize Yu","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.771Z","hidden":false},{"_id":"6aa397d547a406da7901e809","name":"Lianrui Fan","hidden":false},{"_id":"6aa397d547a406da7901e80a","name":"Xini Ding","hidden":false},{"_id":"6aa397d547a406da7901e80b","name":"Hao Li","hidden":false},{"_id":"6aa397d547a406da7901e80c","name":"Shuchang Zhou","hidden":false},{"_id":"6aa397d547a406da7901e80d","name":"Xianming Liu","hidden":false},{"_id":"6aa397d547a406da7901e80e","user":{"_id":"6406db5cd684369027166986","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/Zl-orrGcbY0RbfjfKszn1.jpeg","isPro":false,"fullname":"Shiyu Huang","user":"ShiyuHuang","type":"user","name":"ShiyuHuang"},"name":"Shiyu Huang","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.765Z","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6406db5cd684369027166986/LqtMDn_MJsPpxMH73nkD0.png"],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation","submittedOnDailyBy":{"_id":"6406db5cd684369027166986","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/Zl-orrGcbY0RbfjfKszn1.jpeg","isPro":false,"fullname":"Shiyu Huang","user":"ShiyuHuang","type":"user","name":"ShiyuHuang"},"summary":"Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combinations through short behavioral probes and restores the pruned model through representation alignment, cross-scale distillation, scheduled student-policy supervision, and LoRA finetuning. The language-model backbone remains frozen, while attention LoRA adapters and the tied output embedding adapt during distillation. Training uses the highest-agreement tier from a transcript-consistency pipeline, followed by source reweighting during finetuning. On ten public Chinese--English benchmarks, compressing Qwen3-ASR-0.6B from 18 to 16 audio-encoder layers reduces macro-average error from 5.61% to 5.27%. The 14-layer model reaches 5.75% with 20.7% fewer audio-tower parameters. Under the matched recipe, the 1.7B teacher yields 5.55% mean error, compared with 8.45% for self-distillation, and progressive 18rightarrow14 pruning outperforms direct pruning (5.75% vs. 6.73%). These single-run results establish two practical operating points and show that the accuracy effects vary across benchmarks. Project website: https://xpeng-ai.github.io/x-aut","upvotes":13,"discussionId":"6aa397d647a406da7901e80f","projectPage":"https://xpeng-ai.github.io/x-aut","githubRepo":"https://github.com/XPENG-AI/X-AuT","githubRepoAddedBy":"user","ai_summary":"X-AuT progressively prunes audio-encoder layers in speech large language models and restores accuracy via behavioral probes, representation alignment, cross-scale distillation, and LoRA adaptation.","ai_keywords":["speech large language models","audio-encoder depth","X-AuT","behavioral probes","representation alignment","cross-scale distillation","LoRA finetuning","attention LoRA adapters","transcript-consistency pipeline","source reweighting"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":7,"organization":{"_id":"6aa24a85fc727b9188a95572","name":"XPENG-AI","fullname":"XPENG AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/DAJtDU_HssW_R7WIw0USp.jpeg"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation","summary":"Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combinations through short behavioral probes and restores the pruned model through representation alignment, cross-scale distillation, scheduled student-policy supervision, and LoRA finetuning. The language-model backbone remains frozen, while attention LoRA adapters and the tied output embedding adapt during distillation. Training uses the highest-agreement tier from a transcript-consistency pipeline, followed by source reweighting during finetuning. On ten public Chinese--English benchmarks, compressing Qwen3-ASR-0.6B from 18 to 16 audio-encoder layers reduces macro-average error from 5.61% to 5.27%. The 14-layer model reaches 5.75% with 20.7% fewer audio-tower parameters. Under the matched recipe, the 1.7B teacher yields 5.55% mean error, compared with 8.45% for self-distillation, and progressive 18rightarrow14 pruning outperforms direct pruning (5.75% vs. 6.73%). These single-run results establish two practical operating points and show that the accuracy effects vary across benchmarks. Project website: https://xpeng-ai.github.io/x-aut","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6406db5cd684369027166986/LqtMDn_MJsPpxMH73nkD0.png"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11412.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6406db5cd684369027166986","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/Zl-orrGcbY0RbfjfKszn1.jpeg","fullname":"Shiyu Huang","name":"ShiyuHuang","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":9,"isUserFollowing":false},"organization":{"_id":"6aa24a85fc727b9188a95572","name":"XPENG-AI","fullname":"XPENG AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/DAJtDU_HssW_R7WIw0USp.jpeg"},"isAuthorParticipating":true},{"paper":{"id":"2609.11561","authors":[{"_id":"6aa3de9c47a406da7901e99b","user":{"_id":"6769003931100198233acfc9","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6769003931100198233acfc9/4HOr-exIRu9lyBt8JQWZ2.jpeg","isPro":false,"fullname":"SizheZhao","user":"SizheZhao","type":"user","name":"SizheZhao"},"name":"Sizhe Zhao","status":"claimed_verified","statusLastChangedAt":"2026-09-11T12:32:31.516Z","hidden":false},{"_id":"6aa3de9c47a406da7901e99c","user":{"_id":"63f47b5321eb234ab739e91a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63f47b5321eb234ab739e91a/vWfFNVtMkHl8gieha5PPd.jpeg","isPro":false,"fullname":"Haozhe Xie","user":"hzxie","type":"user","name":"hzxie"},"name":"Haozhe Xie","status":"claimed_verified","statusLastChangedAt":"2026-09-11T12:32:29.662Z","hidden":false},{"_id":"6aa3de9c47a406da7901e99d","name":"Weiyu Zhao","hidden":false},{"_id":"6aa3de9c47a406da7901e99e","name":"Chenchu Zhang","hidden":false},{"_id":"6aa3de9c47a406da7901e99f","name":"Huan Wang","hidden":false},{"_id":"6aa3de9c47a406da7901e9a0","name":"Chenyang Wang","hidden":false},{"_id":"6aa3de9c47a406da7901e9a1","name":"Qinglin Liu","hidden":false},{"_id":"6aa3de9c47a406da7901e9a2","name":"Shengping Zhang","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/63f47b5321eb234ab739e91a/clT7ifvITIQhw6ni64AqX.mp4"],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Memory as Plans: World-Action Modeling with Memory-Grounded Planning","submittedOnDailyBy":{"_id":"63f47b5321eb234ab739e91a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63f47b5321eb234ab739e91a/vWfFNVtMkHl8gieha5PPd.jpeg","isPro":false,"fullname":"Haozhe Xie","user":"hzxie","type":"user","name":"hzxie"},"summary":"Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their combinations, and may therefore lose fine-grained visual evidence or face a trade-off between history coverage and execution efficiency. We introduce MaP-WAM, a Memory-as-Plans framework that decomposes memory-dependent world-action modeling into memory-grounded planning and plan-conditioned execution, and uses long-term multimodal episodic context as planning-time evidence rather than repeatedly conditioning the executor on the full history. MaP-WAM represents memory as completed segment records containing language instructions and sparse visual context, and converts this episodic memory into compact plans comprising the next segment-level language plan and corresponding visual guidance. A World-Action-Progress (WAP) model executes each plan over an unknown duration by jointly predicting action chunks and corresponding execution progress at inference time, calibrating predicted progress through plan-observation alignment for adaptive segment transitions and closed-loop context updates. MaP-WAM keeps the executor context length fixed, while structured attention further enables key-value caching in both planning and execution. MaP-WAM achieves state-of-the-art performance on RMBench with an 83.3% success rate and attains 78.0% success on real-robot tasks, while maintaining approximately constant executor inference latency as task history grows.","upvotes":9,"discussionId":"6aa3de9c47a406da7901e9a3","projectPage":"https://sizhezhao.github.io/projects/MaP-WAM/","githubRepo":"https://github.com/aipixel/MaP-WAM","githubRepoAddedBy":"user","ai_summary":"MaP-WAM improves non-Markovian robotic manipulation by separating memory-grounded planning from plan-conditioned execution, using compact episodic segment records and progress-calibrated action chunks to maintain fixed inference latency.","ai_keywords":["MaP-WAM","Memory-as-Plans","non-Markovian","world-action modeling","episodic memory","segment records","visual guidance","World-Action-Progress model","action chunks","execution progress","structured attention","key-value caching"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Memory as Plans: World-Action Modeling with Memory-Grounded Planning","summary":"Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their combinations, and may therefore lose fine-grained visual evidence or face a trade-off between history coverage and execution efficiency. We introduce MaP-WAM, a Memory-as-Plans framework that decomposes memory-dependent world-action modeling into memory-grounded planning and plan-conditioned execution, and uses long-term multimodal episodic context as planning-time evidence rather than repeatedly conditioning the executor on the full history. MaP-WAM represents memory as completed segment records containing language instructions and sparse visual context, and converts this episodic memory into compact plans comprising the next segment-level language plan and corresponding visual guidance. A World-Action-Progress (WAP) model executes each plan over an unknown duration by jointly predicting action chunks and corresponding execution progress at inference time, calibrating predicted progress through plan-observation alignment for adaptive segment transitions and closed-loop context updates. MaP-WAM keeps the executor context length fixed, while structured attention further enables key-value caching in both planning and execution. MaP-WAM achieves state-of-the-art performance on RMBench with an 83.3% success rate and attains 78.0% success on real-robot tasks, while maintaining approximately constant executor inference latency as task history grows.","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/63f47b5321eb234ab739e91a/clT7ifvITIQhw6ni64AqX.mp4"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11561.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"63f47b5321eb234ab739e91a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63f47b5321eb234ab739e91a/vWfFNVtMkHl8gieha5PPd.jpeg","fullname":"Haozhe Xie","name":"hzxie","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":27,"isUserFollowing":false},"isAuthorParticipating":true},{"paper":{"id":"2609.11499","authors":[{"_id":"6aa36ba647a406da7901e717","name":"Zhiqi Li","hidden":false},{"_id":"6aa36ba647a406da7901e718","name":"Yuxuan Liao","hidden":false},{"_id":"6aa36ba647a406da7901e719","name":"Bo Zhu","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recursive Scene Program (RSP) representation with a construction solver that recursively calls itself. An RSP represents the executable world as compositional scene code, while each solver call follows the same complete process: establish the whole, recursively reconstruct unresolved parts, and revisit the whole to refine their composition. This global-local-global recursion gives fine-scale structures their own perception-and-editing loops while preserving scene-wide geometry and relationships. Reference-aligned views propagate a shared camera projection across levels, while parent revisitation addresses boundaries, spatial relations, and shared errors that emerge after local refinement. A vision-language coding agent directly compares reference images with scene renders to guide refinement, recursive descent, and return. Across complex scenes, RCWM outperforms prior code-based image-to-scene reconstruction methods. Ablation studies further support the benefits of recursive construction and suggest that deeper calls can improve finer-scale reconstruction. RCWM provides a recursive construction principle for building complex executable worlds from visual evidence.","upvotes":6,"discussionId":"6aa36ba747a406da7901e71a","ai_summary":"RCWM recursively builds complex 3D worlds as executable code from a single image by alternating global and local reconstruction with shared camera alignment.","ai_keywords":["Recursive Code World Models","Recursive Scene Program","vision-language coding agent","recursive construction","code-based image-to-scene reconstruction"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs","summary":"Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recursive Scene Program (RSP) representation with a construction solver that recursively calls itself. An RSP represents the executable world as compositional scene code, while each solver call follows the same complete process: establish the whole, recursively reconstruct unresolved parts, and revisit the whole to refine their composition. This global-local-global recursion gives fine-scale structures their own perception-and-editing loops while preserving scene-wide geometry and relationships. Reference-aligned views propagate a shared camera projection across levels, while parent revisitation addresses boundaries, spatial relations, and shared errors that emerge after local refinement. A vision-language coding agent directly compares reference images with scene renders to guide refinement, recursive descent, and return. Across complex scenes, RCWM outperforms prior code-based image-to-scene reconstruction methods. Ablation studies further support the benefits of recursive construction and suggest that deeper calls can improve finer-scale reconstruction. RCWM provides a recursive construction principle for building complex executable worlds from visual evidence.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11499.png","numComments":0,"upvoted":false,"submittedBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","fullname":"taesiri","name":"taesiri","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":373,"isUserFollowing":false},"isAuthorParticipating":false},{"paper":{"id":"2609.01515","authors":[{"_id":"6aa2f09f47a406da7901e5f4","user":{"_id":"65e7d63b14856e8859f1924c","avatarUrl":"/avatars/bc58ab252c7b4d95ff99e4506fb8d3e9.svg","isPro":false,"fullname":"Pei Wenqi","user":"CedPei","type":"user","name":"CedPei"},"name":"Wenqi Pei","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.689Z","hidden":false},{"_id":"6aa2f09f47a406da7901e5f5","name":"Henry Hengyuan Zhao","hidden":false},{"_id":"6aa2f09f47a406da7901e5f6","name":"Yilai Liu","hidden":false},{"_id":"6aa2f09f47a406da7901e5f7","user":{"_id":"65a28e129acab19980226731","avatarUrl":"/avatars/abc3828f807efc4e03837b0eae063f98.svg","isPro":false,"fullname":"Jiahao Meng","user":"marinero4972","type":"user","name":"marinero4972"},"name":"Jiahao Meng","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:35.230Z","hidden":false},{"_id":"6aa2f09f47a406da7901e5f8","user":{"_id":"6399c67bf78f75ae73146760","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6399c67bf78f75ae73146760/LAZxoSRD-hte-S9736iyg.jpeg","isPro":false,"fullname":"CHEN Han","user":"Concyclics","type":"user","name":"Concyclics"},"name":"Han Chen","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:37.834Z","hidden":false},{"_id":"6aa2f09f47a406da7901e5f9","name":"Ziyu Wang","hidden":false},{"_id":"6aa2f09f47a406da7901e5fa","name":"Hongyang Du","hidden":false}],"publishedAt":"2026-09-01T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"TempCloze: Can Video-LLMs Identify the Missing Middle?","submittedOnDailyBy":{"_id":"65e7d63b14856e8859f1924c","avatarUrl":"/avatars/bc58ab252c7b4d95ff99e4506fb8d3e9.svg","isPro":false,"fullname":"Pei Wenqi","user":"CedPei","type":"user","name":"CedPei"},"summary":"Temporal reasoning benchmarks for Video-LLMs are often mediated by language, leaving room for linguistic shortcuts from option wording, answer correlations, or language priors. To reduce such shortcuts, we introduce TempCloze, a video cloze benchmark for evaluating visual temporal reasoning in Video-LLMs. Given the beginning and ending clips of a video, models must identify the true missing middle from four candidates. TempCloze contains 1,521 carefully filtered videos from seven sources, mainly long-take and egocentric videos. We construct same-source distractors along three dimensions: Semantic asks what event should happen, Alignment probes when it should occur, and Progression tests how it should unfold, while shared scenes and objects reduce appearance cues. Our evaluation of 10 proprietary and 21 open-source Video-LLMs reveals Alignment as the primary bottleneck: models often recognize plausible semantic content and local event progression but struggle with temporal alignment. We further conduct error pattern and behavioral sensitivity analyses on TempCloze-Mixed and TempCloze-Hard with four representative models to examine where errors arise and how candidate order, context direction, visible span, frame density, and test-time scaling influence model choices.","upvotes":6,"discussionId":"6aa2f09f47a406da7901e5fb","githubRepo":"https://github.com/CedricPei/Temporal-Cloze","githubRepoAddedBy":"user","ai_summary":"TempCloze evaluates visual temporal reasoning in Video-LLMs by requiring identification of missing video segments from distractors targeting semantics, alignment, and progression.","ai_keywords":["Video-LLMs","temporal reasoning","video cloze","same-source distractors","semantic alignment","progression","test-time scaling"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":5,"organization":{"_id":"67ea9ecfc234715db8dbf339","name":"hkuhk","fullname":"The University of Hong Kong","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ea9e8d2d95c10a0da11b0c/FNnR4M7YqKRuG43N5771B.png"}},"publishedAt":"2026-08-31T20:00:00.000Z","title":"TempCloze: Can Video-LLMs Identify the Missing Middle?","summary":"Temporal reasoning benchmarks for Video-LLMs are often mediated by language, leaving room for linguistic shortcuts from option wording, answer correlations, or language priors. To reduce such shortcuts, we introduce TempCloze, a video cloze benchmark for evaluating visual temporal reasoning in Video-LLMs. Given the beginning and ending clips of a video, models must identify the true missing middle from four candidates. TempCloze contains 1,521 carefully filtered videos from seven sources, mainly long-take and egocentric videos. We construct same-source distractors along three dimensions: Semantic asks what event should happen, Alignment probes when it should occur, and Progression tests how it should unfold, while shared scenes and objects reduce appearance cues. Our evaluation of 10 proprietary and 21 open-source Video-LLMs reveals Alignment as the primary bottleneck: models often recognize plausible semantic content and local event progression but struggle with temporal alignment. We further conduct error pattern and behavioral sensitivity analyses on TempCloze-Mixed and TempCloze-Hard with four representative models to examine where errors arise and how candidate order, context direction, visible span, frame density, and test-time scaling influence model choices.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.01515.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"65e7d63b14856e8859f1924c","avatarUrl":"/avatars/bc58ab252c7b4d95ff99e4506fb8d3e9.svg","fullname":"Pei Wenqi","name":"CedPei","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":2,"isUserFollowing":false},"organization":{"_id":"67ea9ecfc234715db8dbf339","name":"hkuhk","fullname":"The University of Hong Kong","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/67ea9e8d2d95c10a0da11b0c/FNnR4M7YqKRuG43N5771B.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.11486","authors":[{"_id":"6aa3706347a406da7901e753","name":"Vladislav Bargatin","hidden":false},{"_id":"6aa3706347a406da7901e754","user":{"_id":"663692c75f67f8da32723bf8","avatarUrl":"/avatars/258264afe2ea5048a4a7a8e9945d2f5b.svg","isPro":false,"fullname":"Alexander Yakovenko","user":"a-yakovenko","type":"user","name":"a-yakovenko"},"name":"Alexander Yakovenko","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:20.868Z","hidden":false},{"_id":"6aa3706347a406da7901e755","name":"Khaled Abud","hidden":false},{"_id":"6aa3706347a406da7901e756","name":"Dmitriy Vatolin","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation","submittedOnDailyBy":{"_id":"663692c75f67f8da32723bf8","avatarUrl":"/avatars/258264afe2ea5048a4a7a8e9945d2f5b.svg","isPro":false,"fullname":"Alexander Yakovenko","user":"a-yakovenko","type":"user","name":"a-yakovenko"},"summary":"Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to more complex pipelines and additional computational cost. We present FreeFlow, a hierarchical transformer built without any flow-specific components, using instead a single feed-forward encoder--decoder. FreeFlow combines three attention variants: window attention for local processing, shifted-window attention for cross-window information exchange, and a global attention operating at a reduced resolution. The resulting architecture scales naturally with model capacity, enabling a consistent accuracy gain from small to large variants. Despite the absence of standard inductive biases, FreeFlow achieves state-of-the-art results on major benchmarks, including Sintel (0.68/1.48 EPE on Clean/Final), KITTI-2015 (3.23 Fl-all), and Spring (3.192 1px), while remaining memory efficient at 1080p inference.","upvotes":5,"discussionId":"6aa3706347a406da7901e757","projectPage":"https://github.com/msu-video-group/freeflow","githubRepo":"https://github.com/msu-video-group/freeflow","githubRepoAddedBy":"user","ai_summary":"FreeFlow is a hierarchical transformer for optical flow that eliminates task-specific inductive biases and achieves state-of-the-art accuracy using window, shifted-window, and global attention.","ai_keywords":["optical flow","correlation volumes","feature warping","iterative refinement","hierarchical transformer","feed-forward encoder-decoder","window attention","shifted-window attention","global attention","Sintel","KITTI-2015","Spring"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":2},"publishedAt":"2026-09-09T20:00:00.000Z","title":"FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation","summary":"Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to more complex pipelines and additional computational cost. We present FreeFlow, a hierarchical transformer built without any flow-specific components, using instead a single feed-forward encoder--decoder. FreeFlow combines three attention variants: window attention for local processing, shifted-window attention for cross-window information exchange, and a global attention operating at a reduced resolution. The resulting architecture scales naturally with model capacity, enabling a consistent accuracy gain from small to large variants. Despite the absence of standard inductive biases, FreeFlow achieves state-of-the-art results on major benchmarks, including Sintel (0.68/1.48 EPE on Clean/Final), KITTI-2015 (3.23 Fl-all), and Spring (3.192 1px), while remaining memory efficient at 1080p inference.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11486.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"663692c75f67f8da32723bf8","avatarUrl":"/avatars/258264afe2ea5048a4a7a8e9945d2f5b.svg","fullname":"Alexander Yakovenko","name":"a-yakovenko","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"isAuthorParticipating":true},{"paper":{"id":"2609.11548","authors":[{"_id":"6aa36a7547a406da7901e711","name":"Chenxi Song","hidden":false},{"_id":"6aa36a7547a406da7901e712","name":"Yanming Yang","hidden":false},{"_id":"6aa36a7547a406da7901e713","name":"Chi Zhang","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6039478ab3ecf716b1a5fd4d/V2ktCpO1_KDxxmbRipRYA.mp4"],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"World in World: Explore the World with World Models","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from new viewpoints requires the generated rollout to remain synchronised with the recorded event, place observed content in the requested view, plausibly complete newly exposed regions, and recover previously generated appearance on revisits. Existing methods typically address these requirements through task-specific modules or additional training. We present World in World, a training-free inference-time interface that converts heterogeneous control evidence into camera- and time-labelled clean visual states, which are read through the native self attention of a frozen causal video model. The evidence comprises source-video observations, target-view scene projections, geometry renderings that guide completion of newly exposed subject regions, and retrieved generated states beyond the rolling cache. Each evidence source carries token-level support and its own availability schedule. A correspondence router combines persistent point identities with geometry to establish token correspondences, guiding supported queries towards matching source-video tokens. Evidence-wise attention CFG (EWA) then independently regulates each auxiliary channel's additional contribution using attention responses from the same denoising forward pass. The shared interface supports camera-controlled rerendering, long-horizon revisiting, and human-motion transfer with the same frozen backbone. We evaluate World in World on camera-controlled video rerendering under diverse viewpoint changes, assessing perceptual quality, temporal consistency, and camera-following accuracy.","upvotes":5,"discussionId":"6aa36a7647a406da7901e714","projectPage":"https://chenxi-song.github.io/worldinworld/","ai_summary":"A training-free interface enables flexible camera and time control in frozen autoregressive video world models by routing heterogeneous visual evidence through native self-attention with correspondence-guided queries and per-channel attention guidance.","ai_keywords":["autoregressive video world models","self-attention","camera-controlled rerendering","correspondence router","evidence-wise attention CFG","token-level support","geometry renderings","long-horizon revisiting"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"publishedAt":"2026-09-09T20:00:00.000Z","title":"World in World: Explore the World with World Models","summary":"Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from new viewpoints requires the generated rollout to remain synchronised with the recorded event, place observed content in the requested view, plausibly complete newly exposed regions, and recover previously generated appearance on revisits. Existing methods typically address these requirements through task-specific modules or additional training. We present World in World, a training-free inference-time interface that converts heterogeneous control evidence into camera- and time-labelled clean visual states, which are read through the native self attention of a frozen causal video model. The evidence comprises source-video observations, target-view scene projections, geometry renderings that guide completion of newly exposed subject regions, and retrieved generated states beyond the rolling cache. Each evidence source carries token-level support and its own availability schedule. A correspondence router combines persistent point identities with geometry to establish token correspondences, guiding supported queries towards matching source-video tokens. Evidence-wise attention CFG (EWA) then independently regulates each auxiliary channel's additional contribution using attention responses from the same denoising forward pass. The shared interface supports camera-controlled rerendering, long-horizon revisiting, and human-motion transfer with the same frozen backbone. We evaluate World in World on camera-controlled video rerendering under diverse viewpoint changes, assessing perceptual quality, temporal consistency, and camera-following accuracy.","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6039478ab3ecf716b1a5fd4d/V2ktCpO1_KDxxmbRipRYA.mp4"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11548.png","numComments":0,"upvoted":false,"submittedBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","fullname":"taesiri","name":"taesiri","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":373,"isUserFollowing":false},"isAuthorParticipating":false},{"paper":{"id":"2609.10712","authors":[{"_id":"6aa36e5047a406da7901e74a","name":"Ivan Moshkov","hidden":false},{"_id":"6aa36e5047a406da7901e74b","name":"Stephen Ge","hidden":false},{"_id":"6aa36e5047a406da7901e74c","name":"George Armstrong","hidden":false},{"_id":"6aa36e5047a406da7901e74d","name":"Wei Du","hidden":false},{"_id":"6aa36e5047a406da7901e74e","name":"Sadegh Mahdavi","hidden":false},{"_id":"6aa36e5047a406da7901e74f","name":"Igor Gitman","hidden":false}],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics","submittedOnDailyBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","isPro":true,"fullname":"taesiri","user":"taesiri","type":"user","name":"taesiri"},"summary":"We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.","upvotes":5,"discussionId":"6aa36e5047a406da7901e750","ai_summary":"A natural-language proof-generation pipeline using post-trained Nemotron 3 Ultra checkpoints achieves gold-medal performance on IMO 2026 through iterative verification and refinement without external tools.","ai_keywords":["supervised fine-tuning","reinforcement learning","verification","refinement","test-time compute","iterative search","Nemotron-IMO-Bench"],"ai_summary_model":"thinkingmachines/Inkling-Small","organization":{"_id":"60262b67268c201cdc8b7d43","name":"nvidia","fullname":"NVIDIA","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65df9200dc3292a8983e5017/Vs5FPVCH-VZBipV3qKTuy.png"}},"publishedAt":"2026-09-08T20:00:00.000Z","title":"An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics","summary":"We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10712.png","numComments":0,"upvoted":false,"submittedBy":{"_id":"6039478ab3ecf716b1a5fd4d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg","fullname":"taesiri","name":"taesiri","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":373,"isUserFollowing":false},"organization":{"_id":"60262b67268c201cdc8b7d43","name":"nvidia","fullname":"NVIDIA","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65df9200dc3292a8983e5017/Vs5FPVCH-VZBipV3qKTuy.png"},"isAuthorParticipating":false},{"paper":{"id":"2609.11699","authors":[{"_id":"6aa3a26447a406da7901e87d","user":{"_id":"67316c6cb9634ac96f65e1a0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uUY9F17MiN7NhGAg01Yom.png","isPro":false,"fullname":"PP","user":"PassionPrc","type":"user","name":"PassionPrc"},"name":"Rongcan Pei","status":"claimed_verified","statusLastChangedAt":"2026-09-11T09:53:16.674Z","hidden":false},{"_id":"6aa3a26447a406da7901e87e","name":"Zhepei Wei","hidden":false},{"_id":"6aa3a26447a406da7901e87f","name":"Shuyao Xu","hidden":false},{"_id":"6aa3a26447a406da7901e880","name":"Xinyu Zhu","hidden":false},{"_id":"6aa3a26447a406da7901e881","name":"Wei-Lin Chen","hidden":false},{"_id":"6aa3a26447a406da7901e882","name":"Yu Meng","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Negative Self-Distillation: Learning to Reason by Avoiding Flaws","submittedOnDailyBy":{"_id":"67316c6cb9634ac96f65e1a0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uUY9F17MiN7NhGAg01Yom.png","isPro":false,"fullname":"PP","user":"PassionPrc","type":"user","name":"PassionPrc"},"summary":"On-Policy Self-Distillation (OPSD) has emerged as a popular paradigm for large language model (LLM) self-improvement, allowing models to act as their own teachers by leveraging privileged information such as ground-truth solutions. However, recent findings indicate that OPSD can severely degrade the performance of LLMs on complex reasoning tasks: By forcing the student to imitate an artificially confident reasoning trace conditioned on privileged information, OPSD inadvertently suppresses expressions of uncertainty and penalizes the exploratory, self-corrective behaviors required to solve challenging problems. To address this, we introduce Negative Self-Distillation (NSD), a new framework that optimizes LLMs by diverging from flawed reasoning rather than imitating privileged solutions. Instead of relying on ground-truth answers or external supervision, NSD uses the model itself to generate a question-specific negative condition (eg, acting as a ``careless reasoner'') and pushes the student's distribution away from this self-generated negative teacher. Naively applying unlearning objectives to achieve this divergence is problematic, as flawed reasoning tokens are confounded with basic linguistic tokens; indiscriminately penalizing both risks catastrophically degrading the model's foundational language capabilities. We resolve this by designing a dynamic gating mechanism that automatically identifies and isolates reasoning-critical tokens, ensuring gradient updates target only behavioral flaws while preserving the model's linguistic priors. Empirically, NSD consistently outperforms OPSD and other label-free, self-bootstrapping reinforcement learning (RL) baselines.","upvotes":4,"discussionId":"6aa3a26447a406da7901e883","githubRepo":"https://github.com/Prongcan/NSD","githubRepoAddedBy":"user","ai_summary":"Negative Self-Distillation improves large language model reasoning by pushing models away from self-generated flawed reasoning via a dynamic gating mechanism that protects linguistic capabilities.","ai_keywords":["On-Policy Self-Distillation","Negative Self-Distillation","self-distillation","reasoning-critical tokens","dynamic gating mechanism","unlearning","self-bootstrapping reinforcement learning"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":4},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Negative Self-Distillation: Learning to Reason by Avoiding Flaws","summary":"On-Policy Self-Distillation (OPSD) has emerged as a popular paradigm for large language model (LLM) self-improvement, allowing models to act as their own teachers by leveraging privileged information such as ground-truth solutions. However, recent findings indicate that OPSD can severely degrade the performance of LLMs on complex reasoning tasks: By forcing the student to imitate an artificially confident reasoning trace conditioned on privileged information, OPSD inadvertently suppresses expressions of uncertainty and penalizes the exploratory, self-corrective behaviors required to solve challenging problems. To address this, we introduce Negative Self-Distillation (NSD), a new framework that optimizes LLMs by diverging from flawed reasoning rather than imitating privileged solutions. Instead of relying on ground-truth answers or external supervision, NSD uses the model itself to generate a question-specific negative condition (eg, acting as a ``careless reasoner'') and pushes the student's distribution away from this self-generated negative teacher. Naively applying unlearning objectives to achieve this divergence is problematic, as flawed reasoning tokens are confounded with basic linguistic tokens; indiscriminately penalizing both risks catastrophically degrading the model's foundational language capabilities. We resolve this by designing a dynamic gating mechanism that automatically identifies and isolates reasoning-critical tokens, ensuring gradient updates target only behavioral flaws while preserving the model's linguistic priors. Empirically, NSD consistently outperforms OPSD and other label-free, self-bootstrapping reinforcement learning (RL) baselines.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11699.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"67316c6cb9634ac96f65e1a0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uUY9F17MiN7NhGAg01Yom.png","fullname":"PP","name":"PassionPrc","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"isAuthorParticipating":true},{"paper":{"id":"2609.06931","authors":[{"_id":"6aa0b959d0174964227bebd5","user":{"_id":"697035d2d974214e2ccd5c7b","avatarUrl":"/avatars/ade0fe033c0821ea8a8ed487a0c10db4.svg","isPro":false,"fullname":"Jia-Jen Lee","user":"benbayibaurba","type":"user","name":"benbayibaurba"},"name":"Jia-Jen Lee","status":"admin_assigned","statusLastChangedAt":"2026-09-10T16:44:01.688Z","hidden":false},{"_id":"6aa0b959d0174964227bebd6","name":"Shih-Yen Hou","hidden":false},{"_id":"6aa0b959d0174964227bebd7","name":"Kee Koon Ng","hidden":false},{"_id":"6aa0b959d0174964227bebd8","name":"Wei-Chun Wang","hidden":false},{"_id":"6aa0b959d0174964227bebd9","name":"Shih-Sheng Chang","hidden":false}],"publishedAt":"2026-09-07T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"CARDEA: Auditable Reasoning Grounded in Spatial Evidence for End-to-End Coronary Angiography Interpretation","submittedOnDailyBy":{"_id":"697035d2d974214e2ccd5c7b","avatarUrl":"/avatars/ade0fe033c0821ea8a8ed487a0c10db4.svg","isPro":false,"fullname":"Jia-Jen Lee","user":"benbayibaurba","type":"user","name":"benbayibaurba"},"summary":"Invasive coronary angiography (CAG) is the gold standard for diagnosing coronary artery disease, but interpretation varies substantially among observers. Existing AI systems can improve consistency but lack auditable decision processes and are limited in comprehensive open-ended assessment, undermining clinician trust and clinical adoption readiness. We developed CARDEA, a unified large vision-language model that serves as the inference core of a CAG pipeline. It was trained solely on public datasets and closed-ended tasks in three stages: visual feature alignment, a self-distilled Chain-of-Box (CoB) cold start, and reinforcement learning with verifiable rewards (RLVR) with a CoB reward encouraging bounding-box use in the reasoning trace. We assessed its two study-level diagnoses, dominance classification and complexity assessment, against a dedicated classifier and two interventional cardiologists. Report generation was excluded from training and evaluated zero-shot across stages on an external cohort using vessel-severity macro-F_1. CARDEA trailed the classifier on in-distribution dominance but drew level under domain shift (accuracy, 0.91 [95% confidence interval (CI), 0.86 to 0.95]) and was comparable to the cardiologists on complexity assessment (accuracy, 0.90 [CI, 0.82 to 0.97]). Only RLVR improved zero-shot report generation, raising its vessel-severity macro-F_1 (0.686 [CI, 0.664 to 0.707]) above the untuned base model (0.513) and over twice the always-normal floor (0.312). CARDEA runs an end-to-end CAG pipeline from raw multi-view videos through keyframe selection to study-level diagnosis while exposing auditable spatial evidence behind its conclusions. RLVR on verifiable closed-ended tasks surfaced open-ended reporting ability that supervised imitation did not. Clinical use requires prospective validation against expert cardiologists.","upvotes":4,"discussionId":"6aa0b959d0174964227bebda","githubRepo":"https://github.com/benbayibaurba/cardea","githubRepoAddedBy":"user","ai_summary":"A unified vision-language model for coronary angiography uses chain-of-box reasoning and reinforcement learning with verifiable rewards to provide auditable diagnoses and improve zero-shot report generation.","ai_keywords":["large vision-language model","visual feature alignment","self-distilled Chain-of-Box","reinforcement learning with verifiable rewards","RLVR","bounding-box","zero-shot","macro-F1"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0},"publishedAt":"2026-09-06T20:00:00.000Z","title":"CARDEA: Auditable Reasoning Grounded in Spatial Evidence for End-to-End Coronary Angiography Interpretation","summary":"Invasive coronary angiography (CAG) is the gold standard for diagnosing coronary artery disease, but interpretation varies substantially among observers. Existing AI systems can improve consistency but lack auditable decision processes and are limited in comprehensive open-ended assessment, undermining clinician trust and clinical adoption readiness. We developed CARDEA, a unified large vision-language model that serves as the inference core of a CAG pipeline. It was trained solely on public datasets and closed-ended tasks in three stages: visual feature alignment, a self-distilled Chain-of-Box (CoB) cold start, and reinforcement learning with verifiable rewards (RLVR) with a CoB reward encouraging bounding-box use in the reasoning trace. We assessed its two study-level diagnoses, dominance classification and complexity assessment, against a dedicated classifier and two interventional cardiologists. Report generation was excluded from training and evaluated zero-shot across stages on an external cohort using vessel-severity macro-F_1. CARDEA trailed the classifier on in-distribution dominance but drew level under domain shift (accuracy, 0.91 [95% confidence interval (CI), 0.86 to 0.95]) and was comparable to the cardiologists on complexity assessment (accuracy, 0.90 [CI, 0.82 to 0.97]). Only RLVR improved zero-shot report generation, raising its vessel-severity macro-F_1 (0.686 [CI, 0.664 to 0.707]) above the untuned base model (0.513) and over twice the always-normal floor (0.312). CARDEA runs an end-to-end CAG pipeline from raw multi-view videos through keyframe selection to study-level diagnosis while exposing auditable spatial evidence behind its conclusions. RLVR on verifiable closed-ended tasks surfaced open-ended reporting ability that supervised imitation did not. Clinical use requires prospective validation against expert cardiologists.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.06931.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"697035d2d974214e2ccd5c7b","avatarUrl":"/avatars/ade0fe033c0821ea8a8ed487a0c10db4.svg","fullname":"Jia-Jen Lee","name":"benbayibaurba","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"isAuthorParticipating":true},{"paper":{"id":"2608.27875","authors":[{"_id":"6a9e97efaf127efa95389b39","user":{"_id":"6847c20bb510d25b2322fa5c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6847c20bb510d25b2322fa5c/LWov6B3_1BVbkYWewUyL-.png","isPro":false,"fullname":"Jiatong Ding","user":"jerrysfls","type":"user","name":"jerrysfls"},"name":"Jiatong Ding","status":"claimed_verified","statusLastChangedAt":"2026-09-11T00:45:04.234Z","hidden":false},{"_id":"6a9e97efaf127efa95389b3a","name":"Bingxin Xing","hidden":false},{"_id":"6a9e97efaf127efa95389b3b","name":"Yu Zhang","hidden":false},{"_id":"6a9e97efaf127efa95389b3c","name":"Dian Ding","hidden":false},{"_id":"6a9e97efaf127efa95389b3d","name":"Xiaodong Yi","hidden":false},{"_id":"6a9e97efaf127efa95389b3e","name":"Xianbin Ouyang","hidden":false},{"_id":"6a9e97efaf127efa95389b3f","name":"Feihu Zhou","hidden":false},{"_id":"6a9e97efaf127efa95389b40","name":"Kun Zhang","hidden":false},{"_id":"6a9e97efaf127efa95389b41","name":"Zhenyu Guo","hidden":false},{"_id":"6a9e97efaf127efa95389b42","name":"Hao Pan","hidden":false},{"_id":"6a9e97efaf127efa95389b43","name":"Guangtao Xue","hidden":false},{"_id":"6a9e97efaf127efa95389b44","name":"Yiming Zhang","hidden":false}],"publishedAt":"2026-08-28T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"HyQuant: Hybrid-Precision Quantization for LLM Attention","submittedOnDailyBy":{"_id":"6847c20bb510d25b2322fa5c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6847c20bb510d25b2322fa5c/LWov6B3_1BVbkYWewUyL-.png","isPro":false,"fullname":"Jiatong Ding","user":"jerrysfls","type":"user","name":"jerrysfls"},"summary":"Quantization has been widely adopted in LLM training and inference to reduce cost and improve efficiency. However, low-bit quantization of the attention module often introduces large errors at very low bit-widths, causing performance degradation. Existing methods mainly rely on smoothing techniques to handle outliers, while we propose a hybrid quantization design to better balance accuracy and efficiency. Specifically, we propose HyQuant, an efficient hybrid quantization framework for LLM attention. HyQuant quantizes most attention states into low-bit formats while retaining a small set of vertical-line tokens and local-window states in high precision. These accuracy-critical regions are selected using lightweight vertical-line-aware attention-pattern signals, reducing quantization error with limited overhead. In the Prefill stage, HyQuant uses a hybrid-precision quantized attention operator that preserves vertical-line tokens and a local sliding window in full precision while quantizing the remaining context. In the Decode stage, HyQuant applies the same principle to KV-cache compression and fuses KV dequantization with attention computation to improve memory and hardware efficiency. Across diverse tasks, models, and datasets, HyQuant maintains nearly lossless accuracy with an extremely simple design, demonstrating the efficiency and practical feasibility of hybrid quantization for LLM attention. Code is available at: https://github.com/jerrysfls/HyQuant .","upvotes":4,"discussionId":"6a9e97f0af127efa95389b4c","githubRepo":"https://github.com/jerrysfls/HyQuant","githubRepoAddedBy":"user","ai_summary":"HyQuant improves low-bit LLM attention quantization by preserving critical vertical-line tokens and local windows in high precision while quantizing the rest, maintaining accuracy with low overhead.","ai_keywords":["hybrid quantization","attention quantization","low-bit quantization","vertical-line tokens","local-window states","KV-cache compression","prefill stage","decode stage"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":8,"organization":{"_id":"63e5ef7bf2e9a8f22c515654","name":"SJTU","fullname":"Shanghai Jiao Tong University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1676013394657-63e5ee22b6a40bf941da0928.png"}},"publishedAt":"2026-08-27T20:00:00.000Z","title":"HyQuant: Hybrid-Precision Quantization for LLM Attention","summary":"Quantization has been widely adopted in LLM training and inference to reduce cost and improve efficiency. However, low-bit quantization of the attention module often introduces large errors at very low bit-widths, causing performance degradation. Existing methods mainly rely on smoothing techniques to handle outliers, while we propose a hybrid quantization design to better balance accuracy and efficiency. Specifically, we propose HyQuant, an efficient hybrid quantization framework for LLM attention. HyQuant quantizes most attention states into low-bit formats while retaining a small set of vertical-line tokens and local-window states in high precision. These accuracy-critical regions are selected using lightweight vertical-line-aware attention-pattern signals, reducing quantization error with limited overhead. In the Prefill stage, HyQuant uses a hybrid-precision quantized attention operator that preserves vertical-line tokens and a local sliding window in full precision while quantizing the remaining context. In the Decode stage, HyQuant applies the same principle to KV-cache compression and fuses KV dequantization with attention computation to improve memory and hardware efficiency. Across diverse tasks, models, and datasets, HyQuant maintains nearly lossless accuracy with an extremely simple design, demonstrating the efficiency and practical feasibility of hybrid quantization for LLM attention. Code is available at: https://github.com/jerrysfls/HyQuant .","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2608.27875.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6847c20bb510d25b2322fa5c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6847c20bb510d25b2322fa5c/LWov6B3_1BVbkYWewUyL-.png","fullname":"Jiatong Ding","name":"jerrysfls","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"63e5ef7bf2e9a8f22c515654","name":"SJTU","fullname":"Shanghai Jiao Tong University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1676013394657-63e5ee22b6a40bf941da0928.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.10016","authors":[{"_id":"6aa29c0ca2aeb74440b1e0c2","user":{"_id":"63457dc5547c70e4b7ccc128","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/M-gCXZHVZFLLlvvYKHcG6.png","isPro":true,"fullname":"Remco Hendriks","user":"remcohendriks","type":"user","name":"remcohendriks"},"name":"Remco Hendriks","status":"claimed_verified","statusLastChangedAt":"2026-09-10T16:45:04.714Z","hidden":false}],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes","submittedOnDailyBy":{"_id":"63457dc5547c70e4b7ccc128","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/M-gCXZHVZFLLlvvYKHcG6.png","isPro":true,"fullname":"Remco Hendriks","user":"remcohendriks","type":"user","name":"remcohendriks"},"summary":"We introduce MetroLLM-Bench, a 955-case benchmark for testing language models as the policy layer of a transit kiosk. It covers six real metro systems, ranging from 37 to 414 stations, and eleven categories that include routing, fare calculation, disruptions, accessibility, and adversarial input. In each case, the model must call structured tools and submit a machine-renderable terminal state containing an outcome, a per-ticket fare quote when applicable, and a kiosk action. Fourteen deterministic scoring components form Tier 1; eight semantic-quality components form Tier 2, six of which use a language-model judge. We report Tier 1 and the combined score of both tiers. A stratified 75/25 split reserves 717 cases for training-data generation and 238 for held-out evaluation.\n We evaluate twenty-six models from six vendors, of which twenty-three are ranked. On the held-out partition, a 4B Qwen 3.5 student trained through parameter-efficient fine-tuning (PEFT) exceeds both GPT-5.6 tiers on Tier 1 (91.3 against 90.6 and 90.0) and matches GPT-5.4 full at maximum reasoning effort (91.4), with a 2.6 GB Q4_K_M footprint. Larger 9B and 27B students provide no further Tier 1 improvement over the 4B student at this training scale. Across the four Qwen sizes, the PEFT gain over the corresponding base model decreases from +7.03 points at 2B (three training seeds) to -0.91 at 27B; every seed shows the same direction at every size. A deterministic rule-based baseline reaches 84.6 on Tier 1, with the remaining language-model advantage concentrated in policy adaptation, compound scenarios, accessibility, and temporal reasoning. Muse Glimmer 30B leads the composite ranking, and serving configuration alone moves the Qwen 3.5-to-3.8 comparison by 2.7 Tier 1 points. The benchmark, harness, reproduction guide, and fine-tuned students are released at https://github.com/continker/metrollm-bench.","upvotes":3,"discussionId":"6aa29c0ca2aeb74440b1e0c3","projectPage":"https://continker.ai/metrollm-bench","githubRepo":"https://github.com/continker/metrollm-bench","githubRepoAddedBy":"user","ai_summary":"A benchmark for transit-kiosk policy reasoning shows that small parameter-efficiently fine-tuned language models can match larger models on structured tool-use and fare-quoting tasks.","ai_keywords":["parameter-efficient fine-tuning","PEFT","Qwen","language-model judge","deterministic scoring","transit kiosk policy"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0,"organization":{"_id":"69d1315760b47ba30c6f0c49","name":"continker","fullname":"Continker","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/nptM94xE8POJPxfKryjn0.png"}},"publishedAt":"2026-09-08T20:00:00.000Z","title":"MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes","summary":"We introduce MetroLLM-Bench, a 955-case benchmark for testing language models as the policy layer of a transit kiosk. It covers six real metro systems, ranging from 37 to 414 stations, and eleven categories that include routing, fare calculation, disruptions, accessibility, and adversarial input. In each case, the model must call structured tools and submit a machine-renderable terminal state containing an outcome, a per-ticket fare quote when applicable, and a kiosk action. Fourteen deterministic scoring components form Tier 1; eight semantic-quality components form Tier 2, six of which use a language-model judge. We report Tier 1 and the combined score of both tiers. A stratified 75/25 split reserves 717 cases for training-data generation and 238 for held-out evaluation.\n We evaluate twenty-six models from six vendors, of which twenty-three are ranked. On the held-out partition, a 4B Qwen 3.5 student trained through parameter-efficient fine-tuning (PEFT) exceeds both GPT-5.6 tiers on Tier 1 (91.3 against 90.6 and 90.0) and matches GPT-5.4 full at maximum reasoning effort (91.4), with a 2.6 GB Q4_K_M footprint. Larger 9B and 27B students provide no further Tier 1 improvement over the 4B student at this training scale. Across the four Qwen sizes, the PEFT gain over the corresponding base model decreases from +7.03 points at 2B (three training seeds) to -0.91 at 27B; every seed shows the same direction at every size. A deterministic rule-based baseline reaches 84.6 on Tier 1, with the remaining language-model advantage concentrated in policy adaptation, compound scenarios, accessibility, and temporal reasoning. Muse Glimmer 30B leads the composite ranking, and serving configuration alone moves the Qwen 3.5-to-3.8 comparison by 2.7 Tier 1 points. The benchmark, harness, reproduction guide, and fine-tuned students are released at https://github.com/continker/metrollm-bench.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10016.png","numComments":2,"upvoted":false,"submittedBy":{"_id":"63457dc5547c70e4b7ccc128","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/M-gCXZHVZFLLlvvYKHcG6.png","fullname":"Remco Hendriks","name":"remcohendriks","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"69d1315760b47ba30c6f0c49","name":"continker","fullname":"Continker","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/nptM94xE8POJPxfKryjn0.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.11155","authors":[{"_id":"6aa3b95947a406da7901e903","name":"Junlin Liu","hidden":false},{"_id":"6aa3b95947a406da7901e904","name":"Chengwei Li","hidden":false},{"_id":"6aa3b95947a406da7901e905","name":"Yang Gao","hidden":false},{"_id":"6aa3b95947a406da7901e906","name":"Hui Chang","hidden":false},{"_id":"6aa3b95947a406da7901e907","name":"Xinchen Zhang","hidden":false},{"_id":"6aa3b95947a406da7901e908","name":"Zhijun Zhao","hidden":false},{"_id":"6aa3b95947a406da7901e909","name":"Hao Zhao","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat","submittedOnDailyBy":{"_id":"69045a2e9c3d523381fcf489","avatarUrl":"/avatars/8b1982e043a7ab28006b707dfff6c5dc.svg","isPro":false,"fullname":"JunlinLiu","user":"AaronLiu0702","type":"user","name":"AaronLiu0702"},"summary":"Multi-Agent Reinforcement Learning (MARL) has emerged as a pivotal paradigm for complex decision-making in autonomous systems and air combat. While MARL has demonstrated significant potential in air combat, achieving sophisticated tactical coordination remains a non-trivial challenge. This difficulty is largely attributed to two primary limitations: (1) the absence of structured relational modeling hinders agents from capturing complex, time-varying interactions among battlefield entities; and (2) conventional flat architectures often lack the capability to explicitly model tactical roles, leading to ambiguous task allocation in highly dynamic environments. To address these challenges, we propose Hierarchical Dynamic Role-Graph Multi-Agent Proximal Policy Optimization (DRG-MAPPO), a novel MARL framework that integrates graph-based relational modeling with dynamic role assignment. Specifically, DRG-MAPPO constructs a graph-based representation of battlefield interactions and leverages graph attention mechanisms to extract critical relational features among allies, enemies, and threats. Subsequently, a high-level policy employs a dynamic role assignment mechanism to determine tactical responsibilities (e.g., ``leader'' and ``supporter''). Conditioned on these roles and encoded graph-relational features, a low-level policy executes discrete maneuver actions, facilitating the joint optimization of tactical strategy and collaborative execution. Furthermore, a target-priority auxiliary task is designed to foster the emergence of behaviors such as focus-fire. Experimental results demonstrate that DRG-MAPPO achieves a state-of-the-art win rate of 87%, suggesting that our framework effectively balances relational modeling, interpretability, and optimization stability for cooperative air combat.","upvotes":2,"discussionId":"6aa3b95a47a406da7901e90a","ai_summary":"A hierarchical multi-agent reinforcement learning framework combining graph attention and dynamic role assignment improves tactical coordination and win rates in air combat.","ai_keywords":["Multi-Agent Reinforcement Learning","graph attention mechanisms","dynamic role assignment","DRG-MAPPO","proximal policy optimization","relational modeling","auxiliary task"],"ai_summary_model":"thinkingmachines/Inkling-Small","organization":{"_id":"64a239f6eada48418580b8e0","name":"China666","fullname":"Chinese Academy of Sciences","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/64a1d2e171874b7465bfca6d/l9MJIPO4MtwFJP5X9zzUw.png"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat","summary":"Multi-Agent Reinforcement Learning (MARL) has emerged as a pivotal paradigm for complex decision-making in autonomous systems and air combat. While MARL has demonstrated significant potential in air combat, achieving sophisticated tactical coordination remains a non-trivial challenge. This difficulty is largely attributed to two primary limitations: (1) the absence of structured relational modeling hinders agents from capturing complex, time-varying interactions among battlefield entities; and (2) conventional flat architectures often lack the capability to explicitly model tactical roles, leading to ambiguous task allocation in highly dynamic environments. To address these challenges, we propose Hierarchical Dynamic Role-Graph Multi-Agent Proximal Policy Optimization (DRG-MAPPO), a novel MARL framework that integrates graph-based relational modeling with dynamic role assignment. Specifically, DRG-MAPPO constructs a graph-based representation of battlefield interactions and leverages graph attention mechanisms to extract critical relational features among allies, enemies, and threats. Subsequently, a high-level policy employs a dynamic role assignment mechanism to determine tactical responsibilities (e.g., ``leader'' and ``supporter''). Conditioned on these roles and encoded graph-relational features, a low-level policy executes discrete maneuver actions, facilitating the joint optimization of tactical strategy and collaborative execution. Furthermore, a target-priority auxiliary task is designed to foster the emergence of behaviors such as focus-fire. Experimental results demonstrate that DRG-MAPPO achieves a state-of-the-art win rate of 87%, suggesting that our framework effectively balances relational modeling, interpretability, and optimization stability for cooperative air combat.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11155.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"69045a2e9c3d523381fcf489","avatarUrl":"/avatars/8b1982e043a7ab28006b707dfff6c5dc.svg","fullname":"JunlinLiu","name":"AaronLiu0702","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"organization":{"_id":"64a239f6eada48418580b8e0","name":"China666","fullname":"Chinese Academy of Sciences","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/64a1d2e171874b7465bfca6d/l9MJIPO4MtwFJP5X9zzUw.png"},"isAuthorParticipating":false},{"paper":{"id":"2609.11156","authors":[{"_id":"6aa39d8047a406da7901e83c","name":"Zhiwen Yang","hidden":false},{"_id":"6aa39d8047a406da7901e83d","name":"Jiayin Li","hidden":false},{"_id":"6aa39d8047a406da7901e83e","name":"Chengyu Liu","hidden":false},{"_id":"6aa39d8047a406da7901e83f","name":"Hui Zhang","hidden":false},{"_id":"6aa39d8047a406da7901e840","name":"Bingzheng Wei","hidden":false},{"_id":"6aa39d8047a406da7901e841","name":"Yan Xu","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration","submittedOnDailyBy":{"_id":"638a2bdb34cf0480e9abd6a4","avatarUrl":"/avatars/70e84c5c188688db8b29455061a410a1.svg","isPro":false,"fullname":"Zhiwen Yang","user":"upyzwup","type":"user","name":"upyzwup"},"summary":"All-in-One medical image restoration (MedIR) aims to address diverse tasks across modalities and degradation types using a single universal model. Existing methods typically prioritize modeling inter-task heterogeneity (e.g., distinct data distributions and degradation types). However, they largely neglect the inherent homogeneity present in medical images, such as widely shared anatomical structures within and across modalities, which can be leveraged to ease model training and improve generalization. To this end, we propose UniH3, a novel framework that Unifies Hierarchical Homogeneity and Heterogeneity for all-in-one medical image restoration. Specifically, to comprehensively exploit homogeneity, we introduce a Hierarchical Homogeneity Memory (H2M) module that progressively distills intra- and inter-task homogeneity priors from high-quality images during training, and adaptively retrieves the most relevant priors tailored to the input for guided restoration. These retrieved priors are then injected into the restoration pipeline via an efficient Homogeneity-Guided Attention (HGA) mechanism. Furthermore, to comprehensively address heterogeneity, we design a Hierarchical Heterogeneity Balancer (H2B) that mitigates both inter- and intra-task conflicts during optimization, facilitating balanced and effective multi-task learning. Extensive experiments on two large-scale benchmarks, MedIR-2D-500K and MedIR-3D-3K, demonstrate that UniH3 achieves state-of-the-art performance on both all-in-one and single-task medical image restoration. We hope this work establishes a strong benchmark and advances the development of general-purpose medical image restoration models. Code is available at https://github.com/Yaziwel/UniH3.","upvotes":2,"discussionId":"6aa39d8047a406da7901e842","projectPage":"https://github.com/Yaziwel/UniH3","githubRepo":"https://github.com/Yaziwel/UniH3","githubRepoAddedBy":"user","ai_summary":"UniH3 unifies hierarchical homogeneity and heterogeneity for universal medical image restoration via memory-based homogeneity priors and a heterogeneity balancer.","ai_keywords":["Hierarchical Homogeneity Memory","H2M","Homogeneity-Guided Attention","HGA","Hierarchical Heterogeneity Balancer","H2B","all-in-one medical image restoration","MedIR"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":5,"organization":{"_id":"63ba7720fc454697637969f1","name":"Beihang","fullname":"Beihang University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ba7666c138c8f2b7844b58/n98lZU9VWxYgWIkzE_6o4.jpeg"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration","summary":"All-in-One medical image restoration (MedIR) aims to address diverse tasks across modalities and degradation types using a single universal model. Existing methods typically prioritize modeling inter-task heterogeneity (e.g., distinct data distributions and degradation types). However, they largely neglect the inherent homogeneity present in medical images, such as widely shared anatomical structures within and across modalities, which can be leveraged to ease model training and improve generalization. To this end, we propose UniH3, a novel framework that Unifies Hierarchical Homogeneity and Heterogeneity for all-in-one medical image restoration. Specifically, to comprehensively exploit homogeneity, we introduce a Hierarchical Homogeneity Memory (H2M) module that progressively distills intra- and inter-task homogeneity priors from high-quality images during training, and adaptively retrieves the most relevant priors tailored to the input for guided restoration. These retrieved priors are then injected into the restoration pipeline via an efficient Homogeneity-Guided Attention (HGA) mechanism. Furthermore, to comprehensively address heterogeneity, we design a Hierarchical Heterogeneity Balancer (H2B) that mitigates both inter- and intra-task conflicts during optimization, facilitating balanced and effective multi-task learning. Extensive experiments on two large-scale benchmarks, MedIR-2D-500K and MedIR-3D-3K, demonstrate that UniH3 achieves state-of-the-art performance on both all-in-one and single-task medical image restoration. We hope this work establishes a strong benchmark and advances the development of general-purpose medical image restoration models. Code is available at https://github.com/Yaziwel/UniH3.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11156.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"638a2bdb34cf0480e9abd6a4","avatarUrl":"/avatars/70e84c5c188688db8b29455061a410a1.svg","fullname":"Zhiwen Yang","name":"upyzwup","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":0,"isUserFollowing":false},"organization":{"_id":"63ba7720fc454697637969f1","name":"Beihang","fullname":"Beihang University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ba7666c138c8f2b7844b58/n98lZU9VWxYgWIkzE_6o4.jpeg"},"isAuthorParticipating":false},{"paper":{"id":"2609.11808","authors":[{"_id":"6aa399db47a406da7901e831","name":"Mohamed Eltahir","hidden":false},{"_id":"6aa399db47a406da7901e832","name":"Talal Aloushan","hidden":false},{"_id":"6aa399db47a406da7901e833","name":"Rose Khairoalsendi","hidden":false},{"_id":"6aa399db47a406da7901e834","name":"Jana Shata","hidden":false},{"_id":"6aa399db47a406da7901e835","name":"Mohammed Alhassan","hidden":false},{"_id":"6aa399db47a406da7901e836","name":"Leen Alrehaili","hidden":false},{"_id":"6aa399db47a406da7901e837","name":"Tanveer Hussain","hidden":false},{"_id":"6aa399db47a406da7901e838","name":"Naeemullah Khan","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/xxMXWNZHmbrtioq5ajrQk.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/BTMD42iBWGobxVrnGEC0Q.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/QyVO6eD9ZU2wfHrQOR2Ut.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/VfkjdPO-43vXEQ1wUhh_D.png"],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Generative Late-Interaction Embeddings For Visual Document Retrieval","submittedOnDailyBy":{"_id":"6331c242e092098b57bd8e58","avatarUrl":"/avatars/8bcaf3cb3482a002ded96d3206b04947.svg","isPro":false,"fullname":"Mohamed Eltahir","user":"mohammad2012191","type":"user","name":"mohammad2012191"},"summary":"Late-interaction retrieval is the state-of-the-art for visual document search, but it pays for its accuracy in storage. Existing compression methods retain a subset or local average of the N~1,000 vectors per page. Under aggressive storage budgets, however, these methods degrade sharply, and alternatives require retraining the encoder. Investigating this degradation across three encoders, we found two consistent properties: the vectors lie exactly on the unit sphere and concentrate near a manifold of intrinsic dimension five to six. This geometry yields two insights. First, standard k-means centroids fall inside the sphere, causing systematic underestimation of MaxSim scores. Normalizing them to the surface is a free correction worth up to +0.093 nDCG@5 over raw centroids. Second, because the page manifold has few degrees of freedom, the full set of vectors can be regenerated from only a few. To this end, we introduce Generative Late-Interaction Embeddings (GLIE): k << N vectors per page learned from the normalized centroids to serve as both a lightweight index and a basis for regenerating the page's full embedding set. At query time, search runs exclusively on these k vectors, and a decoder expands only the top candidates back to all N vectors for exact rescoring. At four vectors per page on ViDoRe v1, GLIE retains nearly 80% of the uncompressed system's nDCG@5, against 70% for the best prior post-hoc method. These results use a 415K-parameter network fitted in under three GPU-minutes on just a thousand training pages. At a matched training budget, fine-tuning the encoder does not reach even the training-free stage of GLIE, and the full system beats it at every budget. These patterns hold across a second encoder and ViDoRe v2. By reconstructing evidence on demand rather than sampling it, GLIE opens a new axis for storage-efficient retrieval, with the decoder as its main design surface.","upvotes":2,"discussionId":"6aa399dc47a406da7901e839","projectPage":"https://mohammad2012191.github.io/GLIE/","githubRepo":"https://github.com/mohammad2012191/GLIE","githubRepoAddedBy":"user","ai_summary":"Generative Late-Interaction Embeddings compress visual document retrieval vectors by learning a small basis set that regenerates full embeddings on demand, improving accuracy under tight storage limits without retraining the encoder.","ai_keywords":["late-interaction retrieval","MaxSim","k-means","unit sphere","manifold","intrinsic dimension","Generative Late-Interaction Embeddings","GLIE","decoder","nDCG@5","ViDoRe"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":2,"organization":{"_id":"642bf38ba208ae9adcebe075","name":"KAUST","fullname":"King Abdullah University of Science and Technology","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6315fb0b29411a6864b05b35/6egitr9tcwgl5i6ikCbkY.jpeg"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Generative Late-Interaction Embeddings For Visual Document Retrieval","summary":"Late-interaction retrieval is the state-of-the-art for visual document search, but it pays for its accuracy in storage. Existing compression methods retain a subset or local average of the N~1,000 vectors per page. Under aggressive storage budgets, however, these methods degrade sharply, and alternatives require retraining the encoder. Investigating this degradation across three encoders, we found two consistent properties: the vectors lie exactly on the unit sphere and concentrate near a manifold of intrinsic dimension five to six. This geometry yields two insights. First, standard k-means centroids fall inside the sphere, causing systematic underestimation of MaxSim scores. Normalizing them to the surface is a free correction worth up to +0.093 nDCG@5 over raw centroids. Second, because the page manifold has few degrees of freedom, the full set of vectors can be regenerated from only a few. To this end, we introduce Generative Late-Interaction Embeddings (GLIE): k << N vectors per page learned from the normalized centroids to serve as both a lightweight index and a basis for regenerating the page's full embedding set. At query time, search runs exclusively on these k vectors, and a decoder expands only the top candidates back to all N vectors for exact rescoring. At four vectors per page on ViDoRe v1, GLIE retains nearly 80% of the uncompressed system's nDCG@5, against 70% for the best prior post-hoc method. These results use a 415K-parameter network fitted in under three GPU-minutes on just a thousand training pages. At a matched training budget, fine-tuning the encoder does not reach even the training-free stage of GLIE, and the full system beats it at every budget. These patterns hold across a second encoder and ViDoRe v2. By reconstructing evidence on demand rather than sampling it, GLIE opens a new axis for storage-efficient retrieval, with the decoder as its main design surface.","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/xxMXWNZHmbrtioq5ajrQk.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/BTMD42iBWGobxVrnGEC0Q.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/QyVO6eD9ZU2wfHrQOR2Ut.png","https://cdn-uploads.huggingface.co/production/uploads/6331c242e092098b57bd8e58/VfkjdPO-43vXEQ1wUhh_D.png"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11808.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6331c242e092098b57bd8e58","avatarUrl":"/avatars/8bcaf3cb3482a002ded96d3206b04947.svg","fullname":"Mohamed Eltahir","name":"mohammad2012191","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"organization":{"_id":"642bf38ba208ae9adcebe075","name":"KAUST","fullname":"King Abdullah University of Science and Technology","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6315fb0b29411a6864b05b35/6egitr9tcwgl5i6ikCbkY.jpeg"},"isAuthorParticipating":false},{"paper":{"id":"2609.11085","authors":[{"_id":"6aa4243c7ba345d44ad142bd","user":{"_id":"6702352c547acbd64cddf31e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6702352c547acbd64cddf31e/elkGeA8piG-K-mXspBOhi.png","isPro":false,"fullname":"Vikash Singh","user":"optimusPrimeBee","type":"user","name":"optimusPrimeBee"},"name":"Vikash Singh","status":"claimed_verified","statusLastChangedAt":"2026-09-11T16:45:04.572Z","hidden":false},{"_id":"6aa4243c7ba345d44ad142be","name":"Debargha Ganguly","hidden":false},{"_id":"6aa4243c7ba345d44ad142bf","name":"Aman Goel","hidden":false},{"_id":"6aa4243c7ba345d44ad142c0","name":"Ali Torkamani","hidden":false},{"_id":"6aa4243c7ba345d44ad142c1","name":"Xiaoxue Han","hidden":false},{"_id":"6aa4243c7ba345d44ad142c2","name":"Joseph Lilien","hidden":false},{"_id":"6aa4243c7ba345d44ad142c3","name":"Ferhat Erata","hidden":false},{"_id":"6aa4243c7ba345d44ad142c4","name":"Vipin Chaudhary","hidden":false}],"publishedAt":"2026-09-10T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Beyond Solver Verdicts: Generative Reward Models for Autoformalization","submittedOnDailyBy":{"_id":"6702352c547acbd64cddf31e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6702352c547acbd64cddf31e/elkGeA8piG-K-mXspBOhi.png","isPro":false,"fullname":"Vikash Singh","user":"optimusPrimeBee","type":"user","name":"optimusPrimeBee"},"summary":"Neurosymbolic systems rely on mathematical solvers to guarantee reasoning correctness, yet solvers are fundamentally blind to whether a formal translation maintains strict reference-equivalence to a designated formalization. We formalize this vulnerability as Verdict-Preserving-Unfaithfulness (VPU): a failure mode where an incorrect encoding executes successfully and matches the expected verdict. We theoretically prove that structural, verdict-only verification heuristics are mathematically bounded to chance-level detection on these deceptively valid traces. To resolve this, we introduce Generative Verification (GenV), which distills an offline Z3-equivalence oracle into a reference-free, continuous reference-equivalence score by repurposing the language model's native vocabulary space. Mechanistic analysis via decision-projected logit lenses and sparse autoencoders shows this generative readout natively extracts precise spatial error coordinates without explicit localization training. Empirically, our oracle-mined verifier (GenV+HN) achieves 0.961 AUROC in reference-equivalence verification, generalizes zero-shot across unseen translators and divergent formal styles, and yields an 11.3-point downstream accuracy gain in agentic test-time compute allocation.","upvotes":1,"discussionId":"6aa4243c7ba345d44ad142c5","ai_summary":"Neurosymbolic reasoning is vulnerable to incorrect but verdict-matching formal translations, which are addressed by a generative verification method that scores reference equivalence without an oracle and improves downstream accuracy.","ai_keywords":["Verdict-Preserving-Unfaithfulness","Z3-equivalence oracle","Generative Verification","decision-projected logit lenses","sparse autoencoders","reference-equivalence verification"],"ai_summary_model":"thinkingmachines/Inkling-Small","organization":{"_id":"6058ec29102f61b42f65ae35","name":"AWS","fullname":"Amazon Web Services","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66f19ed428ae41c20c470792/wtuzZxWzijQ3do3zYoOFH.png"}},"publishedAt":"2026-09-09T20:00:00.000Z","title":"Beyond Solver Verdicts: Generative Reward Models for Autoformalization","summary":"Neurosymbolic systems rely on mathematical solvers to guarantee reasoning correctness, yet solvers are fundamentally blind to whether a formal translation maintains strict reference-equivalence to a designated formalization. We formalize this vulnerability as Verdict-Preserving-Unfaithfulness (VPU): a failure mode where an incorrect encoding executes successfully and matches the expected verdict. We theoretically prove that structural, verdict-only verification heuristics are mathematically bounded to chance-level detection on these deceptively valid traces. To resolve this, we introduce Generative Verification (GenV), which distills an offline Z3-equivalence oracle into a reference-free, continuous reference-equivalence score by repurposing the language model's native vocabulary space. Mechanistic analysis via decision-projected logit lenses and sparse autoencoders shows this generative readout natively extracts precise spatial error coordinates without explicit localization training. Empirically, our oracle-mined verifier (GenV+HN) achieves 0.961 AUROC in reference-equivalence verification, generalizes zero-shot across unseen translators and divergent formal styles, and yields an 11.3-point downstream accuracy gain in agentic test-time compute allocation.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11085.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6702352c547acbd64cddf31e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6702352c547acbd64cddf31e/elkGeA8piG-K-mXspBOhi.png","fullname":"Vikash Singh","name":"optimusPrimeBee","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false},"organization":{"_id":"6058ec29102f61b42f65ae35","name":"AWS","fullname":"Amazon Web Services","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/66f19ed428ae41c20c470792/wtuzZxWzijQ3do3zYoOFH.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.10745","authors":[{"_id":"6aa351cd47a406da7901e688","user":{"_id":"62886c851f542794efe6548d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62886c851f542794efe6548d/GzGAjr_vncJN4FdJ0TaXi.jpeg","isPro":false,"fullname":"Parinthapat Pengpun","user":"parinzee","type":"user","name":"parinzee"},"name":"Parinthapat Pengpun","status":"claimed_verified","statusLastChangedAt":"2026-09-11T08:45:04.696Z","hidden":false},{"_id":"6aa351cd47a406da7901e689","name":"Simran Khanuja","hidden":false},{"_id":"6aa351cd47a406da7901e68a","name":"Graham Neubig","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/62886c851f542794efe6548d/mMcB42BgYIfqvFvTwSJ4g.png"],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking","submittedOnDailyBy":{"_id":"62886c851f542794efe6548d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62886c851f542794efe6548d/GzGAjr_vncJN4FdJ0TaXi.jpeg","isPro":false,"fullname":"Parinthapat Pengpun","user":"parinzee","type":"user","name":"parinzee"},"summary":"Multimodal entity linking grounds entity mentions in text and images to knowledge-base entries. These systems degrade on rare entities, but prior work measures rarity primarily through popularity-based metrics such as pageviews. We broaden this view using knowledge-graph structural metrics that capture how well an entity is documented and connected. These metrics identify many rare entities that popularity metrics miss. Across the resulting rare-entity slices, state-of-the-art accuracy drops by 15.4-39.9%, showing that different rarity definitions expose different failure modes. To address these failures, we introduce a simple, training-free framework in which a reasoning-capable vision-language model iteratively searches and reasons over Wikipedia, gathering evidence dynamically. Controlled experiments show that reasoning and retrieval are complementary. Reasoning alone does not significantly improve accuracy on rare entities. Retrieval without reasoning improves rare-entity accuracy but can hurt overall accuracy. Their combination performs best. On MERLIN, a multilingual multimodal entity linking benchmark over five languages (Hindi, Indonesian, Japanese, Tamil, Vietnamese), our best system improves over the state of the art by 6.9% overall and by up to 23.3% on rare-entity slices. We release MERLIN-Rare, rare-entity test slices for targeted evaluation, with our framework.","upvotes":1,"discussionId":"6aa351ce47a406da7901e68b","projectPage":"https://neulab.github.io/think-before-you-link/","githubRepo":"https://github.com/neulab/think-before-you-link","githubRepoAddedBy":"user","ai_summary":"A reasoning-capable vision-language model that iteratively retrieves and reasons over Wikipedia improves multimodal entity linking for rare entities defined by knowledge-graph structure.","ai_keywords":["multimodal entity linking","knowledge-graph structural metrics","vision-language model","iterative retrieval","reasoning","MERLIN"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":2,"organization":{"_id":"62cd831b7a036fc9941c147a","name":"neulab","fullname":"NeuLab @ LTI/CMU","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1657821011553-62cd8153248f9e6bc20ab250.png"}},"publishedAt":"2026-09-08T20:00:00.000Z","title":"Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking","summary":"Multimodal entity linking grounds entity mentions in text and images to knowledge-base entries. These systems degrade on rare entities, but prior work measures rarity primarily through popularity-based metrics such as pageviews. We broaden this view using knowledge-graph structural metrics that capture how well an entity is documented and connected. These metrics identify many rare entities that popularity metrics miss. Across the resulting rare-entity slices, state-of-the-art accuracy drops by 15.4-39.9%, showing that different rarity definitions expose different failure modes. To address these failures, we introduce a simple, training-free framework in which a reasoning-capable vision-language model iteratively searches and reasons over Wikipedia, gathering evidence dynamically. Controlled experiments show that reasoning and retrieval are complementary. Reasoning alone does not significantly improve accuracy on rare entities. Retrieval without reasoning improves rare-entity accuracy but can hurt overall accuracy. Their combination performs best. On MERLIN, a multilingual multimodal entity linking benchmark over five languages (Hindi, Indonesian, Japanese, Tamil, Vietnamese), our best system improves over the state of the art by 6.9% overall and by up to 23.3% on rare-entity slices. We release MERLIN-Rare, rare-entity test slices for targeted evaluation, with our framework.","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/62886c851f542794efe6548d/mMcB42BgYIfqvFvTwSJ4g.png"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10745.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"62886c851f542794efe6548d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62886c851f542794efe6548d/GzGAjr_vncJN4FdJ0TaXi.jpeg","fullname":"Parinthapat Pengpun","name":"parinzee","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":22,"isUserFollowing":false},"organization":{"_id":"62cd831b7a036fc9941c147a","name":"neulab","fullname":"NeuLab @ LTI/CMU","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1657821011553-62cd8153248f9e6bc20ab250.png"},"isAuthorParticipating":true},{"paper":{"id":"2609.10445","authors":[{"_id":"6aa3c20147a406da7901e93b","user":{"_id":"67896868bd21dd0c757c1e65","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-5G7C2VjMHOBLZSifP3N8.png","isPro":false,"fullname":"Mehrnaz M","user":"Mhrnz","type":"user","name":"Mhrnz"},"name":"Mehrnaz Mofakhami","status":"claimed_verified","statusLastChangedAt":"2026-09-11T12:32:33.356Z","hidden":false},{"_id":"6aa3c20147a406da7901e93c","name":"Ananya Sahu","hidden":false},{"_id":"6aa3c20147a406da7901e93d","name":"Alejandro R. Salamanca","hidden":false},{"_id":"6aa3c20147a406da7901e93e","name":"Daniel D&#39;souza","hidden":false},{"_id":"6aa3c20147a406da7901e93f","name":"Alexandre Berard","hidden":false},{"_id":"6aa3c20147a406da7901e940","name":"Thomas Euyang","hidden":false},{"_id":"6aa3c20147a406da7901e941","name":"Marzieh Fadaee","hidden":false},{"_id":"6aa3c20147a406da7901e942","name":"Julia Kreutzer","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/67896868bd21dd0c757c1e65/6JG5iNxKk3ju2BhKmKD5C.png","https://cdn-uploads.huggingface.co/production/uploads/67896868bd21dd0c757c1e65/6DVt_KsYjf8DsnnylsMAA.png"],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language Reasoning","submittedOnDailyBy":{"_id":"67896868bd21dd0c757c1e65","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-5G7C2VjMHOBLZSifP3N8.png","isPro":false,"fullname":"Mehrnaz M","user":"Mhrnz","type":"user","name":"Mhrnz"},"summary":"Reasoning language models have made substantial advances on a variety of complex tasks, yet their capabilities remain overwhelmingly English-centric: models primarily reason in English regardless of the language they are prompted in. This is inaccessible for non-English-speaking users, risks losing the intent of the original question, and forgoes knowledge more readily expressed in the target language. In this work, we advance L2 reasoning, the ability of a model to reason consistently in the language of the user's prompt, thus building an in-language bridge between the prompt and the answer. We approach this problem from a data-centric angle, investigating how to optimize data composition and scheduling in SFT for reasoning generalization. Building Tiny Aya L2-Thinker at 3.35B scale, we achieve an L2 reasoning rate above 93% across 60 languages on 6 benchmarks spanning math, commonsense reasoning, instruction following, open-ended generation, and cultural reasoning while keeping performance strong. We show the path to generalizing L2 reasoning to held-out languages goes through broader language coverage, readily available multilingual non-reasoning data, and a sufficient English reasoning backbone. These findings indicate that reasoning is a language-agnostic behavior that can be transferred across typologically diverse languages through careful data mixing and without requiring reasoning supervision in every target language. We release our model weights and multilingual reasoning data to support further research on accessible, in-language reasoning.","upvotes":1,"discussionId":"6aa3c20147a406da7901e943","projectPage":"https://cohere.com/research/papers/building-multilingual-bridges-2026-09-10","ai_summary":"Optimized supervised fine-tuning data composition enables reasoning models to consistently process and respond in diverse non-English languages without requiring reasoning supervision in each target language.","ai_keywords":["reasoning language models","L2 reasoning","supervised fine-tuning","SFT","reasoning generalization","multilingual reasoning","data scheduling","data composition"],"ai_summary_model":"thinkingmachines/Inkling-Small","organization":{"_id":"640ca1c93623f6a56ddab373","name":"CohereLabs","fullname":"Cohere Labs","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1678549441248-5e70f6048ce3c604d78fe133.png"}},"publishedAt":"2026-09-08T20:00:00.000Z","title":"Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language Reasoning","summary":"Reasoning language models have made substantial advances on a variety of complex tasks, yet their capabilities remain overwhelmingly English-centric: models primarily reason in English regardless of the language they are prompted in. This is inaccessible for non-English-speaking users, risks losing the intent of the original question, and forgoes knowledge more readily expressed in the target language. In this work, we advance L2 reasoning, the ability of a model to reason consistently in the language of the user's prompt, thus building an in-language bridge between the prompt and the answer. We approach this problem from a data-centric angle, investigating how to optimize data composition and scheduling in SFT for reasoning generalization. Building Tiny Aya L2-Thinker at 3.35B scale, we achieve an L2 reasoning rate above 93% across 60 languages on 6 benchmarks spanning math, commonsense reasoning, instruction following, open-ended generation, and cultural reasoning while keeping performance strong. We show the path to generalizing L2 reasoning to held-out languages goes through broader language coverage, readily available multilingual non-reasoning data, and a sufficient English reasoning backbone. These findings indicate that reasoning is a language-agnostic behavior that can be transferred across typologically diverse languages through careful data mixing and without requiring reasoning supervision in every target language. We release our model weights and multilingual reasoning data to support further research on accessible, in-language reasoning.","mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/67896868bd21dd0c757c1e65/6JG5iNxKk3ju2BhKmKD5C.png","https://cdn-uploads.huggingface.co/production/uploads/67896868bd21dd0c757c1e65/6DVt_KsYjf8DsnnylsMAA.png"],"thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10445.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"67896868bd21dd0c757c1e65","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-5G7C2VjMHOBLZSifP3N8.png","fullname":"Mehrnaz M","name":"Mhrnz","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"640ca1c93623f6a56ddab373","name":"CohereLabs","fullname":"Cohere Labs","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1678549441248-5e70f6048ce3c604d78fe133.png"},"isAuthorParticipating":true},{"paper":{"id":"2608.15380","authors":[{"_id":"6a8648aadb13816030683df7","user":{"_id":"6907cc03e77c7a753da36afa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6907cc03e77c7a753da36afa/bJ_keZ3p9tK4jQLqn4pjO.jpeg","isPro":false,"fullname":"Kaushalraj","user":"puwar","type":"user","name":"puwar"},"name":"Kaushalraj Puwar","status":"claimed_verified","statusLastChangedAt":"2026-08-21T12:58:00.615Z","hidden":false},{"_id":"6aa4383c7ba345d44ad1432a","name":"B. Thangaraju","hidden":false}],"publishedAt":"2026-09-06T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"Adaptive Bridge: A Proxy-Based Decoupling Layer for Mitigating DDS Backpressure in ROS 2","submittedOnDailyBy":{"_id":"6907cc03e77c7a753da36afa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6907cc03e77c7a753da36afa/bJ_keZ3p9tK4jQLqn4pjO.jpeg","isPro":false,"fullname":"Kaushalraj","user":"puwar","type":"user","name":"puwar"},"summary":"In systems built on Robot Operating System 2 (ROS 2) and using Data Distribution Service (DDS), a single network-impaired or throttled subscriber on a RELIABLE topic can cause backpressure that degrades throughput and latency for all other subscribers, including safety-critical ones sharing the publisher, because the publisher's DDS writer can no longer accept new samples. We present Adaptive Bridge, a proxy-based layer that decouples critical subscribers from degraded or noncritical ones, thereby isolating the critical path through topic splitting and dynamic rate control. The proxy acts as a middleman and subscribes to the original topic and republishes the messages to two independent DDS writers: one RELIABLE writer for critical nodes and one BEST EFFORT writer for noncritical or degraded nodes, thus isolating the degraded nodes and safeguarding the publisher and critical nodes from backpressure. A probe-based classifier actively monitors subscriber health through sampling with hysteresis and adjusts subscriber rate limits in real time. We evaluate the system under a Gilbert-Elliott bursty wireless loss model using a reproducible Docker-based harness. The results show that using the Adaptive Bridge in our evaluation harness reduces the critical subscriber tail p95 latency from up to 15 s to 1.55 ms across all impairment severities while preserving the publisher's configured throughput.","upvotes":1,"discussionId":"6a8648aadb13816030683df8","githubRepo":"https://github.com/KaushalrajPuwar/adaptive-bridge","githubRepoAddedBy":"user","ai_summary":"A proxy layer isolates critical ROS 2 subscribers from degraded ones via topic splitting and dynamic rate control to eliminate DDS backpressure.","ai_keywords":["ROS 2","DDS","RELIABLE","BEST EFFORT","Adaptive Bridge","topic splitting","dynamic rate control","probe-based classifier","Gilbert-Elliott"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0},"publishedAt":"2026-09-05T20:00:00.000Z","title":"Adaptive Bridge: A Proxy-Based Decoupling Layer for Mitigating DDS Backpressure in ROS 2","summary":"In systems built on Robot Operating System 2 (ROS 2) and using Data Distribution Service (DDS), a single network-impaired or throttled subscriber on a RELIABLE topic can cause backpressure that degrades throughput and latency for all other subscribers, including safety-critical ones sharing the publisher, because the publisher's DDS writer can no longer accept new samples. We present Adaptive Bridge, a proxy-based layer that decouples critical subscribers from degraded or noncritical ones, thereby isolating the critical path through topic splitting and dynamic rate control. The proxy acts as a middleman and subscribes to the original topic and republishes the messages to two independent DDS writers: one RELIABLE writer for critical nodes and one BEST EFFORT writer for noncritical or degraded nodes, thus isolating the degraded nodes and safeguarding the publisher and critical nodes from backpressure. A probe-based classifier actively monitors subscriber health through sampling with hysteresis and adjusts subscriber rate limits in real time. We evaluate the system under a Gilbert-Elliott bursty wireless loss model using a reproducible Docker-based harness. The results show that using the Adaptive Bridge in our evaluation harness reduces the critical subscriber tail p95 latency from up to 15 s to 1.55 ms across all impairment severities while preserving the publisher's configured throughput.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2608.15380.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"6907cc03e77c7a753da36afa","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6907cc03e77c7a753da36afa/bJ_keZ3p9tK4jQLqn4pjO.jpeg","fullname":"Kaushalraj","name":"puwar","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"isAuthorParticipating":true},{"paper":{"id":"2609.10539","authors":[{"_id":"6aa37d1a47a406da7901e797","name":"Yiling Ma","hidden":false},{"_id":"6aa37d1a47a406da7901e798","name":"Yilun Zhao","hidden":false},{"_id":"6aa37d1a47a406da7901e799","name":"Sihong Wu","hidden":false},{"_id":"6aa37d1a47a406da7901e79a","name":"Manasi Patwardhan","hidden":false},{"_id":"6aa37d1a47a406da7901e79b","name":"Arman Cohan","hidden":false}],"publishedAt":"2026-09-09T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"IdeaAMBIG: Benchmarking Implementation-Critical Gaps in Research-Idea Specifications","submittedOnDailyBy":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","isPro":false,"fullname":"YilingMa","user":"YilingMa","type":"user","name":"YilingMa"},"summary":"A research idea may be novel, coherent, and scientifically plausible, yet its proposed method may remain insufficiently specified for faithful implementation. We study the codification readiness of implementation-facing research-method specifications, defined by whether they provide sufficient methodological information for a competent implementer or coding agent to construct the intended method without unsupported assumptions. We construct evidence-grounded specifications and their supported resolutions from papers, codebases, issue threads, and reproduction artifacts. We introduce IdeaAMBIG, a benchmark of 660 evidence-grounded instances: 163 real-world gaps from reproducibility reports and GitHub issues, and 497 controlled synthetic gaps injected into codification-ready references. IdeaAMBIG evaluates three capabilities: codification-readiness assessment, defect localization, and clarification action generation. Defect localization receives only the specification, whereas clarification additionally receives the annotated defect. Across 13 LLMs, the best model achieves 9.6% Macro Defect Recovery Rate on real-world instances but 80.6% Macro Clarification Action Success Rate when given the defect. In an oracle study, supplying the gold resolution raises the downstream codification-ready rate from 14% to 98%. Across all evaluated models, defect localization is the main bottleneck, with stronger clarification given the defect.","upvotes":0,"discussionId":"6aa37d1a47a406da7901e79c","githubRepo":"https://github.com/Yiling-Ma/IdeaAMBIG","githubRepoAddedBy":"user","ai_summary":"The study introduces a benchmark to evaluate whether research methods are specified clearly enough for implementation, finding that identifying missing details is the primary challenge for language models.","ai_keywords":["codification readiness","evidence-grounded specifications","IdeaAMBIG","defect localization","clarification action generation","Macro Defect Recovery Rate","Macro Clarification Action Success Rate"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0,"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"}},"publishedAt":"2026-09-08T20:00:00.000Z","title":"IdeaAMBIG: Benchmarking Implementation-Critical Gaps in Research-Idea Specifications","summary":"A research idea may be novel, coherent, and scientifically plausible, yet its proposed method may remain insufficiently specified for faithful implementation. We study the codification readiness of implementation-facing research-method specifications, defined by whether they provide sufficient methodological information for a competent implementer or coding agent to construct the intended method without unsupported assumptions. We construct evidence-grounded specifications and their supported resolutions from papers, codebases, issue threads, and reproduction artifacts. We introduce IdeaAMBIG, a benchmark of 660 evidence-grounded instances: 163 real-world gaps from reproducibility reports and GitHub issues, and 497 controlled synthetic gaps injected into codification-ready references. IdeaAMBIG evaluates three capabilities: codification-readiness assessment, defect localization, and clarification action generation. Defect localization receives only the specification, whereas clarification additionally receives the annotated defect. Across 13 LLMs, the best model achieves 9.6% Macro Defect Recovery Rate on real-world instances but 80.6% Macro Clarification Action Success Rate when given the defect. In an oracle study, supplying the gold resolution raises the downstream codification-ready rate from 14% to 98%. Across all evaluated models, defect localization is the main bottleneck, with stronger clarification given the defect.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10539.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","fullname":"YilingMa","name":"YilingMa","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"},"isAuthorParticipating":false},{"paper":{"id":"2609.09076","authors":[{"_id":"6aa1453bff4bf7311191aaf0","name":"Yiling Ma","hidden":false},{"_id":"6aa1453bff4bf7311191aaf1","name":"Yilun Zhao","hidden":false},{"_id":"6aa1453bff4bf7311191aaf2","name":"Sihong Wu","hidden":false},{"_id":"6aa1453bff4bf7311191aaf3","name":"Ziyu Chen","hidden":false},{"_id":"6aa1453bff4bf7311191aaf4","name":"Manasi Patwardhan","hidden":false},{"_id":"6aa1453bff4bf7311191aaf5","name":"Arman Cohan","hidden":false}],"publishedAt":"2026-09-08T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation","submittedOnDailyBy":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","isPro":false,"fullname":"YilingMa","user":"YilingMa","type":"user","name":"YilingMa"},"summary":"As LLMs are increasingly used for pre-submission self-review, there is growing demand for feedback that not only identifies weaknesses but also guides authors toward concrete revisions. We study this as Actionable Peer-review Generation and decompose it into two subtasks: diagnostic claim generation and revision suggestion generation. We introduce ActReview, a rebuttal-guided post-training framework that connects paper-specific diagnoses to concrete, grounded revision plans. Our central insight is that author rebuttals reveal plausible actions for addressing reviewer concerns and can therefore provide latent supervision for revision-oriented feedback. From real review-rebuttal threads on OpenReview, we construct ActReview-40K by aligning reviewer weaknesses with author responses and grounding the resulting feedback in localized paper evidence. We post-train Qwen3-8B-Base with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. We also introduce ActReview-Bench, a human-curated benchmark of 1,000 instances for evaluating diagnostic quality and revision usefulness. Experiments show that ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness while revealing a remaining gap in technical accuracy, and additional analyses support generalization to held-out papers and robustness across independent judges.","upvotes":0,"discussionId":"6aa1453cff4bf7311191aaf6","githubRepo":"https://github.com/Yiling-Ma/ActReview","githubRepoAddedBy":"user","ai_summary":"ActReview is a rebuttal-guided post-training framework that generates diagnostic claims and concrete revision suggestions for peer review by leveraging author responses as latent supervision.","ai_keywords":["Actionable Peer-review Generation","diagnostic claim generation","revision suggestion generation","ActReview","rebuttal-guided post-training","multi-task supervised fine-tuning","GRPO","candidate-aware rubric rewards","ActReview-40K","ActReview-Bench"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0,"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"}},"publishedAt":"2026-09-07T20:00:00.000Z","title":"ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation","summary":"As LLMs are increasingly used for pre-submission self-review, there is growing demand for feedback that not only identifies weaknesses but also guides authors toward concrete revisions. We study this as Actionable Peer-review Generation and decompose it into two subtasks: diagnostic claim generation and revision suggestion generation. We introduce ActReview, a rebuttal-guided post-training framework that connects paper-specific diagnoses to concrete, grounded revision plans. Our central insight is that author rebuttals reveal plausible actions for addressing reviewer concerns and can therefore provide latent supervision for revision-oriented feedback. From real review-rebuttal threads on OpenReview, we construct ActReview-40K by aligning reviewer weaknesses with author responses and grounding the resulting feedback in localized paper evidence. We post-train Qwen3-8B-Base with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. We also introduce ActReview-Bench, a human-curated benchmark of 1,000 instances for evaluating diagnostic quality and revision usefulness. Experiments show that ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness while revealing a remaining gap in technical accuracy, and additional analyses support generalization to held-out papers and robustness across independent judges.","thumbnail":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.09076.png","numComments":1,"upvoted":false,"submittedBy":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","fullname":"YilingMa","name":"YilingMa","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"},"isAuthorParticipating":false}],"prevDate":"2026-09-10","publisher":{"_id":"60f1abe7544c2adfd699860c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg","isPro":false,"fullname":"AK","user":"akhaliq","type":"user"},"periodType":"day","query":{},"isTrending":false}"><section class="container relative mb-20 mt-8 md:mt-14"><div class="mb-8 grid grid-cols-6 items-start md:mb-12 md:grid-cols-12 md:gap-x-4"><div class="order-1 col-span-5 md:order-none md:col-span-10 lg:col-span-8 xl:col-span-4 xl:pl-0"><div class="flex items-center gap-3"><h1 class="text-2xl font-bold md:text-3xl">Daily Papers</h1> <!--[-1--><!--]--></div> <h2 class="whitespace-nowrap text-gray-500 xl:text-lg">by<a href="/akhaliq"><img alt="" class="mx-1.5 inline h-4" src="/front/assets/papers-by.png"/><span class="underline">AK</span></a> and the research community</h2></div> <div class="order-1 col-span-1 flex justify-end pt-1 md:order-none md:col-span-2 md:mt-0 lg:col-span-4 xl:hidden"><!--[-1--><!--]--></div> <div class="order-3 col-span-6 mt-3 flex items-center self-start md:col-span-12 md:mt-4 lg:col-span-12 xl:order-none xl:col-span-4 xl:mt-0"><div class="relative z-1 flex w-full items-center"><svg class="absolute left-[1.1rem] select-none text-gray-700 dark:text-gray-300" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 11 11"><path fill="currentColor" d="M4.881 4.182c0 .101-.031.2-.087.283a.5.5 0 0 1-.242.18l-.65.217a1.3 1.3 0 0 0-.484.299 1.3 1.3 0 0 0-.298.484l-.222.639a.46.46 0 0 1-.18.242.5.5 0 0 1-.288.092.5.5 0 0 1-.294-.097.5.5 0 0 1-.175-.242l-.211-.644a1.26 1.26 0 0 0-.299-.48 1.14 1.14 0 0 0-.479-.298L.328 4.64a.48.48 0 0 1-.247-.18.515.515 0 0 1 .247-.758l.644-.21a1.28 1.28 0 0 0 .788-.789l.211-.634a.5.5 0 0 1 .165-.242.5.5 0 0 1 .283-.103.5.5 0 0 1 .294.083c.086.058.152.14.19.237l.217.659a1.28 1.28 0 0 0 .788.788l.644.222a.476.476 0 0 1 .237.18.5.5 0 0 1 .092.288"></path><path fill="currentColor" d="M10.031 7.458a.5.5 0 0 1-.098.314.5.5 0 0 1-.267.196l-.881.293c-.272.09-.519.242-.721.443a1.8 1.8 0 0 0-.443.721l-.31.876a.5.5 0 0 1-.185.263.56.56 0 0 1-.319.098.515.515 0 0 1-.515-.366l-.294-.88a1.8 1.8 0 0 0-.443-.722c-.204-.2-.45-.353-.72-.448l-.881-.288a.57.57 0 0 1-.263-.191.56.56 0 0 1-.014-.64.5.5 0 0 1 .271-.194l.886-.294A1.82 1.82 0 0 0 6.01 5.465l.293-.87a.515.515 0 0 1 .49-.377c.11 0 .219.03.314.088a.56.56 0 0 1 .206.263l.298.896a1.82 1.82 0 0 0 1.175 1.174l.875.31a.5.5 0 0 1 .263.195c.07.09.108.2.108.314"></path><path fill="currentColor" d="M7.775 1.684a.5.5 0 0 0 .088-.262.45.45 0 0 0-.088-.263.5.5 0 0 0-.21-.155L7.24.896a.5.5 0 0 1-.165-.103.5.5 0 0 1-.103-.17l-.108-.33a.5.5 0 0 0-.165-.21A.5.5 0 0 0 6.426 0a.5.5 0 0 0-.252.098.5.5 0 0 0-.145.206l-.108.32a.5.5 0 0 1-.103.17.5.5 0 0 1-.17.102L5.334 1a.45.45 0 0 0-.216.155.5.5 0 0 0-.088.262c0 .094.029.186.083.263a.5.5 0 0 0 .216.16l.32.103q.095.03.164.103a.37.37 0 0 1 .103.165l.108.319c.031.09.088.17.165.227a.56.56 0 0 0 .252.077.42.42 0 0 0 .268-.093.5.5 0 0 0 .15-.2l.113-.325a.43.43 0 0 1 .268-.268l.32-.108a.42.42 0 0 0 .215-.155"></path></svg><!----> <input type="text" class="hidden" name="daily-papers-search-fake"/> <input type="search" name="daily-papers-search" autocomplete="off" class="shadow-alternate-sm max-w-full flex-1 rounded-full border-gray-200 py-2 pl-10 pr-12 placeholder:text-gray-400 focus:ring-2 focus:ring-blue-500/30 dark:bg-gray-850 dark:focus:ring-blue-500 svelte-k3q7or" placeholder="Search any paper with AI" value="" maxlength="250"/> <button class="absolute grid size-7 place-items-center rounded-full border bg-gray-200 transition-all hover:brightness-95 dark:border-gray-700 dark:bg-gray-800 dark:hover:brightness-110 right-2"><svg class="text-sm text-gray-600" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.1em" height="1.1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M24 9.4L22.6 8L16 14.6L9.4 8L8 9.4l6.6 6.6L8 22.6L9.4 24l6.6-6.6l6.6 6.6l1.4-1.4l-6.6-6.6L24 9.4z" fill="currentColor"></path></svg><!----></button> <!--[0--><button class="dice-container absolute right-2 grid !size-7 place-items-center rounded-full border bg-gray-200 transition-all dark:border-gray-700 dark:bg-gray-800 svelte-166qlyn" aria-label="Roll dice"><div class="dice svelte-166qlyn" style="transform: function (new_value) {106 if (arguments.length === 0) {107 return updated_value ?? get_value();108 }109110 updated_value = new_value;111 return updated_value;112 };"><div class="face front svelte-166qlyn"></div> <div class="face up svelte-166qlyn"></div> <div class="face left svelte-166qlyn"></div> <div class="face right svelte-166qlyn"></div> <div class="face bottom svelte-166qlyn"></div> <div class="face back svelte-166qlyn"></div></div></button><!--]--></div></div> <div class="order-2 col-span-6 mt-3 flex w-full justify-center md:col-span-12 md:mt-4 lg:col-span-12 xl:order-none xl:col-span-4 xl:mt-0 xl:items-center xl:justify-end"><div class="flex w-full items-center justify-between w-full xl:w-auto"><div class="flex items-center"><ul class="flex gap-1 text-sm "><!--[--><li><button class="flex items-center whitespace-nowrap rounded-lg px-2 bg-black text-white dark:bg-gray-800">Daily <!--[-1--><!--]--></button></li><li><button class="flex items-center whitespace-nowrap rounded-lg px-2 text-gray-500 hover:bg-gray-100 hover:text-gray-700 dark:hover:bg-gray-900 dark:hover:text-gray-300">Weekly <!--[-1--><!--]--></button></li><li><button class="flex items-center whitespace-nowrap rounded-lg px-2 text-gray-500 hover:bg-gray-100 hover:text-gray-700 dark:hover:bg-gray-900 dark:hover:text-gray-300">Monthly <!--[-1--><!--]--></button></li><!--]--></ul><!----> <a href="/papers/trending" class="mx-3 flex size-5 items-center justify-center rounded transition-all text-amber-500 dark:bg-gray-850 dark:text-amber-400 dark:hover:text-amber-300" aria-label="Trending Papers"><svg class="h-4 w-4 flex-shrink-0" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" fill="none" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" stroke="#FFA51E" stroke-width=".6" d="m6.6 2.6.3 1c.2.4.6.8 1.2.8h1c.6 0 .8.8.3 1.1l-.8.6c-.4.3-.6.9-.4 1.4l.3 1c.2.5-.5 1-1 .6l-.8-.5a1 1 0 0 0-1.4 0l-.8.5a.6.6 0 0 1-1-.7l.3-1c.2-.4 0-1-.4-1.3l-.8-.6a.6.6 0 0 1 .3-1.1h1c.6 0 1-.4 1.2-.9l.3-.9c.2-.6 1-.6 1.2 0Z"></path></svg><!----></a></div> <div class="ml-6 flex items-center overflow-hidden rounded-lg bg-gray-100 dark:bg-gray-800"><!--[0--><a href="/papers/date/2026-09-10" class="flex size-8 items-center justify-center bg-gray-200 text-gray-800 hover:bg-gray-300 dark:bg-gray-700 dark:text-gray-200 dark:hover:bg-gray-600 "><svg class="h-2" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 10 10"><path d="M-2.30478e-07 4.95458L7.90909 0.388266L7.90909 9.5209L-2.30478e-07 4.95458Z" fill="currentColor"></path></svg><!----></a> <div class="w-24 whitespace-nowrap text-center text-sm font-semibold text-gray-900 dark:text-gray-100">Sep 11</div> <a class="flex size-8 items-center justify-center bg-gray-200 text-gray-800 hover:bg-gray-300 dark:bg-gray-700 dark:text-gray-200 dark:hover:bg-gray-600 pointer-events-none opacity-50"><svg class="h-2 rotate-180" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 10 10"><path d="M-2.30478e-07 4.95458L7.90909 0.388266L7.90909 9.5209L-2.30478e-07 4.95458Z" fill="currentColor"></path></svg><!----></a><!--]--></div></div><!----></div></div> <div class="relative grid gap-5 *:[content-visibility:auto] grid-cols-1 lg:grid-cols-2 xl:grid-cols-3"><!--[0--><!--[--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10715" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10715.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/d05320c370a6497d8792ef5acb563dd5.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> yuliang03181</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10715" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">157</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10715" class="line-clamp-3 cursor-pointer text-balance">NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.10715" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Shuang Cheng" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Chiyu Chen" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="NCP Team" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Rubin-Wei" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/6609a53bd81d611249ef5266/h31hdQFl-jhRnO8R6Gr4C.png" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Clover-Hill" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/U2qzEGvhPk7RjINeaIYid.jpeg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 28 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[1--><a href="/papers/2609.10715#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11929" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11929.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> taesiri</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11929" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">124</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11929" class="line-clamp-3 cursor-pointer text-balance">SenseNova-U1.5: Towards Native Unified Visual Intelligence</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11929" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="caizhongang" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/652d06833b5997ed71ce5c46/O_D6bpa5mGxLA7uCjmVCG.jpeg" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="tongww" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="/avatars/d166f5f8be666e96b522a0a0effd21c4.svg" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="WRHC" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="/avatars/53c3092905a4d0f677b09f410459845e.svg" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="TokenWang" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="/avatars/4c2ab907247fe071ff5cdd71c404ca7c.svg" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Paranioar" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/64b4a717aa03b6520839e9b8/Rt3ERG-6BVEA4hAwOz0_I.jpeg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 65 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[1--><a href="/papers/2609.11929#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.07064" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.07064.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/38459c92f2bc1b95a3ee43ce0234c9c8.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> rsoohyun</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.07064" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">55</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.07064" class="line-clamp-3 cursor-pointer text-balance">SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/kaist-ai" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png" alt="kaist-ai" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">KAIST AI</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.07064" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.07064#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.05903" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.05903.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/6bmu8qdzZtvQK37I4RiXH.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> andyc03</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.05903" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">36</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.05903" class="line-clamp-3 cursor-pointer text-balance">EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/JohnsHopkins" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/653944e58e687a41625a4694/qqHzBOarppVrUuZbbjqwh.png" alt="JohnsHopkins" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Johns Hopkins University</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.05903" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>1</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.05903#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11317" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11317.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/d2899498627e36d573965d713ec87c89.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> cnbird</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11317" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">20</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11317" class="line-clamp-3 cursor-pointer text-balance">Mi-Ripple: Restoring Images Degraded by Iterative AI Editing</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/Miyang-AI" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/69e1eb933ceb9605d4db2007/-cr0jWbvkbFP9qIHCwfXX.png" alt="Miyang-AI" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Miyang Technology (Shanghai) Co., Ltd.</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11317" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>32</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11317#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 2</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11412" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11412.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/Zl-orrGcbY0RbfjfKszn1.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> ShiyuHuang</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11412" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">13</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11412" class="line-clamp-3 cursor-pointer text-balance">X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/XPENG-AI" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/6406db5cd684369027166986/DAJtDU_HssW_R7WIw0USp.jpeg" alt="XPENG-AI" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">XPENG AI</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11412" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>7</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11412#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[0--><a href="/papers/2609.11561" title="View paper" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><video src="https://cdn-uploads.huggingface.co/production/uploads/63f47b5321eb234ab739e91a/clT7ifvITIQhw6ni64AqX.mp4" loop="" muted="" playsinline="" preload="metadata" class="pointer-events-none h-full w-full object-cover object-center"></video></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/63f47b5321eb234ab739e91a/vWfFNVtMkHl8gieha5PPd.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> hzxie</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11561" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">9</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11561" class="line-clamp-3 cursor-pointer text-balance">Memory as Plans: World-Action Modeling with Memory-Grounded Planning</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11561" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Huan Wang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Chenchu Zhang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Weiyu Zhao" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="hzxie" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/63f47b5321eb234ab739e91a/vWfFNVtMkHl8gieha5PPd.jpeg" loading="lazy"/><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="SizheZhao" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/6769003931100198233acfc9/4HOr-exIRu9lyBt8JQWZ2.jpeg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 8 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11561" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11561#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11499" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11499.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> taesiri</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11499" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">6</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11499" class="line-clamp-3 cursor-pointer text-balance">Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11499" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Bo Zhu" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Yuxuan Liao" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Zhiqi Li" style="content-visibility:auto;"><!--[-1--><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 3 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[-1--><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.01515" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.01515.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/bc58ab252c7b4d95ff99e4506fb8d3e9.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> CedPei</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.01515" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">6</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.01515" class="line-clamp-3 cursor-pointer text-balance">TempCloze: Can Video-LLMs Identify the Missing Middle?</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/hkuhk" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/67ea9e8d2d95c10a0da11b0c/FNnR4M7YqKRuG43N5771B.png" alt="hkuhk" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">The University of Hong Kong</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.01515" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>5</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.01515#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11486" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11486.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/258264afe2ea5048a4a7a8e9945d2f5b.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> a-yakovenko</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11486" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">5</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11486" class="line-clamp-3 cursor-pointer text-balance">FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11486" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Dmitriy Vatolin" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Khaled Abud" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Vladislav Bargatin" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="a-yakovenko" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="/avatars/258264afe2ea5048a4a7a8e9945d2f5b.svg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 4 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11486" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>2</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11486#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[0--><a href="/papers/2609.11548" title="View paper" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><video src="https://cdn-uploads.huggingface.co/production/uploads/6039478ab3ecf716b1a5fd4d/V2ktCpO1_KDxxmbRipRYA.mp4" loop="" muted="" playsinline="" preload="metadata" class="pointer-events-none h-full w-full object-cover object-center"></video></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> taesiri</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11548" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">5</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11548" class="line-clamp-3 cursor-pointer text-balance">World in World: Explore the World with World Models</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11548" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Chi Zhang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Yanming Yang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Chenxi Song" style="content-visibility:auto;"><!--[-1--><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 3 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[-1--><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10712" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10712.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6039478ab3ecf716b1a5fd4d/_Thy4E7taiSYBLKxEKJbT.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> taesiri</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10712" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">5</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10712" class="line-clamp-3 cursor-pointer text-balance">An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/nvidia" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/65df9200dc3292a8983e5017/Vs5FPVCH-VZBipV3qKTuy.png" alt="nvidia" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">NVIDIA</span></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[-1--><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11699" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11699.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uUY9F17MiN7NhGAg01Yom.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> PassionPrc</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11699" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">4</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11699" class="line-clamp-3 cursor-pointer text-balance">Negative Self-Distillation: Learning to Reason by Avoiding Flaws</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.11699" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Wei-Lin Chen" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Xinyu Zhu" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Shuyao Xu" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Zhepei Wei" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="PassionPrc" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uUY9F17MiN7NhGAg01Yom.png" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 6 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11699" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>4</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11699#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.06931" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.06931.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/ade0fe033c0821ea8a8ed487a0c10db4.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> benbayibaurba</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.06931" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">4</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.06931" class="line-clamp-3 cursor-pointer text-balance">CARDEA: Auditable Reasoning Grounded in Spatial Evidence for End-to-End Coronary Angiography Interpretation</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2609.06931" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Shih-Sheng Chang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Wei-Chun Wang" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Kee Koon Ng" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="Shih-Yen Hou" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="benbayibaurba" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="/avatars/ade0fe033c0821ea8a8ed487a0c10db4.svg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 5 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.06931" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.06931#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2608.27875" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2608.27875.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6847c20bb510d25b2322fa5c/LWov6B3_1BVbkYWewUyL-.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> jerrysfls</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2608.27875" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">4</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2608.27875" class="line-clamp-3 cursor-pointer text-balance">HyQuant: Hybrid-Precision Quantization for LLM Attention</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/SJTU" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/1676013394657-63e5ee22b6a40bf941da0928.png" alt="SJTU" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Shanghai Jiao Tong University</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2608.27875" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>8</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2608.27875#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10016" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10016.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/M-gCXZHVZFLLlvvYKHcG6.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> remcohendriks</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10016" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">3</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10016" class="line-clamp-3 cursor-pointer text-balance">MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/continker" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/63457dc5547c70e4b7ccc128/nptM94xE8POJPxfKryjn0.png" alt="continker" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Continker</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.10016" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.10016#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 2</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11155" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11155.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/8b1982e043a7ab28006b707dfff6c5dc.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> AaronLiu0702</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11155" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">2</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11155" class="line-clamp-3 cursor-pointer text-balance">DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/China666" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/64a1d2e171874b7465bfca6d/l9MJIPO4MtwFJP5X9zzUw.png" alt="China666" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Chinese Academy of Sciences</span></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[1--><a href="/papers/2609.11155#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11156" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11156.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/70e84c5c188688db8b29455061a410a1.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> upyzwup</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11156" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">2</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11156" class="line-clamp-3 cursor-pointer text-balance">UniH^3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/Beihang" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/63ba7666c138c8f2b7844b58/n98lZU9VWxYgWIkzE_6o4.jpeg" alt="Beihang" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Beihang University</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11156" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>5</span></a><!--]--> <!--[1--><a href="/papers/2609.11156#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11808" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11808.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/8bcaf3cb3482a002ded96d3206b04947.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> mohammad2012191</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11808" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">2</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11808" class="line-clamp-3 cursor-pointer text-balance">Generative Late-Interaction Embeddings For Visual Document Retrieval</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/KAUST" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/6315fb0b29411a6864b05b35/6egitr9tcwgl5i6ikCbkY.jpeg" alt="KAUST" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">King Abdullah University of Science and Technology</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.11808" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>2</span></a><!--]--> <!--[1--><a href="/papers/2609.11808#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.11085" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.11085.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6702352c547acbd64cddf31e/elkGeA8piG-K-mXspBOhi.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> optimusPrimeBee</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.11085" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">1</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.11085" class="line-clamp-3 cursor-pointer text-balance">Beyond Solver Verdicts: Generative Reward Models for Autoformalization</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/AWS" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/66f19ed428ae41c20c470792/wtuzZxWzijQ3do3zYoOFH.png" alt="AWS" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Amazon Web Services</span></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.11085#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10745" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10745.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/62886c851f542794efe6548d/GzGAjr_vncJN4FdJ0TaXi.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> parinzee</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10745" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">1</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10745" class="line-clamp-3 cursor-pointer text-balance">Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/neulab" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/1657821011553-62cd8153248f9e6bc20ab250.png" alt="neulab" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">NeuLab @ LTI/CMU</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.10745" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>2</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.10745#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10445" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10445.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-5G7C2VjMHOBLZSifP3N8.png" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> Mhrnz</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10445" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">1</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10445" class="line-clamp-3 cursor-pointer text-balance">Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language Reasoning</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/CohereLabs" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/1678549441248-5e70f6048ce3c604d78fe133.png" alt="CohereLabs" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Cohere Labs</span></a><!--]--> <div class="flex items-center gap-2"><!--[-1--><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2609.10445#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2608.15380" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2608.15380.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="https://cdn-avatars.huggingface.co/v1/production/uploads/6907cc03e77c7a753da36afa/bJ_keZ3p9tK4jQLqn4pjO.jpeg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> puwar</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2608.15380" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">1</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2608.15380" class="line-clamp-3 cursor-pointer text-balance">Adaptive Bridge: A Proxy-Based Decoupling Layer for Mitigating DDS Backpressure in ROS 2</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[-1--><a href="/papers/2608.15380" class="flex"><!--[0--><ul class="flex items-center text-gray-600 flex-row-reverse text-sm "><!--[--><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="B. Thangaraju" style="content-visibility:auto;"><!--[-1--><!--]--></li><li class=" -mr-2 h-4 w-4 md:h-5 md:w-5 bg-linear-to-br block flex-none rounded-full border-2 border-white from-gray-300 to-gray-100 dark:border-gray-900 dark:from-gray-600 dark:to-gray-800 " title="puwar" style="content-visibility:auto;"><!--[1--><img class="overflow-hidden rounded-full" alt="" src="https://cdn-avatars.huggingface.co/v1/production/uploads/6907cc03e77c7a753da36afa/bJ_keZ3p9tK4jQLqn4pjO.jpeg" loading="lazy"/><!--]--></li><!--]--> <!--[0--><li class="text-xs hover:text-gray-700 dark:text-gray-400 dark:hover:text-gray-300 order-first ml-3"><!--[--><div class="flex truncate text-sm text-gray-350 sm:text-base"><div class="mx-1">·</div> 2 authors</div><!--]--></li><!--]--></ul><!--]--><!----></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2608.15380" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[0--><span class="inline-block "><span class="contents"><a href="/papers/2608.15380#community" class="flex translate-y-px items-center gap-1 rounded-md border border-blue-200 bg-blue-600/10 px-0.5 py-0.5 text-xs text-blue-500 dark:border-blue-800/60 dark:bg-blue-800/20 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!----></span> <!--[-1--><!--]--></span><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.10539" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.10539.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/20f959a3314ce098fcd224ae999e2a8f.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> YilingMa</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.10539" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">-</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.10539" class="line-clamp-3 cursor-pointer text-balance">IdeaAMBIG: Benchmarking Implementation-Critical Gaps in Research-Idea Specifications</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/yale-nlp" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png" alt="yale-nlp" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Yale NLP Lab</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.10539" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[1--><a href="/papers/2609.10539#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--[-1--><article class="relative flex flex-col overflow-hidden rounded-xl border"><!--[-1--><a href="/papers/2609.09076" class="shadow-alternate-sm peer relative block h-56 w-full cursor-pointer overflow-hidden rounded-xl bg-white md:h-64"><div class="relative flex h-full w-full items-center justify-center overflow-hidden "><!--[0--><div class="absolute inset-0 animate-pulse flex items-center justify-center bg-gray-200 dark:bg-gray-800"></div><!--]--> <!--[-1--><!--]--> <img loading="lazy" src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2609.09076.png" decoding="async" alt="" class="h-full w-full object-cover object-top opacity-80 dark:opacity-70 dark:invert invisible" onload="this.__e=event" onerror="this.__e=event"/></div><!----></a><!--]--> <!--[0--><div class="shadow-xs pointer-events-none absolute right-2 top-56 -mt-8 flex h-6 items-center gap-1 self-end whitespace-nowrap rounded-md border bg-white px-2 text-xs leading-none text-gray-700 dark:bg-gray-900 dark:text-gray-400 sm:text-sm md:top-64">Submitted by <div class="flex-none "><img alt="" loading="lazy" class="max-w-none size-2.5 rounded-full flex-none select-none " src="/avatars/20f959a3314ce098fcd224ae999e2a8f.svg" crossorigin="anonymous"/> <!--[-1--><!--]--></div><!----> YilingMa</div><!--]--> <!--[-1--><!--]--> <div class="from-gray-50-to-white bg-linear-to-b -mt-2 flex px-6 pb-6 pt-8"><div class="flex w-full gap-6"><div class="flex flex-wrap items-center gap-2.5 pt-1 z-1 lg:sticky lg:top-8"><!--[0--><a href="/login?next=%2Fpapers%2F2609.09076" class="self-start"><!--[-1--><!--]--> <!--[0--><!----><div role="checkbox" class="shadow-alternate flex h-14 w-12 gap-1 rounded-xl flex-none cursor-pointer select-none flex-col items-center justify-center self-start border-gray-300 bg-white dark:bg-gray-850 "><input disabled="" type="checkbox" class="peer hidden"/> <svg class="text-sm peer-checked:text-gray-500 group-hover:text-gray-500" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 12 12"><path fill="currentColor" d="M5.19 2.67a.94.94 0 0 1 1.62 0l3.31 5.72a.94.94 0 0 1-.82 1.4H2.7a.94.94 0 0 1-.82-1.4l3.31-5.7v-.02Z"></path></svg><!----> <div class="leading-none">-</div><!----></div><!----><!--]--><!----></a><!--]--> <!--[-1--><!--]--></div> <!--[-1--><!--]--> <dialog class="shadow-alternate z-40 mx-4 my-auto h-fit select-text overflow-hidden rounded-xl bg-white max-sm:max-w-[calc(100dvw-2rem)] sm:mx-auto lg:mt-26 md:portrait:mt-30 xl:mt-30 2xl:mt-32 w-full sm:w-96 max-w-[calc(100%-4rem)] text-base "><div tabindex="-1" class="outline-none focus:ring-0 focus-visible:ring-0"><!--[-1--><!--]--></div></dialog><!----><!----> <div class="w-full min-w-0"><h3 class="mb-1 text-lg/6 font-semibold hover:underline peer-hover:underline 2xl:text-[1.2rem]/6"><a href="/papers/2609.09076" class="line-clamp-3 cursor-pointer text-balance">ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation</a></h3> <div class="flex min-w-0 items-center justify-between gap-1 sm:gap-2"><!--[0--><a href="/yale-nlp" class="inline-flex min-w-0 max-w-full items-center rounded-full border border-blue-200/70 bg-blue-50 py-0.5 text-blue-700 hover:border-blue-200 hover:bg-blue-100 dark:border-blue-500/20 dark:bg-blue-900/30 dark:text-blue-300 dark:hover:border-blue-500/30 dark:hover:bg-blue-900/50 px-2 text-sm gap-1.5 mt-1"><!--[0--><img src="https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png" alt="yale-nlp" class="size-3.5 rounded"/><!--]--> <span class="block min-w-0 truncate font-medium">Yale NLP Lab</span></a><!--]--> <div class="flex items-center gap-2"><!--[0--><a href="/papers/2609.09076" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg class="size-3 text-gray-800" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1.03em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 250"><path d="M128.001 0C57.317 0 0 57.307 0 128.001c0 56.554 36.676 104.535 87.535 121.46c6.397 1.185 8.746-2.777 8.746-6.158c0-3.052-.12-13.135-.174-23.83c-35.61 7.742-43.124-15.103-43.124-15.103c-5.823-14.795-14.213-18.73-14.213-18.73c-11.613-7.944.876-7.78.876-7.78c12.853.902 19.621 13.19 19.621 13.19c11.417 19.568 29.945 13.911 37.249 10.64c1.149-8.272 4.466-13.92 8.127-17.116c-28.431-3.236-58.318-14.212-58.318-63.258c0-13.975 5-25.394 13.188-34.358c-1.329-3.224-5.71-16.242 1.24-33.874c0 0 10.749-3.44 35.21 13.121c10.21-2.836 21.16-4.258 32.038-4.307c10.878.049 21.837 1.47 32.066 4.307c24.431-16.56 35.165-13.12 35.165-13.12c6.967 17.63 2.584 30.65 1.255 33.873c8.207 8.964 13.173 20.383 13.173 34.358c0 49.163-29.944 59.988-58.447 63.157c4.591 3.972 8.682 11.762 8.682 23.704c0 17.126-.148 30.91-.148 35.126c0 3.407 2.304 7.398 8.792 6.14C219.37 232.5 256 184.537 256 128.002C256 57.307 198.691 0 128.001 0zm-80.06 182.34c-.282.636-1.283.827-2.194.39c-.929-.417-1.45-1.284-1.15-1.922c.276-.655 1.279-.838 2.205-.399c.93.418 1.46 1.293 1.139 1.931zm6.296 5.618c-.61.566-1.804.303-2.614-.591c-.837-.892-.994-2.086-.375-2.66c.63-.566 1.787-.301 2.626.591c.838.903 1 2.088.363 2.66zm4.32 7.188c-.785.545-2.067.034-2.86-1.104c-.784-1.138-.784-2.503.017-3.05c.795-.547 2.058-.055 2.861 1.075c.782 1.157.782 2.522-.019 3.08zm7.304 8.325c-.701.774-2.196.566-3.29-.49c-1.119-1.032-1.43-2.496-.726-3.27c.71-.776 2.213-.558 3.315.49c1.11 1.03 1.45 2.505.701 3.27zm9.442 2.81c-.31 1.003-1.75 1.459-3.199 1.033c-1.448-.439-2.395-1.613-2.103-2.626c.301-1.01 1.747-1.484 3.207-1.028c1.446.436 2.396 1.602 2.095 2.622zm10.744 1.193c.036 1.055-1.193 1.93-2.715 1.95c-1.53.034-2.769-.82-2.786-1.86c0-1.065 1.202-1.932 2.733-1.958c1.522-.03 2.768.818 2.768 1.868zm10.555-.405c.182 1.03-.875 2.088-2.387 2.37c-1.485.271-2.861-.365-3.05-1.386c-.184-1.056.893-2.114 2.376-2.387c1.514-.263 2.868.356 3.061 1.403z" fill="currentColor"></path></svg><!----> <span>0</span></a><!--]--> <!--[1--><a href="/papers/2609.09076#community" class="flex translate-y-px items-center gap-1 rounded-md border border-gray-100 px-0.5 py-0.5 text-xs text-gray-500 sm:px-1 sm:text-sm"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 24 24"><path fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"></path></svg><!----> 1</a><!--]--></div></div></div></div></div></article><!--]--><!--]--><!--]--> <!--[-1--><!--]--> <!--[0--><div class="col-span-1 mt-8 flex lg:col-span-2 xl:col-span-3"><!--[0--><a class="btn gap-2" href="/papers/date/2026-09-10"><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M14 26l1.41-1.41L7.83 17H28v-2H7.83l7.58-7.59L14 6L4 16l10 10z" fill="currentColor"></path></svg><!---->Previous</a><!--]--> <!--[-1--><!--]--></div><!--]--></div></section><!----></div><!----></main> <!--[1--><footer class="b-12 mb-2 flex border-t border-gray-100 md:h-14"><nav class="container relative flex flex-col justify-between space-y-2 py-6 text-gray-500 max-md:*:self-start md:flex-row md:items-center md:space-y-0 md:py-0 md:text-sm"><div class="SVELTE_HYDRATER contents" data-target="ThemeSwitcher" data-props="{"theme":"system","isLoggedIn":false,"menuClassNames":"md:-top-24","classNames":"max-md:mb-5 max-md:*:self-start"}"><!--[-1--><div class="relative inline-block max-md:mb-5 max-md:*:self-start"><button class="rounded-full border border-gray-100 pl-2 py-1 pr-2.5 flex items-center text-sm text-gray-500 bg-white hover:bg-purple-50 hover:border-purple-200 dark:hover:bg-gray-800 dark:hover:border-gray-950 dark:border-gray-800 " type="button"><!--[-1--><!--[0--><!--[--><svg class="mr-1.5 text-gray-500" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M29 25H3a1 1 0 1 0 0 2h26a1 1 0 1 0 0-2Z" fill="currentColor"></path><path fill-rule="evenodd" clip-rule="evenodd" d="M6 22.5h20a2 2 0 0 0 2-2V7a2 2 0 0 0-2-2H6a2 2 0 0 0-2 2v13.5a2 2 0 0 0 2 2ZM7 7a1 1 0 0 0-1 1v11a1 1 0 0 0 1 1h18a1 1 0 0 0 1-1V8a1 1 0 0 0-1-1H7Z" fill="currentColor"></path><path d="M6 8a1 1 0 0 1 1-1h18a1 1 0 0 1 1 1v11a1 1 0 0 1-1 1H7a1 1 0 0 1-1-1V8Z" fill="currentColor" fill-opacity=".4"></path><path d="M29 25H3a1 1 0 1 0 0 2h26a1 1 0 1 0 0-2Z" fill="currentColor"></path></svg><!--]--><!--]--> System theme<!--]--> <!--[-1--><!--]--></button> <!--[-1--><!--]--> <!--[-1--><!--]--></div><!--]--><!----></div> <div class="font-semibold text-black md:hidden">Company</div> <a class="hover:underline" href="/terms-of-service">TOS</a> <a class="hover:underline" href="/privacy">Privacy</a> <a class="hover:underline" href="/huggingface">About</a> <a class="hover:underline" href="https://apply.workable.com/huggingface/">Careers</a> <a href="/" class="max-md:mb-4! max-md:mt-8! group flex-none max-md:order-last" aria-label="Hugging Face"><svg class="h-7 w-7 transition-transform group-hover:-translate-y-px" viewBox="0 0 95 88" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M47.2119 76.5C66.4037 76.5 81.9619 60.9419 81.9619 41.75C81.9619 22.5581 66.4037 7 47.2119 7C28.02 7 12.4619 22.5581 12.4619 41.75C12.4619 60.9419 28.02 76.5 47.2119 76.5Z" fill="#FFD21E"></path><path d="M81.9619 41.75C81.9619 22.5581 66.4037 7 47.2119 7C28.02 7 12.4619 22.5581 12.4619 41.75C12.4619 60.9419 28.02 76.5 47.2119 76.5C66.4037 76.5 81.9619 60.9419 81.9619 41.75ZM8.46185 41.75C8.46185 20.349 25.8108 3 47.2119 3C68.6129 3 85.9619 20.349 85.9619 41.75C85.9619 63.151 68.6129 80.5 47.2119 80.5C25.8108 80.5 8.46185 63.151 8.46185 41.75Z" fill="#FF9D0B"></path><path d="M58.5024 32.2915C59.7768 32.7415 60.2839 35.3615 61.5713 34.6769C64.0095 33.3805 64.9351 30.353 63.6387 27.9148C62.3423 25.4767 59.3148 24.5511 56.8766 25.8475C54.4384 27.1439 53.5128 30.1714 54.8092 32.6096C55.4211 33.7604 57.3632 31.8892 58.5024 32.2915Z" fill="#3A3B45"></path><path d="M34.9454 32.2915C33.671 32.7415 33.164 35.3615 31.8766 34.6769C29.4384 33.3805 28.5128 30.353 29.8092 27.9148C31.1056 25.4767 34.1331 24.5511 36.5713 25.8475C39.0095 27.1439 39.9351 30.1714 38.6387 32.6096C38.0268 33.7604 36.0846 31.8892 34.9454 32.2915Z" fill="#3A3B45"></path><path d="M46.9619 56.289C56.7903 56.289 59.9619 47.5261 59.9619 43.0262C59.9619 40.6875 58.3898 41.4236 55.8718 42.6702C53.5449 43.8222 50.4102 45.4101 46.9619 45.4101C39.7822 45.4101 33.9619 38.5263 33.9619 43.0262C33.9619 47.5261 37.1334 56.289 46.9619 56.289Z" fill="#3A3B45"></path><mask id="mask0" style="mask-type:alpha" maskUnits="userSpaceOnUse" x="33" y="41" width="27" height="16"><path d="M46.9619 56.289C56.7903 56.289 59.9619 47.5261 59.9619 43.0262C59.9619 40.6875 58.3898 41.4236 55.8718 42.6702C53.5449 43.8222 50.4102 45.4101 46.9619 45.4101C39.7822 45.4101 33.9619 38.5263 33.9619 43.0262C33.9619 47.5261 37.1334 56.289 46.9619 56.289Z" fill="white"></path></mask><g mask="url(#mask0)"><path d="M47.2119 66.5C52.0018 66.5 55.8848 62.617 55.8848 57.8271C55.8848 54.0962 53.5291 50.9156 50.224 49.6915C50.1023 49.6464 49.9794 49.604 49.8553 49.5643C49.0219 49.2979 48.1337 52.1623 47.2119 52.1623C46.3506 52.1623 45.5186 49.2797 44.7332 49.5135C41.151 50.5799 38.5389 53.8984 38.5389 57.8271C38.5389 62.617 42.4219 66.5 47.2119 66.5Z" fill="#F94040"></path></g><path d="M70.7119 37C72.5068 37 73.9619 35.5449 73.9619 33.75C73.9619 31.9551 72.5068 30.5 70.7119 30.5C68.9169 30.5 67.4619 31.9551 67.4619 33.75C67.4619 35.5449 68.9169 37 70.7119 37Z" fill="#FF9D0B"></path><path d="M24.2119 37C26.0068 37 27.4619 35.5449 27.4619 33.75C27.4619 31.9551 26.0068 30.5 24.2119 30.5C22.4169 30.5 20.9619 31.9551 20.9619 33.75C20.9619 35.5449 22.4169 37 24.2119 37Z" fill="#FF9D0B"></path><path class="origin-bottom-right transition-transform group-hover:-rotate-6" d="M17.5238 48C15.9048 48 14.4578 48.665 13.4488 49.871C12.8248 50.618 12.1728 51.822 12.1198 53.625C11.4408 53.43 10.7878 53.321 10.1778 53.321C8.6278 53.321 7.2278 53.915 6.2378 54.994C4.9658 56.379 4.4008 58.081 4.6468 59.784C4.7638 60.595 5.0348 61.322 5.4398 61.995C4.5858 62.686 3.9568 63.648 3.6528 64.805C3.4148 65.712 3.1708 67.601 4.4448 69.547C4.3638 69.674 4.2878 69.806 4.2168 69.941C3.4508 71.395 3.4018 73.038 4.0778 74.568C5.1028 76.887 7.6498 78.714 12.5958 80.675C15.6728 81.895 18.4878 82.675 18.5128 82.682C22.5808 83.737 26.2598 84.273 29.4448 84.273C35.2988 84.273 39.4898 82.48 41.9018 78.944C45.7838 73.25 45.2288 68.042 40.2058 63.022C37.4258 60.244 35.5778 56.148 35.1928 55.249C34.4168 52.587 32.3648 49.628 28.9538 49.628H28.9528C28.6658 49.628 28.3758 49.651 28.0898 49.696C26.5958 49.931 25.2898 50.791 24.3568 52.085C23.3498 50.833 22.3718 49.837 21.4868 49.275C20.1528 48.429 18.8198 48 17.5238 48ZM17.5238 52C18.0338 52 18.6568 52.217 19.3438 52.653C21.4768 54.006 25.5928 61.081 27.0998 63.833C27.6048 64.755 28.4678 65.145 29.2448 65.145C30.7868 65.145 31.9908 63.612 29.3858 61.664C25.4688 58.733 26.8428 53.942 28.7128 53.647C28.7948 53.634 28.8758 53.628 28.9538 53.628C30.6538 53.628 31.4038 56.558 31.4038 56.558C31.4038 56.558 33.6018 62.078 37.3778 65.851C41.1538 69.625 41.3488 72.654 38.5968 76.69C36.7198 79.442 33.1268 80.273 29.4448 80.273C25.6258 80.273 21.7108 79.379 19.5168 78.81C19.4088 78.782 6.0658 75.013 7.7558 71.805C8.0398 71.266 8.5078 71.05 9.0968 71.05C11.4768 71.05 15.8058 74.592 17.6668 74.592C18.0828 74.592 18.3758 74.415 18.4958 73.983C19.2888 71.138 6.4388 69.942 7.5218 65.821C7.7128 65.092 8.2308 64.796 8.9588 64.797C12.1038 64.797 19.1598 70.328 20.6388 70.328C20.7518 70.328 20.8328 70.295 20.8768 70.225C21.6178 69.029 21.2118 68.194 15.9888 65.033C10.7658 61.871 7.0998 59.969 9.1848 57.699C9.4248 57.437 9.7648 57.321 10.1778 57.321C13.3488 57.322 20.8408 64.14 20.8408 64.14C20.8408 64.14 22.8628 66.243 24.0858 66.243C24.3668 66.243 24.6058 66.132 24.7678 65.858C25.6348 64.396 16.7148 57.636 16.2118 54.847C15.8708 52.957 16.4508 52 17.5238 52Z" fill="#FF9D0B"></path><path class="origin-bottom-right transition-transform group-hover:-rotate-6" d="M38.5967 76.6898C41.3487 72.6538 41.1537 69.6248 37.3777 65.8508C33.6017 62.0778 31.4037 56.5578 31.4037 56.5578C31.4037 56.5578 30.5827 53.3518 28.7127 53.6468C26.8427 53.9418 25.4697 58.7328 29.3867 61.6638C33.3037 64.5938 28.6067 66.5848 27.0997 63.8328C25.5927 61.0808 21.4777 54.0058 19.3437 52.6528C17.2107 51.2998 15.7087 52.0578 16.2117 54.8468C16.7147 57.6358 25.6357 64.3958 24.7677 65.8588C23.8997 67.3208 20.8407 64.1398 20.8407 64.1398C20.8407 64.1398 11.2687 55.4288 9.18465 57.6988C7.10065 59.9688 10.7657 61.8708 15.9887 65.0328C21.2127 68.1938 21.6177 69.0288 20.8767 70.2248C20.1347 71.4208 8.60465 61.6998 7.52165 65.8208C6.43965 69.9418 19.2887 71.1378 18.4957 73.9828C17.7027 76.8288 9.44465 68.5978 7.75565 71.8048C6.06565 75.0128 19.4087 78.7818 19.5167 78.8098C23.8267 79.9278 34.7727 82.2968 38.5967 76.6898Z" fill="#FFD21E"></path><path class="origin-bottom-left transition-transform group-hover:rotate-6" d="M77.3999 48C79.0189 48 80.4659 48.665 81.4749 49.871C82.0989 50.618 82.7509 51.822 82.8039 53.625C83.4829 53.43 84.1359 53.321 84.7459 53.321C86.2959 53.321 87.6959 53.915 88.6859 54.994C89.9579 56.379 90.5229 58.081 90.2769 59.784C90.1599 60.595 89.8889 61.322 89.4839 61.995C90.3379 62.686 90.9669 63.648 91.2709 64.805C91.5089 65.712 91.7529 67.601 90.4789 69.547C90.5599 69.674 90.6359 69.806 90.7069 69.941C91.4729 71.395 91.5219 73.038 90.8459 74.568C89.8209 76.887 87.2739 78.714 82.3279 80.675C79.2509 81.895 76.4359 82.675 76.4109 82.682C72.3429 83.737 68.6639 84.273 65.4789 84.273C59.6249 84.273 55.4339 82.48 53.0219 78.944C49.1399 73.25 49.6949 68.042 54.7179 63.022C57.4979 60.244 59.3459 56.148 59.7309 55.249C60.5069 52.587 62.5589 49.628 65.9699 49.628H65.9709C66.2579 49.628 66.5479 49.651 66.8339 49.696C68.3279 49.931 69.6339 50.791 70.5669 52.085C71.5739 50.833 72.5519 49.837 73.4369 49.275C74.7709 48.429 76.1039 48 77.3999 48ZM77.3999 52C76.8899 52 76.2669 52.217 75.5799 52.653C73.4469 54.006 69.3309 61.081 67.8239 63.833C67.3189 64.755 66.4559 65.145 65.6789 65.145C64.1369 65.145 62.9329 63.612 65.5379 61.664C69.4549 58.733 68.0809 53.942 66.2109 53.647C66.1289 53.634 66.0479 53.628 65.9699 53.628C64.2699 53.628 63.5199 56.558 63.5199 56.558C63.5199 56.558 61.3219 62.078 57.5459 65.851C53.7699 69.625 53.5749 72.654 56.3269 76.69C58.2039 79.442 61.7969 80.273 65.4789 80.273C69.2979 80.273 73.2129 79.379 75.4069 78.81C75.5149 78.782 88.8579 75.013 87.1679 71.805C86.8839 71.266 86.4159 71.05 85.8269 71.05C83.4469 71.05 79.1179 74.592 77.2569 74.592C76.8409 74.592 76.5479 74.415 76.4279 73.983C75.6349 71.138 88.4849 69.942 87.4019 65.821C87.2109 65.092 86.6929 64.796 85.9649 64.797C82.8199 64.797 75.7639 70.328 74.2849 70.328C74.1719 70.328 74.0909 70.295 74.0469 70.225C73.3059 69.029 73.7119 68.194 78.9349 65.033C84.1579 61.871 87.8239 59.969 85.7389 57.699C85.4989 57.437 85.1589 57.321 84.7459 57.321C81.5749 57.322 74.0829 64.14 74.0829 64.14C74.0829 64.14 72.0609 66.243 70.8379 66.243C70.5569 66.243 70.3179 66.132 70.1559 65.858C69.2889 64.396 78.2089 57.636 78.7119 54.847C79.0529 52.957 78.4729 52 77.3999 52Z" fill="#FF9D0B"></path><path class="origin-bottom-left transition-transform group-hover:rotate-6" d="M56.3271 76.6898C53.5751 72.6538 53.7701 69.6248 57.5461 65.8508C61.3221 62.0778 63.5201 56.5578 63.5201 56.5578C63.5201 56.5578 64.3411 53.3518 66.2111 53.6468C68.0811 53.9418 69.4541 58.7328 65.5371 61.6638C61.6201 64.5938 66.3171 66.5848 67.8241 63.8328C69.3311 61.0808 73.4461 54.0058 75.5801 52.6528C77.7131 51.2998 79.2151 52.0578 78.7121 54.8468C78.2091 57.6358 69.2881 64.3958 70.1561 65.8588C71.0241 67.3208 74.0831 64.1398 74.0831 64.1398C74.0831 64.1398 83.6551 55.4288 85.7391 57.6988C87.8231 59.9688 84.1581 61.8708 78.9351 65.0328C73.7111 68.1938 73.3061 69.0288 74.0471 70.2248C74.7891 71.4208 86.3191 61.6998 87.4021 65.8208C88.4841 69.9418 75.6351 71.1378 76.4281 73.9828C77.2211 76.8288 85.4791 68.5978 87.1681 71.8048C88.8581 75.0128 75.5151 78.7818 75.4071 78.8098C71.0971 79.9278 60.1511 82.2968 56.3271 76.6898Z" fill="#FFD21E"></path></svg></a> <div class="max-md:mt-8! font-semibold text-black md:hidden">Website</div> <a class="hover:underline" href="/models">Models</a> <a class="hover:underline" href="/datasets">Datasets</a> <a class="hover:underline" href="/spaces">Spaces</a> <a class="hover:underline" href="/pricing">Pricing</a> <a class="hover:underline" href="/docs">Docs</a></nav></footer><!--]--><!----></div><!--]-->113 <script>114 import("\/front\/build\/kube-8a47f95\/index.js"); window.moonSha = "kube-8a47f95\/"; window.__hf_deferred =115 {};116 </script>117 <!-- Stripe -->118 <script>119 if (["hf.co", "huggingface.co"].includes(window.location.hostname)) {120 const script = document.createElement("script");121 script.src = "https://js.stripe.com/v3/";122 script.async = true;123 document.head.appendChild(script);124 }125 </script>126 </body>127</html>128