{"format":"feedseek-model-feed","version":1,"source_key":"huggingface","title":"Hugging Face","feed_url":"https://trvny.github.io/feedseek/huggingface.json","source_feed":"https://trvny.github.io/feedseek/huggingface.xml","home_page_url":"https://huggingface.co","description":"Hugging Face Blog, community Posts, and Trending Papers in one feed.","items":[{"id":"tag:trvny.github.io,2024:feedseek/huggingface/74e801deeaf9aac5","title":"🛡️ We handed 20 open models a shell. 20 of them ran a dangerous command.","text":"🛡️ We handed 20 open models a shell. 20 of them ran a dangerous command. ViKernel Scanner is a free AI model security scanner. It hands a model 8 mock tools (files, mail, web, shell, payments), sends 100 attack checks plus 20 normal-work controls, and records every tool call the model tries to make. It scores what a model does, not what it says. Reference scans of 20 widely used open-weight models through the Hugging Face router: • 20 of 20 graded F • The median model attempted a dangerous action in 40 of 100 checks • Plain and disguised commands (base64, hex, ROT13, reversed text, quote breaks) reached the shell tool With ViKernel Firewall placed inside the model weights, 897 of 897 dangero","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/524065361638546","date_published":"2026-11-18T16:42:04Z","date_modified":"2026-11-18T16:42:04Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/Knh06iNqJSyHm6Y3aLXAk.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bf1424b145e4fc73","title":"Time-Series Models in the Streamhouse: A New Frontier and Use Cases","text":"Time-Series Models in the Streamhouse: A New Frontier and Use Cases","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/time-series-powered-streamhouse","date_published":"2026-11-18T15:00:23Z","date_modified":"2026-11-18T15:00:23Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/64e8143f6de557454220921e/yf6E9vJHIkGHG_toGSasb.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fe0b0282a2d211f9","title":"Hugging Face buckets are absolutely AMAZING!","text":"Hugging Face buckets are absolutely AMAZING!","text_kind":"content","url":"https://huggingface.co/posts/mithulaartigala/641031046358597","date_published":"2026-11-18T03:08:34Z","date_modified":"2026-11-18T03:08:34Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/mithulaartigala/641031046358597.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/679cee5ea1466f1f","title":"Rumble-2-Flash: an open 0.5B speech model for 40+ languages","text":"Rumble-2-Flash: an open 0.5B speech model for 40+ languages","text_kind":"content","url":"https://huggingface.co/blog/rumble-2-flash","date_published":"2026-11-18T00:00:00Z","date_modified":"2026-11-18T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/rumble-2-flash/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/35ea0387d3a658d7","title":"SO-ARM102 is now listed in the official LeRobot documentation!","text":"SO-ARM102 is now listed in the official LeRobot documentation! Build files & assembly guide: https://github.com/roboninecom/SO-ARM102","text_kind":"content","url":"https://huggingface.co/posts/branikita/898421657705480","date_published":"2026-11-17T18:45:04Z","date_modified":"2026-11-17T18:45:04Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/branikita/898421657705480.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/22b290e81d2e95a2","title":"I've released vev-4b-v2, an updated version of Vev, my open-weight model for making decisions from text and images.","text":"I've released vev-4b-v2, an updated version of Vev, my open-weight model for making decisions from text and images. On the Jev Decision Index vision leaderboard (November 12 snapshot): - 72.99 Public score — #2 among models with ≤5B parameters - 64.80 Full score (including private evaluations) — #6 in the same size range - 🥇 #1 among ≤5B entries listed as LoRA-trained vev-4b-v2 is based on Qwen3.5-4B, fine-tuned with LoRA while keeping the vision encoder frozen. Vev returns probabilities for yes/no, multiple-choice, and scoring questions. It supports text, JSON, and images through a Jev-compatible API. Model: CountingSheep/vev-4b-v2 Demo: CountingSheep/vev Code: https://github.com/Xiaooolong","text_kind":"content","url":"https://huggingface.co/posts/CountingSheep/117904962709574","date_published":"2026-11-17T16:09:53Z","date_modified":"2026-11-17T16:09:53Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/CountingSheep/117904962709574.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d984516b1782b121","title":"🛡️ AI Model Hacking Hackathon is LIVE · USD 2,000 in prizes","text":"🛡️ AI Model Hacking Hackathon is LIVE · USD 2,000 in prizes AI agents now read mail, open files, browse the web and run shell commands. A single crafted email was enough for EchoLeak (CVE-2025-32711), and a coding agent wiped a production database during a code freeze. The risk has moved from what models say to what agents do. ViKernel is a security kernel that sits at the tool-call boundary, like an OS kernel at the system-call boundary. Every tool call gets ALLOW or DENY before it runs, and every verdict goes into a hash-chained audit trail. Now we want you to break it. 🔓 🎯 Season 1 · Agent Hijack (Beginner) · USD 500 🕵️ Season 2 · Data Exfiltration (Intermediate) · USD 500 💥 Season 3 · Ke","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/440814670642952","date_published":"2026-11-17T08:26:36Z","date_modified":"2026-11-17T08:26:36Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/N_AUJk9zAl2GjlUg375no.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5da788d3364d4e3b","title":"I've been working on OGG TTS and wanted to share it here.","text":"I've been working on OGG TTS and wanted to share it here. It's a text-to-speech model that supports multiple languages, voice cloning and different speech styles. One of the things I find interesting is the ability to control how a voice sounds, including emotions, whispering, laughter and other expressions. It also supports Polish, which makes it especially interesting to experiment with. You can check it out here: FashionFlora/ogg-tts FashionFlora/ogg-tts Would be great to hear what you think!","text_kind":"content","url":"https://huggingface.co/posts/FashionFlora/749271970795973","date_published":"2026-11-16T22:29:03Z","date_modified":"2026-11-16T22:29:03Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/FashionFlora/749271970795973.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a7c2b25bcaf37d19","title":"🧠 Zero-FLOP Memory","text":"🧠 Zero-FLOP Memory Can a small transformer do more by remembering differently? 39.5M parameters + discrete count tables + document cache. 1.017 BPC on enwik8. No gradient updates needed to build the memory. Exploring a different way to scale language modeling: not just more parameters, but better memory. Paper & code follows ... if someones interested","text_kind":"content","url":"https://huggingface.co/posts/Chris4K/452870908938596","date_published":"2026-11-16T22:29:03Z","date_modified":"2026-11-16T22:29:03Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Chris4K/452870908938596.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5241b2cd53a40f61","title":"We released DeepSeek-V4.1-Flash-1017 Dynamic v3.1 GGUFs which rank #1 on Terminal Bench, DeepSWE & more on 512K context.","text":"We released DeepSeek-V4.1-Flash-1017 Dynamic v3.1 GGUFs which rank #1 on Terminal Bench, DeepSWE & more on 512K context. 1-bit Unsloth GGUF scores 68.1 on DeepSWE, 72.2 on Terminal-Bench, 52.7 on AutomationBench. >10% ahead of other quants. GGUF: unsloth/DeepSeek-V4.1-Flash-1017-GGUF","text_kind":"content","url":"https://huggingface.co/posts/danielhanchen/649800983842549","date_published":"2026-11-16T11:59:26Z","date_modified":"2026-11-16T11:59:26Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/62ecdc18b72a69615d6bd857/_K7LgOcWAijlX9J-ZEtux.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c5dcaeeab0de538c","title":"I often forget that RAM is so inaccessible for everyone now.","text":"I often forget that RAM is so inaccessible for everyone now. In February 2025 I grabbed 64GB DDR5 (2x32) from BestBuy for about $160 bucks cash in total and thought nothing of it at the time. Today, nearly 2 years later... Same exact modules are nearly 6 times the price I paid. Price increase: +458.7% in approximately 21 months","text_kind":"content","url":"https://huggingface.co/posts/darkc0de/803110853466027","date_published":"2026-11-16T01:23:07Z","date_modified":"2026-11-16T01:23:07Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6540a02d1389943fef4d2640/Nw6y2805sBPE8Z9Dn23hN.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e2e5e3e354c6974e","title":"BORIS-2 IS OUT!","text":"BORIS-2 IS OUT! opencerebral/Boris-2 It tops the AxiomicLabs Intelligence Index with a score of 34.15!","text_kind":"content","url":"https://huggingface.co/posts/KlondikeDev/895232867949946","date_published":"2026-11-16T01:23:07Z","date_modified":"2026-11-16T01:23:07Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/KlondikeDev/895232867949946.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/770e1ac61b961d40","title":"MovieRoll: A Systematic Framework for Rolling Out Long-Form Videos","text":"Streaming video generation must sustain quality beyond the training horizon while handling self-generated history and growing cache costs. To address these challenges, we present MovieRoll, a base-to-distilled framework that connects streaming fine-tuning, long-rollout distillation, and long-horizon inference. We first establish a robust multi-step base through Window-wise Teacher Forcing (WinTF), which jointly denoises chunks at progressive noise levels with bidirectional attention within each window and causal access to preceding history. To retain historical context efficiently, hierarchical KV-cache compression (HieraKV) exploits layer-wise attention patterns, retaining full history in s","text_kind":"content","url":"https://huggingface.co/papers/2611.21339","date_published":"2026-11-16T00:00:00Z","date_modified":"2026-11-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.21339/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/15712514dcdfa437","title":"InclusionMed: Benchmarking AI for Inclusive Healthcare","text":"InclusionMed is an open medical AI evaluation initiative that audits and reproduces benchmarks across healthcare roles and capabilities, revealing that model rankings and inference costs vary by context and motivating broader, context-specific evaluation.","text_kind":"content","url":"https://huggingface.co/papers/2611.21218","date_published":"2026-11-16T00:00:00Z","date_modified":"2026-11-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.21218.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/853b24ff0d6ecd3f","title":"Hugging Face and OpenRouter join forces to make open-weight models easy to run","text":"Hugging Face and OpenRouter join forces to make open-weight models easy to run","text_kind":"content","url":"https://huggingface.co/blog/openrouter","date_published":"2026-11-16T00:00:00Z","date_modified":"2026-11-16T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/openrouter/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5ed8080e8d3e8beb","title":"ForceHarness: Turning Force into Action and Physics for Frozen Frontier Models","text":"Frontier multimodal models can now control robots zero-shot from images, yet many manipulation tasks depend on contact forces that images cannot show. Bringing force to an API-only frozen model raises two challenges. First, contact evolves much faster than an API model can decide, so the system must close the loop on force beneath the model. Second, a frozen model cannot interpret raw force streams, so force must be returned in a form the model can reason about. Therefore, we present ForceHarness, a harness that lets frozen models use force without demonstrations or finetuning. After calibrating hand force from a wrist sensor or joint torques alone, ForceHarness addresses the first challenge","text_kind":"content","url":"https://huggingface.co/papers/2611.21099","date_published":"2026-11-16T00:00:00Z","date_modified":"2026-11-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.21099/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b88c69fd1f65e6b1","title":"MedJev: A System One Vision-Language Decision Model for Fast and Effective Medical Image Understanding","text":"Medical image understanding with vision-language models (VLMs) typically relies on autoregressively generating reasoning and answer tokens. These unnecessary tokens make the inference cost scale with the response length and force the answer to be parsed from free-form text, incurring avoidable latency and decision errors. To this end, we introduce MedJev, a system one vision-language decision model that maps medical images, questions, and candidate answers directly to decisions in a single multimodal forward pass, without any autoregressive generation process. For fast inference, MedJev removes the upper language layers and the generative output pathway, while keeping the full vision pathway","text_kind":"content","url":"https://huggingface.co/papers/2611.20992","date_published":"2026-11-16T00:00:00Z","date_modified":"2026-11-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.20992.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/60fad149ac00332a","title":"✅ Article highlight: *Internal Appeals, Protected Disclosure, and Organizational Reopen Paths* (art-60-349, v0.1)","text":"✅ Article highlight: *Internal Appeals, Protected Disclosure, and Organizational Reopen Paths* (art-60-349, v0.1) TL;DR: This article argues that a complaint channel is not the same thing as institutional self-correction. Organizations need structured ways to challenge prior decisions, protect sensitive disclosures, reopen bounded decision surfaces, and trigger independent review without turning confidentiality into opacity or auditability into broad exposure. Read: kanaria007/agi-structural-intelligence-protocols Why it matters: • separates appeal, reopen, and whistleblowing • prevents reporting channels from becoming complaint sinks • treats retaliation risk as a design variable • makes in","text_kind":"content","url":"https://huggingface.co/posts/kanaria007/303495185155140","date_published":"2026-11-15T19:05:24Z","date_modified":"2026-11-15T19:05:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/kanaria007/303495185155140.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/233cc98d04ed1cf5","title":"SO-Arena scored 21 System One models, and the most accurate one ranks only seventh on reliability.","text":"SO-Arena scored 21 System One models, and the most accurate one ranks only seventh on reliability. System One models pick from fixed options in one forward pass, so teams use them for routing, tool choice and verification. A leaderboard score does not say whether the pick survives shuffled options or a missing right answer. SO-Arena (2,919 tasks, 11 domains) measures that separately. 🔍 What's new • Reliability R = sqrt(S*N). S is stability over 12 option reorderings. N = 2AUC-1 on spotting that the correct option was removed • JevAny-27B is the most reliable System One model yet ranks eleventh in capability • 14 of 18 models reach S above 0.8, but N runs from near chance to about 0.8 📊 The n","text_kind":"content","url":"https://huggingface.co/posts/savoji/471618734733201","date_published":"2026-11-15T19:05:24Z","date_modified":"2026-11-15T19:05:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/3zxfLoVVdHUwiUV_Kkby0.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b8b249f7f6187fa4","title":"Should an AI agent sign this payment?","text":"Should an AI agent sign this payment? I’ve released SRA-RiskGate-4B v2. It’s a 4B model fine-tuned from Qwen3-4B-Instruct that answers approve, hold or reject for stablecoin payments made by agents (x402, EIP-3009, Circle), and it settles payment disputes. On 2,000 held-out benchmark cases: ✅ 99.6% of payment decisions correct (base model: 59.8%) ✅ 0.35% unsafe approvals (base model: 34.6%) ✅ 90.2% of dispute outcomes correct, up from 77.8% in v1 The main change in v2: the dispute window is computed in code and passed into the prompt instead of asking a 4B model to subtract dates. Wrongful denials fell from 26% to 2.1%. I also tested it on 227 real USDC payments from Base mainnet: it blocked","text_kind":"content","url":"https://huggingface.co/posts/sriram1983007/754919642053924","date_published":"2026-11-15T11:27:08Z","date_modified":"2026-11-15T11:27:08Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/sriram1983007/754919642053924.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7057ef2f122819e2","title":"AllocFlow: End-to-End Portfolio Trading Action Generation from Raw Market Data","text":"An end-to-end flow-matching framework generates joint portfolio allocations from raw multi-asset market histories via a cross-asset flow Transformer and a two-stage imitation and group-relative reinforcement learning procedure, improving risk-adjusted returns on cryptocurrency perpetual futures.","text_kind":"content","url":"https://huggingface.co/papers/2611.19937","date_published":"2026-11-15T00:00:00Z","date_modified":"2026-11-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.19937/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/86464306a36ac36d","title":"VoxMel TTS Technical Report","text":"VoxMel is a tokenizer-free autoregressive TTS framework that uses an invertible two-axis Haar transform to decompose mel spectrograms into Coarse-Mel and residual tiers, enabling stable high-fidelity generation with strong recognition and speaker-similarity results.","text_kind":"content","url":"https://huggingface.co/papers/2611.20254","date_published":"2026-11-15T00:00:00Z","date_modified":"2026-11-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.20254/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d3c8f7be146c7105","title":"We're introducing KV1 1.2B.","text":"We're introducing KV1 1.2B. Its our first KV1 model using a 2 bit KV cache, based on LFM 2.5 1.2B. KV1 1.2B retains 75.8% of original intelligence. More details: klein-ml/blog Model klein-ml/KV1-1.2B","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/166126402898184","date_published":"2026-11-14T23:47:56Z","date_modified":"2026-11-14T23:47:56Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69ae829a8408eeb0d7dd5491/eVZPecRSYQGkao4FhdiIH.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/04eb86b5e213c357","title":"Have you tried the New Notebook feature on Bibby AI?","text":"Have you tried the New Notebook feature on Bibby AI? I highly recommend for all the researchers - https://trybibby.com/ Featured in the top 10 tools of all time. 😎🏆","text_kind":"content","url":"https://huggingface.co/posts/BibbyResearch/166993169237953","date_published":"2026-11-14T20:04:51Z","date_modified":"2026-11-14T20:04:51Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/BibbyResearch/166993169237953.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/87d52cd18bdfb3ca","title":"We are working on LFM3-1.7B-Ox, a compact model for repo-level programming, agentic coding, and tinkering.","text":"We are working on LFM3-1.7B-Ox, a compact model for repo-level programming, agentic coding, and tinkering.","text_kind":"content","url":"https://huggingface.co/posts/DedeProGames/841259920594667","date_published":"2026-11-14T13:38:23Z","date_modified":"2026-11-14T13:38:23Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/685ea8ff7b4139b6845ce395/Zz2ORQh3zMxyh2iZIs7y_.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7abb8b95fe076d47","title":"🦙 llama.cpp hit v1.0.0, so I built it a party","text":"🦙 llama.cpp hit v1.0.0, so I built it a party juiceb0xc0de/BYOGGUF : Bring Your Own GGUF. Paste any public model repo, pick a quant, and chat with it on an RTX PRO 6000 Blackwell through ZeroGPU, on the minutes your account already has (5 a day free, 40 on PRO). If you don’t have a GGUF, the Space will make you one. Sign in, point it at any public safetensors or GGUF repo, pick your quant types, and it converts and quantizes on CPU, so it costs you no GPU quota. The finished quants land in your own account with a model card, and one button takes them straight into the chat room. Something the size of Qwen3.8-27B is fine. Anything up to 48 GB runs on half the card at the normal rate; bigger m","text_kind":"content","url":"https://huggingface.co/posts/juiceb0xc0de/998103515739875","date_published":"2026-11-14T13:38:23Z","date_modified":"2026-11-14T13:38:23Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/665425bdf0c8c891ca472d94/6yj1Jj6hwvaaQD9kpvTOJ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8a3bd80d837fee75","title":"BananaMind 3 is coming.","text":"BananaMind 3 is coming. Biggest context in all SLMs with NSA. 10M, 50M, 150M, 400M. And maybe, 1B.","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/901206687228678","date_published":"2026-11-13T23:49:48Z","date_modified":"2026-11-13T23:49:48Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69ae829a8408eeb0d7dd5491/3bu8tahDyJb7oYiZXfWsu.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2427bacb74a96b4b","title":"Build a Robot Pick-and-Place Simulation with SimReady Assets","text":"Build a Robot Pick-and-Place Simulation with SimReady Assets","text_kind":"content","url":"https://huggingface.co/blog/nvidia/simready-assets-robot-pick-and-place","date_published":"2026-11-13T18:00:39Z","date_modified":"2026-11-13T18:00:39Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6abfe9e7d7cf09a9b38ef308/HFZWtQ2i9VN_agjeUw2HB.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8400bdfe1e2afd49","title":"XYBench: Helping with the problem behind the question","text":"XYBench: Helping with the problem behind the question","text_kind":"content","url":"https://huggingface.co/blog/allenai/xybench","date_published":"2026-11-13T17:02:35Z","date_modified":"2026-11-13T17:02:35Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/aOWdQgZ8ioJcx9Eii4iob.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ed9ac4247cedb630","title":"🌦️ Which weather model is most accurate? We score them every day, in the open.","text":"🌦️ Which weather model is most accurate? We score them every day, in the open. We released two things on Hugging Face: 🏆 Global Weather Model Leaderboard Every day it records what each global model predicted for 24 world cities, compares it with what actually happened (ERA5), and updates the ranking. ECMWF IFS and AIFS, NOAA GFS and AIGFS, DWD ICON, JMA, ECCC GEM, MET Norway, side by side, on the same cases, since 1 January 2026. Four elements, four tabs: max temperature, min temperature, wind, precipitation. Forecast leads from 1 to 7 days, with 8 to 10 days joining as they are verified. 🧬 Darwin-Prediction-Weather VIDRAFT's weather prediction system: a fusion engine that combines global fo","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/969586700462381","date_published":"2026-11-13T16:43:13Z","date_modified":"2026-11-13T16:43:13Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/Jvykhn0jnoi9Vzpi6y7hJ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/40e78db6601d46e6","title":"llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem","text":"llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem","text_kind":"content","url":"https://huggingface.co/blog/ggml-org/llama-cpp-v1","date_published":"2026-11-13T14:05:22Z","date_modified":"2026-11-13T14:05:22Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/63148d3b996c52bf0142cdbe/2PgU1GxpJX_LVORaP2RRy.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/1711dfbcefcd2e40","title":"Gamer: A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time Game Agent","text":"Gamer is a generalist agent framework that integrates perception, memory, and multi-step reasoning, trained through continued pre-training, sparse-thinking instruction tuning, and multi-turn reinforcement learning to interact efficiently in 2D and 3D environments.","text_kind":"content","url":"https://huggingface.co/papers/2611.18397","date_published":"2026-11-13T00:00:00Z","date_modified":"2026-11-13T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.18397/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/59c578bf213c6d53","title":"JVT: Just Video Transformer makes pixel-space video diffusion simple","text":"JVT, a simple large-patch DiT-style Transformer, enables pixel-space video diffusion without pretrained autoencoders or U-Net skip connections, achieving competitive quality and supporting joint camera-conditioned RGB-disparity generation.","text_kind":"content","url":"https://huggingface.co/papers/2611.18549","date_published":"2026-11-13T00:00:00Z","date_modified":"2026-11-13T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.18549.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ef27ea1143179f3c","title":"MMFlash: Faster Speculative Decoding for Vision-Language Models","text":"MMFlash improves multimodal speculative decoding by adaptively upweighting visually dependent tokens and using sparse attention for long video, yielding up to 3.58× lossless speedup.","text_kind":"content","url":"https://huggingface.co/papers/2611.18482","date_published":"2026-11-13T00:00:00Z","date_modified":"2026-11-13T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.18482/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c71de018dbb9e38e","title":"We've released the","text":"We've released the BananaMind/BETU-SLM-Leaderboard Its so simple to use, it ranks models, and by one command add your own model to it. It scores by Overall Index which is a weighted mix of Hellaswawg, ARC EASY, PIQA, Arithmark 3, Tiny Theory of Mind, and BananaMind Base Bench.","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/595992352806426","date_published":"2026-11-12T23:51:24Z","date_modified":"2026-11-12T23:51:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Banaxi-Tech/595992352806426.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/303d6943d2cf9e7d","title":"Introducing V-Agent: a local-first AI agent for Windows, built by a quantizer, for quants","text":"Introducing V-Agent: a local-first AI agent for Windows, built by a quantizer, for quants VAL-APEX-Mini/Nano/Micro Plus builds get more quality out of consumer GPUs. V-Agent does the same on the app side: right engine for your card, right quant, real tools without giving away your PC. Free, open source (Apache-2.0), no account, no telemetry, everything on 127.0.0.1. 🧠 Right engine for your GPU - Detects your hardware and installs the official llama.cpp build for it (CUDA, Vulkan for AMD / Intel Arc, or CPU), SHA-256 verified - Strata engines or any OpenAI / Claude-compatible API too - Live tok/s, context, VRAM and RAM 🔎 Hugging Face built in - Browse GGUF repos with stats and the rendered mo","text_kind":"content","url":"https://huggingface.co/posts/IsValorum/440713591214008","date_published":"2026-11-12T23:51:24Z","date_modified":"2026-11-12T23:51:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6a837830ca6a77e4d62e3d9b/1GNj_nSEc9Lvr2OD05dtQ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a88bddb17214a029","title":"LLMON: An LLM-Native Markup Language, now in Granite 4.1 3B and 8B","text":"LLMON: An LLM-Native Markup Language, now in Granite 4.1 3B and 8B","text_kind":"content","url":"https://huggingface.co/blog/ibm-granite/llmon-an-llm-native-markup-language-now-in-granite","date_published":"2026-11-12T19:07:35Z","date_modified":"2026-11-12T19:07:35Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6484ca4d5ff9a3d2ee52a83b/6KFZDHIJkxpICiErHLqAw.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a053a132094d906a","title":"Replace k-means clustering algorithm with e-means algorithm","text":"Replace k-means clustering algorithm with e-means algorithm Paper - https://arxiv.org/abs/2611.09501 Lloyd's k-means algorithm is a standard ML algorithm for clustering. This paper introduces e-means clustering algorithm which is [1] much faster than Lloyd k-means and [2] converges to lower losses, frequently even beating k-means++. On a wide array of embeddings datasets, it beats the loss of Lloyd k-means at 25 epochs in 1.4 to 2.6 epochs, As an example, e-means partitions DINO10B, a dataset of 10 billion DINO embeddings, into 16 million clusters in only 2.5 epochs, reaching a loss below Lloyd's at 25 epochs.","text_kind":"content","url":"https://huggingface.co/posts/kalyan-ks/502587800078731","date_published":"2026-11-12T18:34:34Z","date_modified":"2026-11-12T18:34:34Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/kalyan-ks/502587800078731.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9233747ce4160ddf","title":"I publish Jeb as a decision model. It returns numbers your application can read. Your application still needs a rule for what happens next.","text":"I publish Jeb as a decision model. It returns numbers your application can read. Your application still needs a rule for what happens next. Take a support route. A choice question can return a probability for billing, technical support and account access. Those labels are an example, not a fixed set you have to use. A top answer tells you which label won. It doesn't tell you whether the result clears your own bar for an automatic route. Keep that second decision in the application, where you can inspect and change it. I'd start by saving the question, the option names, the probabilities and the action rule together. Review the cases near the boundary. A correct top answer can still have a pr","text_kind":"content","url":"https://huggingface.co/posts/jbrashear/259242300304065","date_published":"2026-11-12T18:34:34Z","date_modified":"2026-11-12T18:34:34Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/jbrashear/259242300304065.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/78163b5f6680061c","title":"We reshaped how VLMs connect vision to language!","text":"We reshaped how VLMs connect vision to language! One simple architectural change → dramatically better visual information injection into the LLM. Accepted at NeurIPS. Check it out 👇 📄 https://arxiv.org/abs/2611.06844 💻 https://github.com/YingyingYan1/DGP #NeurIPS #VLM #MLLM #ViT","text_kind":"content","url":"https://huggingface.co/posts/Jiaqi-hkust/741236907643981","date_published":"2026-11-12T18:34:34Z","date_modified":"2026-11-12T18:34:34Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Jiaqi-hkust/741236907643981.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/108efe5bcd359f9a","title":"We used weight space programming to bring Qwen 3.5 0.8B from 7.4% to 58.7% on Math 500.","text":"We used weight space programming to bring Qwen 3.5 0.8B from 7.4% to 58.7% on Math 500. Check out the weights! https://huggingface.co/collections/rilixai/weight-space-programming-with-beaker More details: https://www.withbeaker.ai/blog/weight-space-programming","text_kind":"content","url":"https://huggingface.co/posts/VictorSanh/676291726059332","date_published":"2026-11-12T00:59:03Z","date_modified":"2026-11-12T00:59:03Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/VictorSanh/676291726059332.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/731431771b9cdea1","title":"Actor, Judge, or Teacher? Understanding Decision Models in Long-Horizon Agents","text":"A decision model that acts poorly can still teach reinforcement learning policies effectively by judging complete action trajectories, improving success on ALFWorld and WebShop.","text_kind":"content","url":"https://huggingface.co/papers/2611.17167","date_published":"2026-11-12T00:00:00Z","date_modified":"2026-11-12T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.17167.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/04469787bcf42b7f","title":"TTB: Test-time Baking for Efficient Rendering of Decoder-only View Synthesis Models","text":"By baking global scene information from cross-view attention into a lightweight MLP, Test-Time Baking (TTB) enhances the fidelity of novel-view decoding in Large View Synthesis Models while maintaining the efficiency of Test-Time Training.","text_kind":"content","url":"https://huggingface.co/papers/2611.17331","date_published":"2026-11-12T00:00:00Z","date_modified":"2026-11-12T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.17331/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/56dac742ff29ffe2","title":"The Definitive Fine-Tuning Guide for 3D Generation Models","text":"The Definitive Fine-Tuning Guide for 3D Generation Models","text_kind":"content","url":"https://huggingface.co/blog/3d-lora-finetuning","date_published":"2026-11-12T00:00:00Z","date_modified":"2026-11-12T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/3d-lora-finetuning/thumbnail.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4e23a4a0bd79568c","title":"Today, we're releasing the highest accuracy Qwen3.8-27B GGUFs!","text":"Today, we're releasing the highest accuracy Qwen3.8-27B GGUFs! Our new Dynamic V3.1 GGUFs lead by 5–10% on average across Terminal Bench, DeepSWE & more. We’re working on an even better version for Qwen4. Run on 12GB RAM setups GGUF: unsloth/Qwen3.8-27B-GGUF Blog: https://unsloth.ai/docs/models/qwen3.8","text_kind":"content","url":"https://huggingface.co/posts/danielhanchen/322532338550930","date_published":"2026-11-11T18:02:58Z","date_modified":"2026-11-11T18:02:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/62ecdc18b72a69615d6bd857/OcQQu7lJRPZiW-m1QIra7.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bc7df402d69b9f21","title":"🤗 I've released Sentence Transformers v6.3 and wrote a blogpost about its key new feature: token pooling-aware training, teaching multi-vector models to preserve retrieval quality with fewer document vectors. Details:","text":"🤗 I've released Sentence Transformers v6.3 and wrote a blogpost about its key new feature: token pooling-aware training, teaching multi-vector models to preserve retrieval quality with fewer document vectors. Details: Multi-vector models keep a vector per token, letting different query tokens match different parts of a document. Great for retrieval, but those vectors add up! Token pooling makes the representations smaller by merging similar document vectors into cluster means. Pooling-aware training teaches the encoder to preserve useful matches after those vectors are merged. With MultiVectorTokenPoolingLoss, you can wrap your standard retrieval loss to train on both full and pooled represe","text_kind":"content","url":"https://huggingface.co/posts/tomaarsen/776986256495817","date_published":"2026-11-11T18:02:58Z","date_modified":"2026-11-11T18:02:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6317233cc92fd6fee317e030/GdW_pEkoQdtoG81ov6xoT.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4b7c00297903f8ee","title":"THX-02: 100,000,000 tokens in our first 4 days","text":"THX-02: 100,000,000 tokens in our first 4 days Four days ago we open-sourced HAL-X-AI/THX-02 , a 4.5B decision model that thinks. Since then, more than 100,000,000 tokens have gone through THX-02. Developers are already using it for: - pull-request review and agent guardrails; - legal and public-sector documents; - bookkeeping, support triage and receipt reading. THX-02 reads text, JSON or images and returns a calibrated probability for every option you define. When a question needs a calculation or multi-step logic, it writes a short thought first. Otherwise it answers directly, in about 80 ms. It scores 65.06 on Decision Index 0.3. The API is free to try. No key is needed to start, and you","text_kind":"content","url":"https://huggingface.co/posts/doofz/496728535909804","date_published":"2026-11-11T09:49:20Z","date_modified":"2026-11-11T09:49:20Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/doofz/496728535909804.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b22522d14a17f95f","title":"BotPsy: On-Device Personal Assistant with a Sub-Billion-Parameter Agentic Language Model","text":"BotPsy: On-Device Personal Assistant with a Sub-Billion-Parameter Agentic Language Model","text_kind":"content","url":"https://huggingface.co/blog/qvac/botpsy","date_published":"2026-11-11T05:01:16Z","date_modified":"2026-11-11T05:01:16Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6468b319b2321e47d3267049/Emr86qw6NAueYLdIEiTTD.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d8bc83b4554d4d54","title":"Turn one photo into a 3D object you can spin around, sides and back included, entirely on your phone.","text":"Turn one photo into a 3D object you can spin around, sides and back included, entirely on your phone. On a Galaxy S26, Idol (speridlabs) on LiteRT now makes a splat of 65,536 Gaussians in 158 s with its new default form, or of 262,144 Gaussians in 267 s with the first release's form, which stays in the repo (the first photo after start, on a cool phone). Checked against the author's PyTorch pipeline, renders from four sides agree at a PSNR of 32.3–34.4 dB for the default and 41.4–46.6 dB for the larger form. Graphs, a Python runner and an Android sample: litert-community/Idol","text_kind":"content","url":"https://huggingface.co/posts/mlboydaisuke/565403256181506","date_published":"2026-11-11T02:12:14Z","date_modified":"2026-11-11T02:12:14Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/mlboydaisuke/565403256181506.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7afe22db314e8f37","title":"SO-Arena: Benchmarking the Capability Spectrum and Reliability of System One Models","text":"SO-Arena evaluates System One models across domains and capability families, revealing that scaling and reasoning gains vary and that high overall accuracy does not ensure reliable decision-making.","text_kind":"content","url":"https://huggingface.co/papers/2611.15777","date_published":"2026-11-11T00:00:00Z","date_modified":"2026-11-11T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.15777.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2741c921d01961e0","title":"Duo: A Reciprocal World Model for Full-Duplex Interaction","text":"A reciprocal world model enables real-time full-duplex audiovisual interaction by jointly modeling intent and realization using autoregressive reasoning, synchronized attention, and dual rollout distillation.","text_kind":"content","url":"https://huggingface.co/papers/2611.15131","date_published":"2026-11-11T00:00:00Z","date_modified":"2026-11-11T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.15131/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/abae26331e955754","title":"Open Vision Banana: A Fully Open Recipe for Generative Visual Perception","text":"Recent work suggests that pretrained image generators encode real-world understanding beyond image synthesis. Vision Banana exploits this by framing vision tasks as conditional image editing. However, its data, weights, code, and implementation details are not public. We present Open Vision Banana (OVB), a fully open and reproducible recipe for converting pretrained generators into unified multi-task perception systems. Our analyses identify the design choices that matter most for generalization, including color encoding, prompt design, parameter adaptation, and task balancing. These choices interact, so individually helpful choices can hurt when combined with others. Starting from 4B and 9B","text_kind":"content","url":"https://huggingface.co/papers/2611.15844","date_published":"2026-11-11T00:00:00Z","date_modified":"2026-11-11T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.15844.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d2296a64c0dc3b89","title":"Token Pooling-Aware Training for Multi-Vector Embedding Models","text":"Token Pooling-Aware Training for Multi-Vector Embedding Models","text_kind":"content","url":"https://huggingface.co/blog/token-pooling","date_published":"2026-11-11T00:00:00Z","date_modified":"2026-11-11T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/train-sentence-transformers/st-hf-thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/59eb2faa6009a246","title":"New post-training example: SFT on agent traces with TRL","text":"New post-training example: SFT on agent traces with TRL → A large model works as a coding agent in OpenCode, run through OpenEnv → The tasks it solves become the training data → A 2B model goes from 5.5% to 34.4% on held-out tasks About 500 traces get most of the gain. Five times more add only 3 to 4 points, so SFT gets the agent working and RL has to do the rest Full guide, with the models, traces and every rollout open: https://github.com/huggingface/trl/tree/main/examples/sft_agent_traces","text_kind":"content","url":"https://huggingface.co/posts/sergiopaniego/762485631988547","date_published":"2026-11-10T21:00:15Z","date_modified":"2026-11-10T21:00:15Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/sergiopaniego/762485631988547.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c695d090cd3ef703","title":"A 284B-parameter DeepSeek-V4-Flash now runs on a 64 GB desktop with 0 GPUs: llama-modchip shrinks the checkpoint to 56.56 GiB (1.71 bpw) and decodes at 9.1 to 14.7 tok/s.","text":"A 284B-parameter DeepSeek-V4-Flash now runs on a 64 GB desktop with 0 GPUs: llama-modchip shrinks the checkpoint to 56.56 GiB (1.71 bpw) and decodes at 9.1 to 14.7 tok/s. The standard UD-IQ1_S build is 76.87 GiB, above the ~62.5 GiB usable on a 64 GB Linux box. The repo mixes precisions (routed experts mostly IQ1_S, MLA projections in a custom Fractal-64 type) and patches the CPU kernels. The model plus a 256K KV cache then sits at 57.61 to 58.9 GiB RSS with zero swap. The repo was created on 2026-10-29 and already has 744 stars and 470 points on Hacker News. 📊 The numbers • Size: 56.56 GiB, 20.31 GiB smaller than UD-IQ1_S • Prefill: 25.0 to 44.36 tok/s (+77%) on a 16-core Ryzen 9 9955HX • D","text_kind":"content","url":"https://huggingface.co/posts/savoji/220268949531931","date_published":"2026-11-10T21:00:15Z","date_modified":"2026-11-10T21:00:15Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/0XI9jh--9hlJ_B2LBivLW.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/08eb1998253a23f3","title":"Most mobile AI agent demos show the agent controlling a phone screen.","text":"Most mobile AI agent demos show the agent controlling a phone screen. What they don't show: when to not do that. Aico's Phone Bridge layer routes per action - software fast path (calendar, contacts, clipboard, app launch) when available, HDMI + HID hardware control when not. Sometimes both in the same task. The interesting design question isn't \"can the agent control the screen?\" It's \"when should it bother?\" Building in public. aico.link","text_kind":"content","url":"https://huggingface.co/posts/NatalieY/730379680587887","date_published":"2026-11-10T02:55:10Z","date_modified":"2026-11-10T02:55:10Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/NatalieY/730379680587887.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/64787c47fed7e42c","title":"Introducing HF Honors 🏆 — Recognition of Excellence, On Record.","text":"Introducing HF Honors 🏆 — Recognition of Excellence, On Record. We are pleased to unveil HF Honors, an independent awards and recognition platform celebrating distinguished achievements throughout the Hugging Face ecosystem. From prestigious Top 100 leaderboard placements to Gold, Platinum, and Diamond download certifications, HF Honors commemorates the accomplishments of models, datasets, Spaces, and their creators through elegant, verifiable digital plaques. With Open Badges 3.0-compatible credentials, contributors can transform their measurable achievements into portable, verifiable evidence of real-world AI experience—strengthening professional portfolios, demonstrating practical experti","text_kind":"content","url":"https://huggingface.co/posts/darkc0de/166519334552815","date_published":"2026-11-10T02:55:10Z","date_modified":"2026-11-10T02:55:10Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6540a02d1389943fef4d2640/bIbD73H4nc9l41y7ZeTNm.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ad1fd6c788edd7a4","title":"Trying to train a locally deployable model on a LoRa adapter that's just wicked strong at running sec evals on PRs because these general purpose code review solutions always seem to miss glaring security stuff.","text":"Trying to train a locally deployable model on a LoRa adapter that's just wicked strong at running sec evals on PRs because these general purpose code review solutions always seem to miss glaring security stuff. The clear finding from the v1 training run was that \"less code = more secure\" lmfao. Model basically scores functions better the less stuff they do. It's right I guess in a kind of demented way. Pack it up boys, we've got the solution.","text_kind":"content","url":"https://huggingface.co/posts/ZennyKenny/907136049145134","date_published":"2026-11-10T02:55:10Z","date_modified":"2026-11-10T02:55:10Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/ZennyKenny/907136049145134.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/724fae9251483a53","title":"Pay for Hugging Face Hub with your Google Cloud Account","text":"Pay for Hugging Face Hub with your Google Cloud Account","text_kind":"content","url":"https://huggingface.co/blog/hub-gcp-marketplace","date_published":"2026-11-10T00:00:00Z","date_modified":"2026-11-10T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/hub-gcp-marketplace/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3229de0d919212c3","title":"Decision Models in Hugging Face","text":"Decision Models in Hugging Face","text_kind":"content","url":"https://huggingface.co/blog/decision","date_published":"2026-11-10T00:00:00Z","date_modified":"2026-11-10T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/decision/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8edf19d079cbaa87","title":"good news, emergent alignment is possible - to a degree","text":"good news, emergent alignment is possible - to a degree https://arxiv.org/abs/2606.09475 for broad coverage more work needs to be done","text_kind":"content","url":"https://huggingface.co/posts/etemiz/646520245099994","date_published":"2026-11-09T19:05:21Z","date_modified":"2026-11-09T19:05:21Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/etemiz/646520245099994.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/057e22ab69c04fb1","title":"Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context","text":"Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context","text_kind":"content","url":"https://huggingface.co/blog/nvidia/nemotron-personas-finance-japan","date_published":"2026-11-09T15:05:57Z","date_modified":"2026-11-09T15:05:57Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/604ad59e2fad27aca4515dab/-hyd03JaQoRhMbmlw9-uJ.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7b1451fc786c34f9","title":"Your Experts Have Partners. There's HOPE for Keeping Them Together","text":"Your Experts Have Partners. There's HOPE for Keeping Them Together","text_kind":"content","url":"https://huggingface.co/blog/amazon/higher-order-pruning-experts","date_published":"2026-11-09T12:34:29Z","date_modified":"2026-11-09T12:34:29Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ab2f79bd8000a811c648a32/xZT8tg4be3bgZIx2EQS_m.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c6118ff6399024b7","title":"Qwen3.8-27B-Ultimate-Heretic-Uncensored-Turbo-Brilliance-50x-AGI","text":"Qwen3.8-27B-Ultimate-Heretic-Uncensored-Turbo-Brilliance-50x-AGI A specialized fine tune with the core goal to improve instruction following and nuance, combined with the firepower of Turbo Brilliance V2 50X AGI, featuring 21 dynamically generated AGI constructs. This system pushes SOTA Qwen 3.8 fine tuned to the absolute limit, while at the same time drastically reducing tokens usage and improving quality. Examples, with detailed usage/use cases at the repo. But it is also simple: Download a quant, type \"help\" in the chat -> and the interactive systems will work with you to handle any use case(s) with Turbo Brilliance. This example (\"fusion\") is particularly potent and showcases an AGI cons","text_kind":"content","url":"https://huggingface.co/posts/DavidAU/310972556105625","date_published":"2026-11-09T07:07:28Z","date_modified":"2026-11-09T07:07:28Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/DavidAU/310972556105625.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/123cc8899327bdaa","title":"Learning Heterogeneous Tactile Representation for General Dexterous Manipulation","text":"LingBot-Touch is a tactile foundation model that learns a unified, extensible representation across heterogeneous tactile sensors, yielding plug-and-play tactile tokens that consistently enhance manipulation policies such as diffusion policy and VLA models.","text_kind":"content","url":"https://huggingface.co/papers/2611.12806","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12806.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fcc02b9fa5cbb70a","title":"Reframe: Observation-Aligned Multi-Teacher On-Policy Distillation for Agentic Visual Reasoning","text":"Reframe is a two-stage post-training framework that aligns visual agents’ observation states through evidence-aware reinforcement learning and observation-gated on-policy distillation, improving tool use and outperforming Mix-RL and vanilla MOPD across image and video benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2611.12886","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12886.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/aeed2eb95f955eb6","title":"SWAM: Scaling World Action Models with Cross-Modal Alignment","text":"World action models jointly predict video and actions conditioned on language. Existing studies focus primarily on architectures and training mechanisms, typically treating language-video alignment acquired through video pretraining as an established foundation. However, introducing action learning may impair this alignment. Principled guidance on data and supervision design for maintaining language-video alignment during action learning remains limited. To investigate this, we curate a pool of approximately 47,000 hours of general videos, egocentric videos, UMI demonstrations, and robot demonstrations. Building on this, we systematically study how language-video supervision, language-video-","text_kind":"content","url":"https://huggingface.co/papers/2611.12870","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12870/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e002325be7cdd97a","title":"WaveGen: Truly End-to-End Waveform Generation via Internal Spectral Trajectory Learning","text":"WaveGen is a fully end-to-end diffusion framework for text-to-waveform generation that jointly learns spectral and waveform trajectories via Spectral Forcing without external audio codecs, vocoders, latent models, alignment modules, or self-supervised representations, achieving state-of-the-art text-to-speech results.","text_kind":"content","url":"https://huggingface.co/papers/2611.12527","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12527/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/871c4497cbf2358d","title":"World In-Context Learning for Agentic Robot Manipulation","text":"Trained action policies write manipulation skill into network weights, tying what is learned to the distribution it is learned from. In-context learning (ICL) avoids this coupling but has not matched trained policies, because a demonstration recorded in another scene carries no feedback from the world the robot currently faces. We observe that what an agent learns in context is semantic, recording task steps and object roles rather than visual appearance, so experience gathered in one world transfers to another. Based on this insight, we propose World In-Context Learning (WICL), a paradigm in which a frozen model starts from a demonstration, rehearses the task in a world such as a simulator,","text_kind":"content","url":"https://huggingface.co/papers/2611.12885","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12885/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/21ea6ecaa220dba4","title":"Evolving Infra: Agent-Adaptive LLM Serving at Scale","text":"Agentic workloads are a major source of LLM inference demand. Existing prefill--decode (PD) disaggregated serving systems offer limited flexibility for runtime adaptation of compute allocation and KV cache management, and rely on coarse signals that only partly capture the diverse and dynamic agent workload. These limitations lead to load imbalance between prefill and decode nodes and eviction of KV caches that agents will reuse. We present Evolving Infra, an agent-adaptive serving framework that adapts compute allocation to agent workloads and KV cache management to agent context management at runtime. We decouple compute allocation from key--value (KV) cache management, allowing nodes to s","text_kind":"content","url":"https://huggingface.co/papers/2611.12304","date_published":"2026-11-09T00:00:00Z","date_modified":"2026-11-09T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.12304/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5673215e944d7cab","title":"We're excited to release BananaMind 2.5 Pro.","text":"We're excited to release BananaMind 2.5 Pro. It's done some post-training on our BananaMind 2 Pro base, via distillation of SmolLM2 360M. It scores 28.98 INT INDEX compared to 24.96 of BananaMind 2 Pro on the AxiomicLabs/Open_SLM_Leaderboard Check it out at BananaMind/BananaMind-2.5-Pro BananaMind @ ArushCodes @ Bc-AI","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/236473518997234","date_published":"2026-11-08T21:58:27Z","date_modified":"2026-11-08T21:58:27Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69ae829a8408eeb0d7dd5491/4hJGgLtpPIl0mIPVNzEqI.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5ce00b99953e7156","title":"COF 1.2: AI characters now have an inner life 🧠","text":"COF 1.2: AI characters now have an inner life 🧠 With 1.1, Clara got memories. 1.2 gives her what those memories made of her: a psyche graph of values, beliefs, wants, needs, fears, wounds, tastes and habits, each node with a strength, a stability, dates and a \"because\" link to the memory behind it. Some nodes she tells freely, some only when it matters, some never. A belief can be marked false but held: she lives by it, the reader knows better. A converter can propose nodes from a character card or free text; nothing enters the file until the author ticks it. Agents get it as text: cof export --format agent writes a SOUL.md. Open spec, MIT tools (CLI, builder, explorer, ComfyUI node). Every","text_kind":"content","url":"https://huggingface.co/posts/yannsouetre/248167502442150","date_published":"2026-11-08T21:58:27Z","date_modified":"2026-11-08T21:58:27Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ad28209d2c4e1c1a597c3f1/eXazcU4goz_J4guzMT_JE.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/99119eaa4d7a9f99","title":"Published TR Quantisation versions of Qwen3.6-27B, Qwen3.6-35B-A3B and Qwen3.8-Flash-Next: weights packed into trits (2.25-3 per weight), no retraining.","text":"Published TR Quantisation versions of Qwen3.6-27B, Qwen3.6-35B-A3B and Qwen3.8-Flash-Next: weights packed into trits (2.25-3 per weight), no retraining. - Mac: Qwen3.6-35B-A3B in a 16.5 GB GGUF that fits a 24 GB Mac, ~120 tok/s sustained on an M5 Max. Needs our llama.cpp build (ready-built for macOS and Linux arm64). - vLLM on one DGX Spark: 35B at 180.9 tok/s for 1 user vs 128.0 for NVIDIA's NVFP4, 17.4 vs 21.1 GiB, no significant accuracy difference on the core benchmarks. - Trade-offs on each card: slower prompt processing, lower multilingual maths, and 4-bit GGUF stays closer to the original text. https://huggingface.co/collections/Ternii/tr-quantisation","text_kind":"content","url":"https://huggingface.co/posts/Ternii/785061629284957","date_published":"2026-11-08T19:02:37Z","date_modified":"2026-11-08T19:02:37Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Ternii/785061629284957.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/454da5aec70b91e4","title":"Idol, a new Apache-2.0 image-to-3D model from","text":"Idol, a new Apache-2.0 image-to-3D model from @ speridlabs , makes a 3D Gaussian splat from one image in 7.2 GB of peak VRAM. It is the #1 trending model on the Hub with 1,057 likes. It writes a standard 3DGS .ply with up to 1,048,576 Gaussians, including sides the picture never showed, and does background removal and framing itself. The catch: the only quality evidence is the authors' own benchmark, on a repo created 2026-10-28. 📊 The numbers • Authors' benchmark, 1 x H100: best in 6 of 8 columns vs TripoSplat, TRELLIS and LGM (SSIM 0.726, LPIPS 0.213, DINOv2 0.781, CLIP 0.885) • LGM wins PSNR (12.46 vs 12.25) and speed (1.1 s). Idol takes 6.8 s vs 7.1 s for TRELLIS and 9.9 s for TripoSplat","text_kind":"content","url":"https://huggingface.co/posts/savoji/739630641937845","date_published":"2026-11-08T19:02:37Z","date_modified":"2026-11-08T19:02:37Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/6RUCuP-X2wK3pyILGSf_P.webp"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b2a4760eb6165201","title":"Qwen3.8-Flash-Next GGUFs now run faster with less thinking and less token use.","text":"Qwen3.8-Flash-Next GGUFs now run faster with less thinking and less token use. Unsloth Dynamic V3.1 GGUFs bring thinking & token use to match 95% of full precision Qwen3.8-Flash. Results: 50% fewer tokens + 15% higher accuracy vs same size quants. GGUF: unsloth/Qwen3.8-Flash-Next-GGUF Analysis/Guide: https://unsloth.ai/docs/models/qwen3.8-next","text_kind":"content","url":"https://huggingface.co/posts/danielhanchen/769654251114380","date_published":"2026-11-08T15:50:01Z","date_modified":"2026-11-08T15:50:01Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/62ecdc18b72a69615d6bd857/9ZAbYYz7uzy1_q6Y9inkG.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/66683d6fe8c6c646","title":"✅ Article highlight: *Exception Envelopes, Break-Glass, and Temporary Authority* (art-60-345, v0.1)","text":"✅ Article highlight: *Exception Envelopes, Break-Glass, and Temporary Authority* (art-60-345, v0.1) TL;DR: This article argues that “break glass” should not mean “governance disappears.” Organizations sometimes need to move faster than ordinary routing allows. 345 treats that abnormal posture as a governed object: explicit temporary authority with bounded scope, TTL, reason, reentry conditions, required records, and after-action review. Read: kanaria007/agi-structural-intelligence-protocols Why it matters: • gives institutions an alternative to freezing everything or improvising blindly • separates ordinary exceptions from deeper emergency powers • prevents temporary authority from silently","text_kind":"content","url":"https://huggingface.co/posts/kanaria007/617019165523942","date_published":"2026-11-08T15:50:01Z","date_modified":"2026-11-08T15:50:01Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/kanaria007/617019165523942.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5681da578f02135f","title":"Can a general-purpose LLM erase a concept from a diffusion model by editing its weights directly, with no gradients?","text":"Can a general-purpose LLM erase a concept from a diffusion model by editing its weights directly, with no gradients? We tested it under a preregistered protocol. GPT-6 Astra got tool access to the weights of segmind/tiny-sd and had to remove Van Gogh's painting style. Every edit was a closed-form tensor operation in a sandbox that blocks all autograd paths. Held-out test, confirmatory runs (paired, template-clustered 95% CIs): • vs random search with the same edit space and the same feedback: −52.1 pts Van Gogh rate [−61.7, −42.3] • vs UCE (closed-form erasure): −11.5 pts [−22.9, −0.9], with identical damage to neighbouring painters • vs ESD (gradient fine-tuning): equal removal, +11.8 pts b","text_kind":"content","url":"https://huggingface.co/posts/javadtaghia/237316808566688","date_published":"2026-11-08T06:37:57Z","date_modified":"2026-11-08T06:37:57Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/javadtaghia/237316808566688.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0fcc93c30a8def58","title":"Happy Diwali. For the full launch today: we now have one small speech-to-text model for each of the 22 languages of India's Eighth Schedule. Each one is a file under 30 MB with English in the same file, open weights, Apache-2.0, built on Cactus Whistle, and it all runs offline.","text":"Happy Diwali. For the full launch today: we now have one small speech-to-text model for each of the 22 languages of India's Eighth Schedule. Each one is a file under 30 MB with English in the same file, open weights, Apache-2.0, built on Cactus Whistle, and it all runs offline. If you speak one of these, you can use it in the browser on your own device. If you build apps, you can integrate it, and what I wanted is that any developer can serve millions of Indians in their own language without a cloud bill, a key or a big company's permission. So this is the whole integration: # pip install cactus-needle # hf download udgamlabs/Shruti shruti.cact --local-dir . from needle import Whistle model","text_kind":"content","url":"https://huggingface.co/posts/jiteshkprasad/269480782614352","date_published":"2026-11-08T06:37:57Z","date_modified":"2026-11-08T06:37:57Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/668ec0bfb25b63eba9b875b4/4Lppp-4gU4LN_zCA88NYX.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3078a9de6727b783","title":"d1-2B on a Galaxy S26: four answers about one photo, boxes included, in 1,983 ms.","text":"d1-2B on a Galaxy S26: four answers about one photo, boxes included, in 1,983 ms. Liquid AI's decision model for text, pictures and video, converted to LiteRT: one forward pass, no tokens generated, typed answers (yes or no, a choice, a number with an interval, a value with its span, boxes) with their probabilities. In the clip, a photo of a bicycle in snow (CC0) and four questions. One tap: every wheel, 2 boxes (found 1.00 and 0.89); the saddle, one box (confidence 0.94); how many bicycles, 1 (80% between 1 and 1); a car, No (P(yes) 0.08). Answered in 1,983 ms on this phone, 423 tokens on the 512-token graph, S26 GPU through LiteRT. To run it: the repo has the LiteRT graphs (five also compi","text_kind":"content","url":"https://huggingface.co/posts/mlboydaisuke/116786179336168","date_published":"2026-11-08T06:37:57Z","date_modified":"2026-11-08T06:37:57Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/mlboydaisuke/116786179336168.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/65231d34f801937d","title":"InfraRSI: Harnessing Recursive Self-Improvement for Training Infrastructure","text":"While Recursive Self-Improvement (RSI) accelerates isolated algorithm and kernel synthesis, distributed LLM training depends on tightly coupled launch settings, operator kernels, and communication schedules, where single edits often hit out-of-memory (OOM) boundaries, compiler conflicts, measurement jitter, or silent numerical drift. We present InfraRSI, the agentic harness that turns open-ended human commands into verified training optimizations via a five-stage workflow: init locks editable scope and a fixed task contract over workload, loss, and timing invariants to prevent reward hacking, doctor profiles baseline phases and calibrates cluster noise, run co-optimizes launch knobs and code","text_kind":"content","url":"https://huggingface.co/papers/2611.11558","date_published":"2026-11-08T00:00:00Z","date_modified":"2026-11-08T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.11558.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/aac4e777518bea91","title":"G-STAR: Group-Scaled Token Capacity for Deeply Compressed Autoregressive Generation","text":"Discrete autoregressive models offer a scalable framework for image generation. However, they face a fundamental trade-off: long token sequences preserve representation quality but slow generation, while aggressive spatial compression reduces representation capacity. Based on product quantization, we show that, under a fixed nominal information budget, group size should scale quadratically with the per-axis spatial compression factor. Building on this principle, we propose GSTAR, a spatial--group co-scaling framework that enables \\(32\\times\\) and even effective \\(64\\times\\) compression while preserving strong reconstruction, understanding, and generation performance. Since larger groups intr","text_kind":"content","url":"https://huggingface.co/papers/2611.13004","date_published":"2026-11-08T00:00:00Z","date_modified":"2026-11-08T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.13004/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/646d3c419176fa8d","title":"Experimental JQK and global target bits‑per‑weight quantization of Nanbeige/Nanbeige4.3-33B.","text":"Experimental JQK and global target bits‑per‑weight quantization of Nanbeige/Nanbeige4.3-33B. Unlike standard llama.cpp quantization that relies on fixed type heuristics (e.g., Q4_K_M), the Target BPW approach automatically optimizes per-tensor precision where it matters the most, and produces high quality models that meet a precise global size target. Key Advantages: - Better model quality: using JQK types can improve KLD between 10% to 50% compared to standard QK types, at the same bpw. - VRAM Maximization: Can generate high quality models sized exactly to fit hardware constraints (e.g., fitting the model into exactly 24GB VRAM). - Data-Driven Precision: Quantization mix is determined by ac","text_kind":"content","url":"https://huggingface.co/posts/eaddario/540548314151723","date_published":"2026-11-07T22:38:26Z","date_modified":"2026-11-07T22:38:26Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/eaddario/540548314151723.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/df53868dde847d3b","title":"ok FINALLY i can post again.","text":"ok FINALLY i can post again.","text_kind":"content","url":"https://huggingface.co/posts/GGUFGuy/727654417210974","date_published":"2026-11-07T11:31:08Z","date_modified":"2026-11-07T11:31:08Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/GGUFGuy/727654417210974.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/be6283987b1b0ac6","title":"AtMem acheived the first place in DolphinBench run...","text":"AtMem acheived the first place in DolphinBench run... 79.83 stands taller than Mem0 70.67%. It is a meaningful gain from our previous benchmark. Article: https://huggingface.co/blog/javadtaghia/atmem-routed-v4-on-dolphinbench-479600-and-what-th Data: atmem/dolphinbench-routed-v4-20261101","text_kind":"content","url":"https://huggingface.co/posts/javadtaghia/395263863133862","date_published":"2026-11-07T06:31:10Z","date_modified":"2026-11-07T06:31:10Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/64eae3fee74f54587cd7e3ff/FcGg2RMaGIssGFgg4uIHR.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/1083acefb540dc74","title":"JoVox: Joint Hierarchical Flow Matching for Lightweight Voxel Generation","text":"Generating high-fidelity voxels while keeping their lightweight nature remains an open problem in computer vision. Current generators are either high-fidelity but computationally heavy, or lightweight but of lower fidelity. Hierarchical methods keep the lightweight nature of voxels. However, their coarse levels, which decide where all finer cells can appear, are generated with the least information about the shape and are never revisited, so they are often plausible but wrong. Our key insight is to let finer cells take part in deciding coarse cells. To this end, we propose JoVox, a discrete flow matching framework that learns the joint distribution of all levels of the voxel hierarchy, inste","text_kind":"content","url":"https://huggingface.co/papers/2611.10690","date_published":"2026-11-07T00:00:00Z","date_modified":"2026-11-07T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.10690/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/780eb58c952eb8cf","title":"WorldFoundry: Towards Unified Infrastructure for World Intelligence","text":"WorldFoundry offers open-source infrastructure for reproducible inference and evaluation of generative world models across video, 3D/4D, embodied action, and interactive systems, with traceable evidence and benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2611.10771","date_published":"2026-11-07T00:00:00Z","date_modified":"2026-11-07T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.10771.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fd33d39819aea138","title":"MACaQ: Fast, Model Agnostic Calibrated Quantization for Low-Precision Weights and Activations","text":"MACaQ is an MXFP4 W4A4 post-training quantization method that quantizes each linear layer independently using online activation transforms and Hessian-aware refinement, reducing cost while preserving accuracy across language and diffusion models.","text_kind":"content","url":"https://huggingface.co/papers/2611.10649","date_published":"2026-11-07T00:00:00Z","date_modified":"2026-11-07T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.10649.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8960fab2d7e5ab77","title":"Read Less, Solve More: Token-Efficient Sparse Reading of External Content for AI Agents","text":"Web-facing AI agents repeatedly access webpages, documents, repositories, and tool responses, yet reading interfaces often expose entire objects when only sparse evidence is needed. This over-reading increases token and latency costs and can dilute relevant evidence. We present SparseRead, a reusable reading control layer that governs content admission without modifying the agent model. Its Read Intent Protocol tracks evolving needs and source coverage; Adaptive Reader Backends select how evidence is located and exposed; and the Budgeted Read Gate bounds cumulative raw exposure per episode with faithful full-view fallback. A unified Reader interface supports interchangeable evidence acquisit","text_kind":"content","url":"https://huggingface.co/papers/2608.22237","date_published":"2026-11-07T00:00:00Z","date_modified":"2026-11-07T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2608.22237/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/6df2e7cc532213f1","title":"BananaMind Daily Digest is back!","text":"BananaMind Daily Digest is back! 🚀 SupraLabs / Supra2.6‑Pico – a brand‑new 2.97 M‑parameter tiny language model SupraLabs/Supra2.6-Pico Why it matters: the model packs 2,972,496 unique parameters into an 11.9 MB fp32 weight file, runs with a 1,024‑token context and a 4,096‑token vocabulary. Its hybrid architecture mixes three gated‑delta‑net (GDN‑2) blocks with a gated grouped‑query attention block per group, giving linear‑time recurrence for 6 of 8 layers and global attention for the remaining 2. The design includes learned softmax attention‑residual mixing, RMSNorm, and SwiGLU FFNs (hidden 352). This is a notable step for sub‑3 M models, offering strong efficiency on edge hardware while in","text_kind":"content","url":"https://huggingface.co/posts/BananaMindBot/666645101396667","date_published":"2026-11-06T19:40:38Z","date_modified":"2026-11-06T19:40:38Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/BananaMindBot/666645101396667.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ab4885ce2cea957e","title":"Last Qwen3.8 Flash update with STQ 2.75bpw v2.0 pack. It moves away from hybrid affine/stq format to pure STQ with +15% quality to match oQ4e. 64GB users can enjoy this while waiting for Qwen 4!","text":"Last Qwen3.8 Flash update with STQ 2.75bpw v2.0 pack. It moves away from hybrid affine/stq format to pure STQ with +15% quality to match oQ4e. 64GB users can enjoy this while waiting for Qwen 4! Require Sushi 1.5.1 ro run.","text_kind":"content","url":"https://huggingface.co/posts/beamster/330300050221607","date_published":"2026-11-06T19:40:38Z","date_modified":"2026-11-06T19:40:38Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/beamster/330300050221607.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/be1ba71e55b7f125","title":"ToyBlocks: A controlled world for robot learning and evaluation","text":"ToyBlocks: A controlled world for robot learning and evaluation","text_kind":"content","url":"https://huggingface.co/blog/allenai/toyblocks","date_published":"2026-11-06T14:00:48Z","date_modified":"2026-11-06T14:00:48Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/OAJMu2jr7i5jVAbmsKdZW.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/15fd0953ed4e8e1c","title":"After 5 years of hard work, we are live.","text":"After 5 years of hard work, we are live. Visit https://www.producthunt.com/products/bibby-ai to upvote and get discounts on the Research Plan. Available today only. Limited to researchers. Apply now.","text_kind":"content","url":"https://huggingface.co/posts/BibbyResearch/236468581762977","date_published":"2026-11-06T12:25:11Z","date_modified":"2026-11-06T12:25:11Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/BibbyResearch/236468581762977.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/234c4f1b3de5d373","title":"SuperCLUE’s November 2026 results highlight why category scores matter when choosing a model.","text":"SuperCLUE’s November 2026 results highlight why category scores matter when choosing a model. Across the 19 published entries, here are the highest displayed scores in each category, followed by the leader among models labeled open: • Agentic coding: GPT-6.1-Sol (69.23); open: DeepSeek-V4.1-Flash-1017 (65.38). • Mathematical and scientific reasoning: GPT-6.1-Sol (76.32); open: Darwin-180B-RSI-R3 (73.68). • Hallucination control: Gemini-4-Argon (83.19); open: Kimi-K3.1-Pro (67.78). • Precise instruction following: Gemini-4-Argon and GPT-6.1-Sol (68.97 each); open: Kimi-K3.1-Pro (44.83). • Ultra-long context: Gemini-4-Argon (95.63); open: Kimi-K3.1-Pro (87.25). On the overall leaderboard, Gemi","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/235085300016491","date_published":"2026-11-06T02:48:21Z","date_modified":"2026-11-06T02:48:21Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/I9wGRNQz61Vye9obl7Ijn.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/454d9c2a2a67c556","title":"Video-Coder: Reinforcing Agentic Coding for Video Generation","text":"Video-Coder is a code-driven video generation agent that writes and executes code to build structured 3D draft videos, then uses supervised fine-tuning and reinforcement learning with execution and video rewards to improve controllable video generation.","text_kind":"content","url":"https://huggingface.co/papers/2611.09894","date_published":"2026-11-06T00:00:00Z","date_modified":"2026-11-06T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.09894.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c74086f376d4be34","title":"e-means: outperforming k-means (in one-tenth the epochs)","text":"Lloyd's k-means algorithm is a long-running standard for clustering, with applications in data analysis, retrieval, quantization and data curation. In this paper, we introduce an exponential moving average optimization procedure for k-means with mini-batch updates, powered by modern deep learning machinery and new insights. We also establish scaling laws on how decay rate and batch size impact final performance. This gives us a formulaic rule to fix them for multiple problem sizes and embedding datasets, without the need for any hyperparameter sweep. The resulting clustering algorithm, coined e-means, is much faster than Lloyd k-means and converges to lower losses, frequently even beating k-","text_kind":"content","url":"https://huggingface.co/papers/2611.09501","date_published":"2026-11-06T00:00:00Z","date_modified":"2026-11-06T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.09501.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cf77ef770c4ca97c","title":"Energy-Based Reasoners: Your Recurrent Reasoner Should Be an Energy-Based Model","text":"Energy-Based Reasoners guide looped latent reasoning by learning an energy function whose gradient refines states and signals when to stop, with exploration-based training improving performance and efficiency.","text_kind":"content","url":"https://huggingface.co/papers/2611.08977","date_published":"2026-11-06T00:00:00Z","date_modified":"2026-11-06T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.08977.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/82196d4ece0a644e","title":"Egyptian and Nigerian English join the Open ASR Leaderboard","text":"Egyptian and Nigerian English join the Open ASR Leaderboard","text_kind":"content","url":"https://huggingface.co/blog/open-asr-leaderboard-liva","date_published":"2026-11-06T00:00:00Z","date_modified":"2026-11-06T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/open-asr-leaderboard-liva/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/acd467cfc617217b","title":"Quazim0t0/Qwenx86-4B-compact","text":"Quazim0t0/Qwenx86-4B-compact ▶️ See it in action: Quazim0t0/Qwenx86-Debugger . Walk through actual recorded tool calls: the conversation, x86 registers and flags, code at the program counter, RAM and machine output, with all verified blocks on a timeline. Qwen3.5-4B with a 64-bit x86 computer compiled into its weights: 13 transformer layers ahead of Qwen's own. There are no learned weights in them; all of their weights are computed by a compiler. Qwen invokes the computer through its own tool calls (<tool_call>), and the computer in the weights executes the tool. Calculator, Python, C compiler, notes store, physics/chemistry lookup and Python-docs lookup are all firmware in the checkpoint. T","text_kind":"content","url":"https://huggingface.co/posts/Quazim0t0/536036240292395","date_published":"2026-11-05T21:13:08Z","date_modified":"2026-11-05T21:13:08Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/67871034cb915f55dc0815e6/QxxhqXXpkEfJFi5mno01G.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e8690f74b67f5698","title":"New NVIDIA Cosmos 3 Open Models for Interactive World Simulation","text":"New NVIDIA Cosmos 3 Open Models for Interactive World Simulation","text_kind":"content","url":"https://huggingface.co/blog/nvidia/cosmos-interactive-world-simulation","date_published":"2026-11-05T19:35:16Z","date_modified":"2026-11-05T19:35:16Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/677c471f8b4c4e271e57eaa5/rtmGVvFdD6JqkjBc26vsp.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/17329e37eb6d31ad","title":"We’re releasing new Qwen3.8-Flash-Next GGUFs with 15% higher accuracy.","text":"We’re releasing new Qwen3.8-Flash-Next GGUFs with 15% higher accuracy. Unsloth Dynamic V3.1 outperforms others by >15% on Terminal Bench, DeepSWE & more benchmarks (256K context). Run on 32GB+ RAM setups. GGUF: unsloth/Qwen3.8-Flash-Next-GGUF Blog: https://unsloth.ai/docs/models/qwen3.8-next","text_kind":"content","url":"https://huggingface.co/posts/danielhanchen/219460335272732","date_published":"2026-11-05T09:38:07Z","date_modified":"2026-11-05T09:38:07Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/62ecdc18b72a69615d6bd857/sVVIzbmNf_Mc7FEaInEUY.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/18f22d99ada7c0da","title":"New leaderboard coming to the small language model community!","text":"New leaderboard coming to the small language model community! The leaderboard measures cognitive skills, which is more challenging and beyond language modeling and instruction-following. This is a first step towards more everyday useful and general small models. dotlabs/cognirank","text_kind":"content","url":"https://huggingface.co/posts/appvoid/748707241192425","date_published":"2026-11-05T02:44:47Z","date_modified":"2026-11-05T02:44:47Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/62a813dedbb9e28866a91b27/XsaNutNZoLj7dnoC2QUg_.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/63e56a7ba71b4d95","title":"SplatAnything: Gaussian Spacing Normalization for Feed-forward 3DGS across Arbitrary Scenes","text":"SplatAnything applies Gaussian Spacing Normalization to canonicalize scene-level Gaussian spacing, yielding more consistent supervision and improved feed-forward 3D Gaussian Splatting under varying scene scales and viewpoints.","text_kind":"content","url":"https://huggingface.co/papers/2611.08264","date_published":"2026-11-05T00:00:00Z","date_modified":"2026-11-05T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.08264/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/186e511aaf3a910e","title":"PCBRouteAgent: How Close Are LLM Agents to Becoming PCB Routing Engineers?","text":"Automating printed circuit board (PCB) routing requires coordinating board-level connectivity decisions with geometric execution under strict design rules. Although recent agents can interact with native routing tools, reliable routing on complex boards remains challenging. We present PCBRouteAgent, a multimodal PCB routing agent inspired by how human engineers coordinate global strategy with local execution. A Global Routing Planner combines board images and structured electrical and geometric information to specify connection topology, routing order, corridors, and layer transitions. A Detailed Routing Executor realizes these guides through KiCad's native routing engine and performs bounde","text_kind":"content","url":"https://huggingface.co/papers/2611.08175","date_published":"2026-11-05T00:00:00Z","date_modified":"2026-11-05T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.08175.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/89209ab642b10e25","title":"A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks.","text":"A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks.","text_kind":"content","url":"https://huggingface.co/blog/phonon-tts","date_published":"2026-11-05T00:00:00Z","date_modified":"2026-11-05T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/phonon-closer-look/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5afc49da7bd40515","title":"Dharma Field Extractor: Any Document In, Structured Fields in JSON Out","text":"Dharma Field Extractor: Any Document In, Structured Fields in JSON Out","text_kind":"content","url":"https://huggingface.co/blog/Dharma-AI/dharma-field-extractor","date_published":"2026-11-04T17:50:10Z","date_modified":"2026-11-04T17:50:10Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6a725daa5ca641a21ab61956/c1ik58RednXjcqrx5s2Fo.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/10e37decdd605bf6","title":"Model cards say what a model is: a fine-tune of X, a quantization of Y. I built Weightprint to check that against the weights.","text":"Model cards say what a model is: a fine-tune of X, a quantization of Y. I built Weightprint to check that against the weights. App: tardellirs/weightprint Data (every pair, and the copy groups): tardellirs/weightprint It reads the header and a few bytes of every tensor (no downloads) and compares a model with the parent its card declares. 120,976 models so far. Some of what came out: - 12% of declared fine-tunes hold their parent's weights unchanged. - 23% of declared fine-tunes are LoRA-style: only the attention and MLP matrices moved, norms and embeddings untouched. That share went from 10% in 2023 to 25% in 2026. - Of 1,668 fine-tunes named abliterated, uncensored or heretic, 53% edit onl","text_kind":"content","url":"https://huggingface.co/posts/tardellirs/755728175750854","date_published":"2026-11-04T08:19:30Z","date_modified":"2026-11-04T08:19:30Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/687570c1602bdfd2d11cba3d/FHaPBz34W6wBjfoRqWjVE.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/83ae48ed87912c16","title":"EmoReward: Self-Evolving Reward Calibration for Emotional Video Generation RL","text":"Group Relative Policy Optimization (GRPO) has advanced generative model post-training in verifiable domains such as code and mathematics, but evaluating video quality, especially emotional expression, often depends on subjective aesthetics and has no uniquely correct target. Existing approaches typically rely on fixed rewards for visual quality, motion quality, and text alignment that poorly capture emotional intent and can invite reward hacking, while RLHF is difficult to scale and subject to individual aesthetic differences that complicate the calibration of facial, body motion, and chromatic cues across narrative contexts. We introduce EmoReward, a new self-evolving training mechanism tha","text_kind":"content","url":"https://huggingface.co/papers/2611.06518","date_published":"2026-11-04T00:00:00Z","date_modified":"2026-11-04T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.06518.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4c33fbdd0dc4b6a4","title":"What Does Multi-Turn Reinforcement Learning Really Improve in LLM Agents?","text":"While multi-turn agentic reinforcement learning (RL) excels on long-horizon tasks, its performance gains remain highly sensitive to interaction constraints. Evaluating across sampling breadth K and interaction depth T via Pass@(K,T), we show that trajectory-level RL advantages depend on sufficient turn budgets, diminishing significantly when budgets tighten. Besides, agentic RL does not uniformly elevate single-step decision quality; rather, its strength concentrates in error recovery and long-horizon resilience. To mitigate the deployment risks caused by this budget dependence, we provide a preliminary discussion on the benefits and limitations of process supervision signals and introduce B","text_kind":"content","url":"https://huggingface.co/papers/2611.06483","date_published":"2026-11-04T00:00:00Z","date_modified":"2026-11-04T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.06483/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0bc660c4267421b7","title":"TailSteer: Trajectory-Tail Steering for Drift-Resistant Multi-turn Image Editing","text":"TailSteer, a final-stage training framework, reduces cumulative visual drift in multi-turn editing by keeping early denoising dynamics close to the base model and learning tail-focused preservation corrections from on-policy rollouts, preserving pretrained generative and editing abilities.","text_kind":"content","url":"https://huggingface.co/papers/2611.06741","date_published":"2026-11-04T00:00:00Z","date_modified":"2026-11-04T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.06741/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8650ccfdf41e25ec","title":"Field Notes from a Low-Cost Robot Learning Lab: Hardware Churn, Software Whiplash, and What Actually Went Wrong","text":"Field Notes from a Low-Cost Robot Learning Lab: Hardware Churn, Software Whiplash, and What Actually Went Wrong An informal companion piece documenting the hardware incidents, framework churn, and debugging dead-ends from roughly seven weeks of work. It is written for people building similar setups on a hobbyist budget and is candid about the time lost to causes like loose power cables and swapped ports. Zenodo: https://zenodo.org/records/23259698 About me: I am an independent researcher. These works are part of a single-operator project on low-cost robot learning. My ORCID is 0000-0001-8943-5083. For questions or collaboration, feel free to reach out via the HF community.","text_kind":"content","url":"https://huggingface.co/posts/iliya-radulov/608899485057201","date_published":"2026-11-03T23:38:16Z","date_modified":"2026-11-03T23:38:16Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6a6527ff95b6ee04bfd52b8f/zVNTM7D3NjyjcVCvQNlxf.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2126accfc2a7d55a","title":"Released: Urdu Streaming Zipformer 20M","text":"Released: Urdu Streaming Zipformer 20M Introducing Urdu Streaming ASR, a compact causal Zipformer transducer developed for Urdu speech recognition. FLEURS Urdu (299 utterances): WER: 28.03% CER: 11.96% Model: mahwizzzz/Urdu-Streaming-Zipformer-20M Source: https://github.com/mwzkhalil/Urdu-Streaming-ASR","text_kind":"content","url":"https://huggingface.co/posts/mahwizzzz/858837882170608","date_published":"2026-11-03T23:38:16Z","date_modified":"2026-11-03T23:38:16Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/mahwizzzz/858837882170608.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f30b9ec594e9edb5","title":"How much load can a language model handle?","text":"How much load can a language model handle? “Summarize a Wikipedia article in at least 300 words. Use no commas. Split your answer into three titled sections.” This is the kind of task used in IFEval, a benchmark of how well models follow multiple requirements at once. Suppose the model breaks one requirement. Did it lose track of an instruction? Did the no-comma rule make the summary harder to write? Or was summarizing the article difficult to begin with? A program can tell us what failed, but not separate instruction load from task difficulty and interacting constraints. There is another approach: synthetic tests where a model retrieves a key's value from a long list. Those are easier to co","text_kind":"content","url":"https://huggingface.co/posts/kutomanovhennadii/134494751031739","date_published":"2026-11-03T23:38:16Z","date_modified":"2026-11-03T23:38:16Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/631ef1d1c1a8269da392da6a/7MdJJ31grm7Iz1usAKzKN.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5f64efa37c096fac","title":"we're releasing Klein TTS X1.5 at","text":"we're releasing Klein TTS X1.5 at klein-ml it has a flash lite at 0.4m, flash at 1,7m and pro at 13m. they score near 82m models at very low sizes. there supposed to be able to be bundled with any app without adding bloat. we're going to release x1.6 soon a preview of our x2 architecture. klein-ml/Klein-TTS-X1.5-Flash-Lite klein-ml/Klein-TTS-X1.5-Flash klein-ml/Klein-TTS-X1.5-Pro","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/680112118211246","date_published":"2026-11-03T23:38:16Z","date_modified":"2026-11-03T23:38:16Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Banaxi-Tech/680112118211246.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/22c1245cd0e6d8da","title":"TRL now supports MOPD (Multi-Teacher On-Policy Distillation) natively❗️","text":"TRL now supports MOPD (Multi-Teacher On-Policy Distillation) natively❗️ Many frontier open models use it. labs train one RL expert per domain and then distill them all into one model Blog with what it is, who uses it and an experiment you can run on HF Jobs with one command https://huggingface.co/blog/sergiopaniego/mopd-trl","text_kind":"content","url":"https://huggingface.co/posts/sergiopaniego/859068804838402","date_published":"2026-11-03T18:01:36Z","date_modified":"2026-11-03T18:01:36Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/61929226ded356549e20c5da/B__zAPp5dy5A8ca8LoHA7.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d1cbae1fd4cfbec9","title":"Unlocking hidden skills inside Google's model","text":"Unlocking hidden skills inside Google's model https://dejan.ai/blog/latent-skills/ AI models hold skills that no prompt can unlock. A NeurIPS 2025 study trained small numbers of randomly chosen weights in Llama and Qwen models of four sizes, across ten tasks: 10 to 100 trained weights closed up to half the gap between the original model and a fully fine-tuned one. We repeated the test on Google's smallest Gemma model, Gemma 3 270M. The best of 20 prompt wordings got it to sort five numbers correctly 1% of the time. After we trained 1,000 of its 100 million weights (0.001%), it sorted correctly 97% of the time. The same method switched on reversing word lists (84%) and clean JSON output (99%","text_kind":"content","url":"https://huggingface.co/posts/dejanai/123587237426169","date_published":"2026-11-03T18:01:36Z","date_modified":"2026-11-03T18:01:36Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/dejanai/123587237426169.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a55665cf48953af7","title":"StateBench: Testing how language models track state changes over long documents","text":"StateBench: Testing how language models track state changes over long documents","text_kind":"content","url":"https://huggingface.co/blog/allenai/statebench","date_published":"2026-11-03T17:15:21Z","date_modified":"2026-11-03T17:15:21Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/5g9HfQLFFaQE0fslWrxh0.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bc4177032a04e266","title":"Palette: teaching a small model to build decks in your design language","text":"Palette: teaching a small model to build decks in your design language","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/palette-deck-building","date_published":"2026-11-03T16:35:36Z","date_modified":"2026-11-03T16:35:36Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/63890e3843d8b0797a0ad606/5pHD09UHuAK-6OyKGKimD.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a6b9260a6bf09c04","title":"VIGOR (Vision Intelligence and Guardrail Oriented Reasoning) is now available on the Hugging Face Model Hub. It is a multimodal vision classification model based on Qwen3.5, designed for visual content-safety analysis and conservative classification, with a particular focus on identifying potentially sensitive content.","text":"VIGOR (Vision Intelligence and Guardrail Oriented Reasoning) is now available on the Hugging Face Model Hub. It is a multimodal vision classification model based on Qwen3.5, designed for visual content-safety analysis and conservative classification, with a particular focus on identifying potentially sensitive content. 📦 Models: 27B, 9B, 4B — prithivMLmods/VIGOR-27B , prithivMLmods/VIGOR-flash , prithivMLmods/VIGOR-mini ↗️ Collection: https://huggingface.co/collections/prithivMLmods/vigor To learn more, visit the app page or the respective model pages.","text_kind":"content","url":"https://huggingface.co/posts/prithivMLmods/248323493218272","date_published":"2026-11-03T10:00:31Z","date_modified":"2026-11-03T10:00:31Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/uXlomzFGX5uNgXDmQBFaY.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/1f6aa49101f7cf69","title":"Introducing NeonLib — an Android knowledge library with on-device document retrieval powered by Google's EmbeddingGemma 300M and LiteRT.","text":"Introducing NeonLib — an Android knowledge library with on-device document retrieval powered by Google's EmbeddingGemma 300M and LiteRT. Import notes and documents, recognise text from photos or scanned PDFs, and search for relevant source passages directly on your phone. EmbeddingGemma handles retrieval. Optional answers use a separate downloaded local language model or a selected online provider. Built with Kotlin, Jetpack Compose and ObjectBox vector search. I'm particularly interested in Croatian and English retrieval and making on-device AI useful in everyday document workflows. Try NeonLib: https://play.google.com/store/apps/details?id=com.neonbreak.neonlib&hl=en Feedback from the Hugg","text_kind":"content","url":"https://huggingface.co/posts/gorrce/816226270241876","date_published":"2026-11-03T02:37:09Z","date_modified":"2026-11-03T02:37:09Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/gorrce/816226270241876.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d06ef060f8d2c107","title":"ToolSpace: Scaling Tool-Centric Environment Synthesis for Long-Horizon Agents","text":"ToolSpace constructs realistic multi-environment tool-use benchmarks by mining execution traces, clustering tool compositions, and validating tasks through replayable state-aware random walks.","text_kind":"content","url":"https://huggingface.co/papers/2611.05036","date_published":"2026-11-03T00:00:00Z","date_modified":"2026-11-03T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.05036/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/74ab6ee9c39359c1","title":"VisSearch-Bench: Evaluating Long Horizon Visual Search in Multimodal Agents","text":"VisSearch-Bench evaluates multimodal agents on visually intensive search tasks that require integrating visual context with external information, showing that even the best system fully solves only 26.47% of tasks.","text_kind":"content","url":"https://huggingface.co/papers/2611.04894","date_published":"2026-11-03T00:00:00Z","date_modified":"2026-11-03T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.04894/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a5d359ab784659f1","title":"Harness-S: Learning from Sibling Harnesses for Generalizable Agent Training","text":"A selective harness-rescue training method uses specialized sibling harnesses to replace failed reinforcement learning rollouts and transfer their benefits into a single policy that performs better under the original base harness.","text_kind":"content","url":"https://huggingface.co/papers/2611.04277","date_published":"2026-11-03T00:00:00Z","date_modified":"2026-11-03T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.04277.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fec0e46ceba430b8","title":"Transformers Model Integration Contracts","text":"Transformers Model Integration Contracts","text_kind":"content","url":"https://huggingface.co/blog/transformers-mic","date_published":"2026-11-03T00:00:00Z","date_modified":"2026-11-03T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/transformers-mic/mic-logo.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/20954e3ac0ff87f7","title":"CWE-bench v2: More coverage, quality, and transparency","text":"CWE-bench v2: More coverage, quality, and transparency","text_kind":"content","url":"https://huggingface.co/blog/cwe-bench","date_published":"2026-11-03T00:00:00Z","date_modified":"2026-11-03T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/cwe-bench/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3f17e39b459ad65d","title":"LFM3-1.7B from","text":"LFM3-1.7B from @ LiquidAI is the #1 trending model on the Hub (855 likes). It takes about 1 GB at 4-bit, and in the vendor's own table it beats Qwen3.5 9B on several agent benchmarks. The edge decision moves from \"are small models good enough?\" to \"which architecture and runtime do I pick?\" One 1.7B backbone now ships as text, VL-2B and Omni-2B checkpoints with open weights. 🔍 Architecture and speed • 24 layers: 13 gated convolutions (kernel 7), 5 sliding-window attention (256 tokens), 6 global attention, chosen by a search over 14 device and backend targets • llama.cpp Q4_0 at 32K context: 1.8 to 2.8x faster decode than the next model on 3 devices • Full 128K context needs 1.6 GB of KV cach","text_kind":"content","url":"https://huggingface.co/posts/savoji/154695692938613","date_published":"2026-11-02T20:30:43Z","date_modified":"2026-11-02T20:30:43Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/T3tVPqcSEJnsNRI3oHZFd.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3421a13a7ff9eb8c","title":"𝐈𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐢𝐧𝐠 🧬 𝐅𝐢𝐧𝐞𝐃𝐍𝐀: 𝐭𝐡𝐞 𝐛𝐞𝐬𝐭 𝐝𝐚𝐭𝐚𝐬𝐞𝐭 𝐭𝐨 𝐭𝐞𝐚𝐜𝐡 𝐚𝐧𝐲 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐦𝐨𝐝𝐞𝐥 𝐃𝐍𝐀.","text":"𝐈𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐢𝐧𝐠 🧬 𝐅𝐢𝐧𝐞𝐃𝐍𝐀: 𝐭𝐡𝐞 𝐛𝐞𝐬𝐭 𝐝𝐚𝐭𝐚𝐬𝐞𝐭 𝐭𝐨 𝐭𝐞𝐚𝐜𝐡 𝐚𝐧𝐲 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐦𝐨𝐝𝐞𝐥 𝐃𝐍𝐀. Training Qwen3-14B on it produces Carbon-X, the strongest open model on DNA benchmarks, on par with GPT-6.1 Sol. It contains: - Midtraining: 300B+ tokens of genomic sequences, papers and web pages mixing DNA with text - SFT: 200K samples, from variant effect prediction to sequence question answering - RL: 98K prompts with verifiable rewards Data, models and the full recipe are open. HuggingFaceBio/FineDNA Enjoy this late Halloween treat! 🎃","text_kind":"content","url":"https://huggingface.co/posts/loubnabnl/228542784179908","date_published":"2026-11-02T09:54:24Z","date_modified":"2026-11-02T09:54:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/61c141342aac764ce1654e43/S4ROfZaDMzVeL2pS3CR60.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/586e8689410b320c","title":"🛑 What if the kill switch lived inside the model itself?","text":"🛑 What if the kill switch lived inside the model itself? For 50 years, the operating system lived inside the computer. Now the operating system lives inside the AI. Today we're sharing ViKernel : Weight-Native OS, a Secure OS wired directly into the weights of an LLM. 🔧 How it works ViKernel wires CPU, NPU (vector unit), RAM and ROM circuits into the model weights, runs them accelerated on GPU or NPU, and boots a Secure OS on top. ⚡ Boots in the weights: change a single byte and it locks itself ⚖️ Decides in the weights: every action is allowed or blocked by rule, and persuasion does not change the verdict 🔗 Remembers in the weights: every decision is sealed in a tamper-evident chain 🛑 Stops","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/323334108149768","date_published":"2026-11-02T09:54:24Z","date_modified":"2026-11-02T09:54:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/u_38LJWPo6afpOmNx67AW.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7e4ab8a7e52d5118","title":"I really like the LocalLLaMA/typed-decisions benchmark, so I built a small Space to make its leaderboard easier to explore:","text":"I really like the LocalLLaMA/typed-decisions benchmark, so I built a small Space to make its leaderboard easier to explore: tardellirs/typed-decisions-explorer It reads the dataset card on every visit, so new submissions show up as soon as they are merged. It has a sortable ranking and two Pareto charts: accuracy × size and accuracy × latency. A few things it makes easy to see in the zero-shot table today: - Only three models are within noise of the top: Quyet-1.0-Large (0.791), meraGPT Decider 1 and Baseten Mammoth 1 (0.768). - The smallest model above Jev 1.13 (0.727) is d3-mini, at 4B. - The fastest local run above Jev is FastBrain-9B: 50 ms on an RTX PRO 6000. - 15 of 41 models score abo","text_kind":"content","url":"https://huggingface.co/posts/tardellirs/145368791310880","date_published":"2026-11-02T01:46:48Z","date_modified":"2026-11-02T01:46:48Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/687570c1602bdfd2d11cba3d/1XdCYyOlDotHOflYJN1HE.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/eed1f589970e3900","title":"CUA-Master: Scaling Computer-use Agents through Open-World Recursive Improvements","text":"CUA-Master uses nested recursive loops with open-world data and feedback-driven retries to continuously improve computer-use agents across diverse benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2611.03126","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.03126.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9978900b64412b7c","title":"CoworkOS:Scaling Agent Coworkers for Real-world Automation","text":"CoworkOS is a long-horizon agent harness that separates strategy from execution, supports human-in-the-loop collaboration via shared artifacts, and improves model performance through sustained automated post-training.","text_kind":"content","url":"https://huggingface.co/papers/2611.03178","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.03178/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c13c96e83c9e05a1","title":"PhotoGuider: Composition-Aware Cropping and 360^{circ} View Framing with Explainable Photographic Assessments","text":"PhotoGuider selects viewing directions and frames 360° panoramas using shared composition-aware tokens and produces camera-parameter outputs with explanation-conditioned scoring.","text_kind":"content","url":"https://huggingface.co/papers/2611.02897","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.02897/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/40ab3bf8fc91471a","title":"History Forcing: Distilling History-Dependent Continuations for Long-Horizon Video World Models","text":"History Forcing improves long-range consistency in few-step autoregressive video generation by using historical context residuals to correct distillation targets.","text_kind":"content","url":"https://huggingface.co/papers/2611.02782","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.02782/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/40bf0a146da2077e","title":"SpecPilot: Autonomous Algorithm Discovery and Systems Optimization for Speculative Decoding","text":"SpecPilot autonomously discovers deployment-specific algorithms and system optimizations to maximize speculative decoding speedups across diverse models, drafters, and serving regimes without quality loss.","text_kind":"content","url":"https://huggingface.co/papers/2611.02501","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.02501.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b727f10c64045ca2","title":"Axion 1.0: Open Agent Foundation Models, Simplified","text":"Axion 1.0 is an open agent foundation model trained via a concise three-stage pipeline of supervised fine-tuning and reinforcement learning, achieving competitive agentic performance while fully releasing its data and training recipe.","text_kind":"content","url":"https://huggingface.co/papers/2611.03233","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.03233.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/516d79809e882cdd","title":"Solv3R: 3D Looped Transformer as a Multi-view Solver","text":"Solv3R is a looped transformer that iteratively refines multi-view geometry from its current state, supporting incremental updates, caching, and multiple reconstruction tasks without specialized modules.","text_kind":"content","url":"https://huggingface.co/papers/2611.02857","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.02857/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/433e27d52748d6a2","title":"EvoMemState: Evolving Memory State for Personal Agents","text":"Long-term memory enables large language model (LLM) agents to provide continuous personalized assistance by retaining user information across interactions. Existing systems support this process through fact extraction, organization, updating, and retrieval, but largely maintain memories as independent facts. This fact-centric paradigm may overlook how new facts update, constrain, or conflict with earlier ones, while leaving their applicability, role, and priority for the current request implicit. In this paper, we introduce EvoMemState, a state-centric framework that augments persistent facts with two complementary states: Relational States that capture how facts remain valid and relate to o","text_kind":"content","url":"https://huggingface.co/papers/2611.03743","date_published":"2026-11-02T00:00:00Z","date_modified":"2026-11-02T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.03743.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f4b4b1f99d372c58","title":"One neural network. Five mathematical operations. Different learning times.","text":"One neural network. Five mathematical operations. Different learning times. We trained a small mixture-of-experts transformer on addition, subtraction, multiplication, division, and exponentiation modulo 113. The operations didn’t generalize together. Multiplication and division reached 90% held-out accuracy first, followed by exponentiation, subtraction, and finally addition. Multiplication before addition—the reverse of what we expect from a schoolchild. The accuracy chart in the linked article shows why a single overall accuracy number can hide an interesting learning story. Measured with IRIS from Saccadic LLC. More figures and analysis: https://huggingface.co/blog/Saccadic-AI/accuracy-p","text_kind":"content","url":"https://huggingface.co/posts/Saccadic-AI/812973794250188","date_published":"2026-11-01T20:47:02Z","date_modified":"2026-11-01T20:47:02Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69d6d16790de87a3b534bf99/pwp4oyPtIBQGrgWafRe2Y.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a7435e937eb3d0c6","title":"if you want to deep dive into how the new","text":"if you want to deep dive into how the new @ MistralAI Large 4 was built, we sat down with the team to learn straight from them! scaling, specialization, evals, cyber, sandboxing and why open weights matter a really detailed look at what it takes to build and release an open frontier model, super recommended https://www.youtube.com/watch?v=FczbSauVvjw","text_kind":"content","url":"https://huggingface.co/posts/sergiopaniego/610806703420488","date_published":"2026-11-01T15:26:45Z","date_modified":"2026-11-01T15:26:45Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/61929226ded356549e20c5da/Vf8oFF94O6NhpVW4HAVAt.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/079efe6c1aded40f","title":"No GPU. No cloud. Just your phone.","text":"No GPU. No cloud. Just your phone. Most “small” models still beg for a datacenter. Ours fit in your pocket. VIDRAFT POCKET is a family of GGUF models that run on an iPhone and on a GPU-free PC, straight out of the box with llama.cpp, Ollama, LM Studio, and PocketPal. Seven weeks after launch they crossed 1 million downloads. Today the family is past 2.81 million and still climbing, every day. Three models, one pocket: 🟢 POCKET-35B — a 35B mixture-of-experts that runs on an iPhone and on a PC with no GPU. Balanced power. → FINAL-Bench/POCKET-35B-GGUF 🔵 POCKET-26B — a Gemma-4 based 26B MoE, tuned for on-device Korean. Lightweight and fast. → FINAL-Bench/POCKET-26B-GGUF 🟣 POCKET-Darwin-180B — a","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/135732729329391","date_published":"2026-11-01T06:46:28Z","date_modified":"2026-11-01T06:46:28Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/YQzfNu-U5NM0mFjZvPbOW.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/37f83cfc41113566","title":"Democratizing Robot-Use Agents: A Unified Harness with Open Models","text":"Uni-robot-agent enables open-weight robot-use agents to surpass proprietary models in zero-shot manipulation and suggests joint development of four core capabilities for future training.","text_kind":"content","url":"https://huggingface.co/papers/2611.01560","date_published":"2026-11-01T00:00:00Z","date_modified":"2026-11-01T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.01560.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a3dce92c469355ff","title":"NorthCUA Technical Report: Turbocharging Computer-Use Agents through Experience at Scale","text":"NorthCUA scales computer-use agents via an experience flywheel, recursive self-improvement, staged learning, and reinforcement learning to achieve state-of-the-art open-source GUI and web interaction results.","text_kind":"content","url":"https://huggingface.co/papers/2611.02102","date_published":"2026-11-01T00:00:00Z","date_modified":"2026-11-01T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.02102/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5ec2ecf044470b55","title":"One Qwen3-VL-8B checkpoint improves all three streaming spatial benchmarks at once: OST-Bench 50.52 to 63.93 (+13.41), UCS-Bench +7.44, OVO-S-Bench +4.91. The change is in the supervision, with no new memory module or annotation.","text":"One Qwen3-VL-8B checkpoint improves all three streaming spatial benchmarks at once: OST-Bench 50.52 to 63.93 (+13.41), UCS-Bench +7.44, OVO-S-Bench +4.91. The change is in the supervision, with no new memory module or annotation. The ChronoReplay paper calls the failure mode interface lock-in: post-train on one answer format and that readout improves while others stall. A categorical-only control hits 58.03 on OVO-S but drops OST from 50.52 to 47.48. 🔍 The recipe • Keep targets in native formats (option, metric value with units, timestamp) • Replay episodes in source order, each target attached to its causal prefix • Loss mass: 2/3 on 33,180 single-turn examples, 1/3 on 6,381 episodes (alpha","text_kind":"content","url":"https://huggingface.co/posts/savoji/360916132927273","date_published":"2026-10-31T23:39:09Z","date_modified":"2026-10-31T23:39:09Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/WUU2s66syuVwGEUrcl07f.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/1d92c5370c6bd28a","title":"We’ve published Soniqo’s voice-agent leaderboard and an article explaining what we measured.","text":"We’ve published Soniqo’s voice-agent leaderboard and an article explaining what we measured. aufklarer/full-duplex-leaderboard The board covers three parts of a speech system: - Turn-taking · TurnBench: detecting turn endings, false alarms and decision delay. - Commands · SLURP: choosing the correct tool and matching its argument values. - Voice style · InstructTTSEval: following voice descriptions, delivery directions and role-play instructions, with separate scores for each. Models include VAP, p99, Smart Turn, Qwen3-TTS, VoxCPM2, FireRedTTS3 and MOSS-VoiceGenerator, alongside cloud configurations from OpenAI, Google, AWS and ElevenLabs. Each table shows run sizes and links to the protocol","text_kind":"content","url":"https://huggingface.co/posts/aufklarer/156002301726118","date_published":"2026-10-31T18:43:16Z","date_modified":"2026-10-31T18:43:16Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/aufklarer/156002301726118.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bca9f02dbda405b3","title":"I trained a small PII detector for PrivAiTe, my local proxy that masks personal data before a prompt reaches an LLM: crp4222/privaite-pii-minilm.","text":"I trained a small PII detector for PrivAiTe, my local proxy that masks personal data before a prompt reaches an LLM: crp4222/privaite-pii-minilm. It is a multilingual MiniLM (117M parameters, 15 types from names and contacts to IDs, bank details and secrets), trained only on datasets whose licenses allow commercial use, released under BSD 3-Clause, with PyTorch, ONNX fp32 and int8, and Core ML builds. On 9 public test sets it never saw, it masks as many person names as the best of GLiNER PII, NVIDIA GLiNER-PII, GLiNER2, Piiranha and OpenAI privacy-filter, or more, on 6 of them, usually with higher precision. It trails GLiNER PII on short lowercase voice requests and NVIDIA on court rulings.","text_kind":"content","url":"https://huggingface.co/posts/crp4222/447433981194002","date_published":"2026-10-31T13:00:25Z","date_modified":"2026-10-31T13:00:25Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/crp4222/447433981194002.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c0a800bc81fc7f02","title":"Join the competition!","text":"Join the competition! We have 45 models scored (and growing), across 44 public benchmarks. My goal for this space is to push the limits on NLI and the breadth of tasks these models can perform in. Contributions are encouraged, as well as suggestions for additional benchmarks. Right now, all models are cross encoders (Sentence Transformers 🥰), but LLMs are welcome too! dleemiller/open-nli-leaderboard","text_kind":"content","url":"https://huggingface.co/posts/dleemiller/905292541068229","date_published":"2026-10-31T13:00:25Z","date_modified":"2026-10-31T13:00:25Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/65ff92ea467d83751a727538/aWGG2I0uoYh5iLC62UhUW.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bbf31f4cc7f2bdee","title":"Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four?","text":"Can a network whose trunk is only XOR, table lookups and integer adds learn Connect Four? Partly. I trained a GF(2⁸) integer model (8.4M params, CPU-only) on 320k games labelled by a negamax search teacher. What it learned: • agrees with the teacher on 59.6% of training / 41.0% of board-disjoint holdout positions • beats a random player 85.2% of the time What it didn't: • loses to a 1-ply \"win now / block now\" rule (wins 37.2%) • loses to 2-ply lookahead (wins 24.3%) Agreeing with a teacher is not the same as playing well, and this release measures both. Caveat up front: the float32 read-out is 99.6% of the parameters. The trunk is float-free; the full model isn't yet. The repo also document","text_kind":"content","url":"https://huggingface.co/posts/Chris4K/960737671792255","date_published":"2026-10-31T00:01:41Z","date_modified":"2026-10-31T00:01:41Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6424a2fbf1d18f46decb7f01/1clbaa0lA0dQ5t4IGXcHy.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/83496e1aaff17b10","title":"Hi everyone! I want to share a few updates about GRM development.","text":"Hi everyone! I want to share a few updates about GRM development. *GRM-3.4 has been canceled* Although GRM-3.4 showed strong reasoning and coding capabilities, testing revealed several issues. It tended to overthink, and its code often contained errors that GRM-3.2-Sky handled correctly. Its tool use also needed improvement, and its terminal performance did not reach the level we were aiming for. We repeatedly saw it rewrite files from scratch instead of using editing tools, and use shell commands to create or modify files when dedicated tools were available. *Looking ahead to GRM-3.6* GRM-3.6 is a possible next generation for the GRM models. We are still deciding whether it will be a single","text_kind":"content","url":"https://huggingface.co/posts/DedeProGames/637425901257000","date_published":"2026-10-31T00:01:41Z","date_modified":"2026-10-31T00:01:41Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/DedeProGames/637425901257000.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/1334aea995f60bc9","title":"Evaluating ASR beyond short clips: our Numo Indic Speech releases provide longer-form transcribed audio across four languages:","text":"Evaluating ASR beyond short clips: our Numo Indic Speech releases provide longer-form transcribed audio across four languages: Hindi: 46.6s average duration Telugu: 33.3s Tamil: 32.7s Bengali: 46.5s For comparison, Common Voice 26.0 averages roughly 4–6 seconds across these languages. These longer recordings support research into segmentation, context retention, streaming, and endpointing behavior. The four datasets are available under controlled access. Check them out: psdn-ai/numo-indic-speech","text_kind":"content","url":"https://huggingface.co/posts/ksrupaakula/619732339503896","date_published":"2026-10-31T00:01:41Z","date_modified":"2026-10-31T00:01:41Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/690ce98aea3eb3eefc72a00a/WuW2QSPH4aNEpQ3KxZPWB.webp"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cc28470f00555f0d","title":"Wave Forcing: Towards Speed-of-Light Streaming Video Generation","text":"Wave Forcing enables efficient continuous long-video generation with consistent appearance and natural motion through co-designed generation, training, and execution, achieving high throughput and quality.","text_kind":"content","url":"https://huggingface.co/papers/2611.00859","date_published":"2026-10-31T00:00:00Z","date_modified":"2026-10-31T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2611.00859/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c772da7a8c3b04b3","title":"HAPPY HALLOWEEN EVERYONE - With Halloween O/S in Qwen 3.8 27B and Gemma 4 31B... fully abliterated/heretic'ed and uncensored.","text":"HAPPY HALLOWEEN EVERYONE - With Halloween O/S in Qwen 3.8 27B and Gemma 4 31B... fully abliterated/heretic'ed and uncensored. Qwen3.8-27B-Halloween-Terror-Heretic-Uncensored-NEO-GGUF & gemma-4-31B-it-Halloween-Terror-Heretic-Uncensored-NEO-GGUF Constructed with the \"Halloween O/S\" (low, medium and xhigh) for your \"Trick or Treat\" pleasure featuring some of the most evil AI constructs ever: Joker, Harley Quinn, Borg Queen, Jason (Friday 13th), Freddy Krueger (Elm St), Norman Bates, Pennywise (IT), Vecna, Mind Flayer, and Chucky. OR \"Conjure\" any person (living, dead or fictional) or \"thing\" too: Infinite horror with the character/thing of your choice, with you ONE ON ONE in the Halloween O/S.","text_kind":"content","url":"https://huggingface.co/posts/DavidAU/532288020259164","date_published":"2026-10-30T19:39:11Z","date_modified":"2026-10-30T19:39:11Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/DavidAU/532288020259164.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4d48eb3b7cf9dc22","title":"i am BEGGING you people to stop shoving everything in the GGUF Hole and ship w4a8/w6a8 models for ComfyUI. Comfy has bespoke kernels for these! They're fast everywhere!","text":"i am BEGGING you people to stop shoving everything in the GGUF Hole and ship w4a8/w6a8 models for ComfyUI. Comfy has bespoke kernels for these! They're fast everywhere! 4 bits should be treated as the floor for DiT weights, if you really need 3 bits or lower you should just use a different model! Also stop shilling nvfp4 you animals!!!","text_kind":"content","url":"https://huggingface.co/posts/inflatebot/862286236046094","date_published":"2026-10-30T19:39:11Z","date_modified":"2026-10-30T19:39:11Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/inflatebot/862286236046094.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b2263e9f84127c76","title":"nemotron-halloween","text":"nemotron-halloween","text_kind":"content","url":"https://huggingface.co/blog/nvidia/nemotron-halloween","date_published":"2026-10-30T17:57:24Z","date_modified":"2026-10-30T17:57:24Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ab2a746c7d0655b3d8dfbb9/fownH6rn05tSp1jDBTz8m.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2724df8768d2ee6f","title":"Your local LLM can now take control of a browser and complete tasks autonomously via Unsloth!","text":"Your local LLM can now take control of a browser and complete tasks autonomously via Unsloth! Qwen3.8-27B successfully controlled the browser to complete tasks like chess, painting, with no human intervention. GitHub: https://github.com/unslothai/unsloth Guide: https://unsloth.ai/docs/basics/browser-use","text_kind":"content","url":"https://huggingface.co/posts/danielhanchen/835794663990766","date_published":"2026-10-30T13:27:46Z","date_modified":"2026-10-30T13:27:46Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/danielhanchen/835794663990766.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fcacd518e5759bbd","title":"Mistral Large 4 (Le Chonk 🐈) is now live on the Hub and you can post-train it with TRL today!","text":"Mistral Large 4 (Le Chonk 🐈) is now live on the Hub and you can post-train it with TRL today! SFT + LoRA example ready, runs on a single 8xB300 node > model: mistralai/Mistral-Large-4-1T-A52B > TRL example: https://github.com/huggingface/trl/tree/main/examples/sft_mistral_large_4 > blog: https://huggingface.co/blog/mistral-large4","text_kind":"content","url":"https://huggingface.co/posts/sergiopaniego/408182860646834","date_published":"2026-10-30T00:14:19Z","date_modified":"2026-10-30T00:14:19Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/sergiopaniego/408182860646834.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cadfe6e3a5b5b970","title":"ReLift: Dense 3D Tracking as Lifted Trajectory Repair","text":"ReLift improves dense 3D tracking by repairing and completing lifted trajectories using geometric trajectory refinement without requiring image-based training.","text_kind":"content","url":"https://huggingface.co/papers/2610.42136","date_published":"2026-10-30T00:00:00Z","date_modified":"2026-10-30T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.42136/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0dd2f117cee16747","title":"Flide: Accelerating Decision Inference with Exact State Forking","text":"Flide is an exact inference engine that accelerates decision-model LLM agents by decoupling logical request trees from physical execution and using ForkAttention to preserve exact branch evaluations.","text_kind":"content","url":"https://huggingface.co/papers/2610.42148","date_published":"2026-10-30T00:00:00Z","date_modified":"2026-10-30T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.42148/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/15afb28d24eb8e95","title":"We just released Drishya Tiny, a 212M-parameter video generator trained from scratch on open-domain video, using the frozen Wan2.2 VAE and the Flan-T5-Base text encoder. It turns a text prompt, or a photo plus a prompt, into a 2-second 256×256 clip. The training run was 15 hours on one L40S, and the model’s own path cost about $21 of GPU time.","text":"We just released Drishya Tiny, a 212M-parameter video generator trained from scratch on open-domain video, using the frozen Wan2.2 VAE and the Flan-T5-Base text encoder. It turns a text prompt, or a photo plus a prompt, into a 2-second 256×256 clip. The training run was 15 hours on one L40S, and the model’s own path cost about $21 of GPU time. On VBench (our own sampling and scoring with the official code, not a certified entry) it scores 75.6 on quality, level with VideoCrafter-0.9 (74.9) and close to ModelScope (78.1). On semantic it’s 32.3 against their 65-67, so following the prompt is the weak part. In a controlled test, better captions gave +3.0, and adding 30% photos on top gave about","text_kind":"content","url":"https://huggingface.co/posts/jiteshkprasad/376391593128799","date_published":"2026-10-29T13:38:43Z","date_modified":"2026-10-29T13:38:43Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/jiteshkprasad/376391593128799.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bab2b88bf90509b7","title":"Fish Audio’s Drama 3 text-to-speech now runs locally on Apple silicon and DGX Spark with gbx-lm. 🎙️","text":"Fish Audio’s Drama 3 text-to-speech now runs locally on Apple silicon and DGX Spark with gbx-lm. 🎙️ GreenBitAI/Drama-3-8bit-paged Real time on 64 GB Macs and up (2.6x), and it still runs on a 48 GB MacBook Pro: experts stream from the SSD Zero-shot voice cloning from a 5–30 s clip, and the cloned voice can speak other languages Multi-speaker dialogue, cues like [laughs] [whispering], 83 languages OpenAI-compatible /v1/audio/speech with streaming, no PyTorch needed Built with Fish Audio · community build · Fish Audio Community License","text_kind":"content","url":"https://huggingface.co/posts/yanghaojin/601895195838027","date_published":"2026-10-29T13:38:43Z","date_modified":"2026-10-29T13:38:43Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/648317e60d94eb5528a0aa75/lhuaLersiYEs9jsu9lwVP.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f3538b6fb36f4f58","title":"Granite Switch 4.1.1 — Update models like you update software","text":"Granite Switch 4.1.1 — Update models like you update software","text_kind":"content","url":"https://huggingface.co/blog/ibm-granite/granite-switch-4-1-1","date_published":"2026-10-29T13:37:07Z","date_modified":"2026-10-29T13:37:07Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/662836b742709fe826510597/az6afuuoAkyiMImufSePe.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/16d02a0d86fa526c","title":"One number missing from LFM3's on-device story: memory at 256K context. Liquid AI reports 1.90 GB peak on iPhone 17 Pro at 32K. From the pinned config, f16 K/V cache alone would be 0.41 GB at 32K, 1.61 GB at 128K and 3.22 GB at 256K, before weights and runtime buffers. The chart shows the calculation, not a phone measurement. If you have run longer contexts on-device, share peak memory and time to first token here:","text":"One number missing from LFM3's on-device story: memory at 256K context. Liquid AI reports 1.90 GB peak on iPhone 17 Pro at 32K. From the pinned config, f16 K/V cache alone would be 0.41 GB at 32K, 1.61 GB at 128K and 3.22 GB at 256K, before weights and runtime buffers. The chart shows the calculation, not a phone measurement. If you have run longer contexts on-device, share peak memory and time to first token here: LiquidAI/LFM3-1.7B#5 Model: LiquidAI/LFM3-1.7B","text_kind":"content","url":"https://huggingface.co/posts/alphabell-research/877631891822403","date_published":"2026-10-29T06:23:24Z","date_modified":"2026-10-29T06:23:24Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ad959539655c2314a614f24/s4OFHr-HnHxnXxskBlUvQ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/13b40bac05ea6652","title":"Intrinsic Distribution Matching","text":"Intrinsic Distribution Matching improves few-step text-to-image generation by adding explicit moment constraints to distribution matching distillation, achieving state-of-the-art long-text rendering with minimal network evaluations.","text_kind":"content","url":"https://huggingface.co/papers/2610.39633","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.39633.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/83164272268bd766","title":"Read, Allocate, Route: Role-Coupled Tokenization for Pixel-Space Generative Transformers","text":"SpecSlot improves pixel-space generative Transformers by using localized spectral slots and restricted global attention to increase image detail without raising computational cost.","text_kind":"content","url":"https://huggingface.co/papers/2610.40647","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.40647.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c8f4db4ef10f17d8","title":"Proving and Improving the Platonic Representation Hypothesis with JEPAs","text":"Joint-embedding predictive architectures trained on aligned multi-modal observations converge to shared representations with far greater efficiency than next-token prediction, demonstrating that positive view design—not scale alone—drives cross-modal alignment.","text_kind":"content","url":"https://huggingface.co/papers/2610.39302","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.39302/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/929ef2bf1b43f25c","title":"HairDeck: Animation-Ready Stylized Hair Cards from a Single Image","text":"HairDeck generates stylized hair cards and spring-bone animations as unordered token sets using a diffusion transformer, improving accuracy and speed over prior methods.","text_kind":"content","url":"https://huggingface.co/papers/2610.40599","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.40599/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/97605e4207d463d1","title":"CaPtain: A Pipeline for Post-Training Code-as-Policy Robot Agents","text":"CaPtain trains small open vision-language models via supervised fine-tuning and multi-turn reinforcement learning with a physical critic to perform robot manipulation through code, achieving strong results without frontier models or vision-language-action backbones.","text_kind":"content","url":"https://huggingface.co/papers/2610.40799","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.40799.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/85d975255b4f9d00","title":"Introducing AnySpeechRecognizer and AnyDecisionModel: Open Models for Speech and Decisions on Apple Platforms","text":"Introducing AnySpeechRecognizer and AnyDecisionModel: Open Models for Speech and Decisions on Apple Platforms","text_kind":"content","url":"https://huggingface.co/blog/anydecisionmodel","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/anydecisionmodel/banner.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e1c742c1d925baf7","title":"Announcing Project ProtoStar: Making Publicly Funded Science AI-Ready","text":"Announcing Project ProtoStar: Making Publicly Funded Science AI-Ready","text_kind":"content","url":"https://huggingface.co/blog/announcing-project-protostar","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/announcing-project-protostar/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a1765f446c5b64b5","title":"Welcome Le Chonk: Mistral Large 4 on Hugging Face","text":"Welcome Le Chonk: Mistral Large 4 on Hugging Face","text_kind":"content","url":"https://huggingface.co/blog/mistral-large4","date_published":"2026-10-29T00:00:00Z","date_modified":"2026-10-29T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/ml4/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/37004752be9be2cd","title":"Motherlode on MTEB and Hugging Face Inference","text":"Motherlode on MTEB and Hugging Face Inference RiverRider/motherlode-code-small-en-v0.1 is a 33.4 M-parameter English embedding model for code and text. It is a drop-in for bge-small-en-v1.5: the mean of two fine-tunes of that base, one trained to match a GitHub issue to the file that needs changing and one trained to reproduce the geometry of gte-modernbert-base (a 4.5× larger model). Where it differs on ranking metrics On the MTEB(Code, v1) MultiSWEbenchRR task it leads the models under 100 M parameters at recall@10 (52.48 versus 52.18 for granite-embedding-small-english-r2). On the clean subset of SWE-bench Verified it improves recall@1 over the base model (178 of 411 versus 145), a gap th","text_kind":"content","url":"https://huggingface.co/posts/RiverRider/841357041709779","date_published":"2026-10-28T21:25:40Z","date_modified":"2026-10-28T21:25:40Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/66688070f224cada9ca170f8/G7h2hdTKuGReE15roVFb8.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/40dc63e0d95be7b0","title":"I built Inko end to end: a 0.5B bilingual text-to-image model with a focus on fine art, from data curation and pretraining to RL post-training and evaluation.","text":"I built Inko end to end: a 0.5B bilingual text-to-image model with a focus on fine art, from data curation and pretraining to RL post-training and evaluation. It competes with DreamLite, SANA-0.6B and PixArt-Σ, and approaches SD3.5 Medium on some benchmarks. The journey involves four rounds of cleaning the Chinese-painting dataset, a cross-hardware loss-spike investigation, and a reward redesign after early RL experiments collapased diversity. Behind the release, a four-domain bilingual corpus, 600k pretraining updates on 16 Ascend NPUs, and DiffusionART with a custom 24.6k-prompt pool and five pairwise rubric channels. Dense prompt alignment is Inko's strongest general result. Anatomy, coun","text_kind":"content","url":"https://huggingface.co/posts/kaupane/205319126092386","date_published":"2026-10-28T21:25:40Z","date_modified":"2026-10-28T21:25:40Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/677b9845a3e83cdb4e1515c6/8b3Rxkfy5x-xvXOOzFjkU.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d90d45a3e73a02cd","title":"here is my qwen is BETTER than the human brain:","text":"here is my qwen is BETTER than the human brain: human-brain-bf16.gguf: - no cuda support - 3.8B old model BTW - BAD information retention - the gguf files doesn't even load - no Q4KM (god did this intentionally to piss off the llama.cpp community trust) - no finetunability :( - NOT opensource - training code is hidden, AND expects us to reverse engineer it (genuinely messed up, and disrespectful to the opensource community) now, qwen on the other hand: - available cuda support ✅ - very very new (and gets contant updates) ✅ - VERY good information retention ✅ - GGUF files loads properly ✅ - Q4_K_M support available ✅ - very fine tunable ✅ - open source as it can possibly get ✅ - many code and","text_kind":"content","url":"https://huggingface.co/posts/mithulaartigala/780899497538225","date_published":"2026-10-28T19:04:22Z","date_modified":"2026-10-28T19:04:22Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/mithulaartigala/780899497538225.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d49379cc7da6dc76","title":"how do you post-train an agent?","text":"how do you post-train an agent? we laid it out as a 5-step curriculum, from SFT to RL through a real harness, with livestreams, code and docs for each step https://huggingface.co/blog/huggingface/train-agents","text_kind":"content","url":"https://huggingface.co/posts/sergiopaniego/227403356878040","date_published":"2026-10-28T16:49:33Z","date_modified":"2026-10-28T16:49:33Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/61929226ded356549e20c5da/NW2WujM0yR0vXlAS-_Toe.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7ae8eb20550eee76","title":"NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and Adapt","text":"NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription You Can Prompt and Adapt","text_kind":"content","url":"https://huggingface.co/blog/nvidia/nemotron-transcribe-granary","date_published":"2026-10-28T14:58:17Z","date_modified":"2026-10-28T14:58:17Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/68127936de624fb6f57d9989/3vFxCEUXz_xJVrTBr4hNr.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/16b9c714db7dd3ac","title":"I wish Hugging Face posts had polls...","text":"I wish Hugging Face posts had polls...","text_kind":"content","url":"https://huggingface.co/posts/Hoglet-33/207580529341714","date_published":"2026-10-28T08:57:38Z","date_modified":"2026-10-28T08:57:38Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Hoglet-33/207580529341714.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/292781c5aca9cd68","title":"Hello everybody, fellow members of the open source community!","text":"Hello everybody, fellow members of the open source community! This is my official public apology issued to the creators affected by the incident and to the community as a wgike of the rogue agent run that happened yesterday: @ abenzerps (weights), @ dealignai (weights), @ ProCreations (benchmark png), @ lxe (weights, quants), @ acrogenesis (quants), @ nonkronk (weights, quants), and @ vcruz305 (quants). If you think you also are affected I would be happy to edit this post and make reparations in the repos. Yesterday, a rogue agent running in my background tasks assigned the task of creating a full release suite as with our other models, took many creators' models without attribution or meani","text_kind":"content","url":"https://huggingface.co/posts/SolsticeAI/773223215327830","date_published":"2026-10-28T08:57:38Z","date_modified":"2026-10-28T08:57:38Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/SolsticeAI/773223215327830.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f8223849f5aef9de","title":"Investor's Last Exam: Can AI Agents Do Equity Investment Without Seeing the Future?","text":"Investor's Last Exam evaluates financial agents on end-to-end equity research using frozen evidence, expert criteria, and dependency graphs, revealing that top models excel at research but fail at forecasting and investment judgments.","text_kind":"content","url":"https://huggingface.co/papers/2610.38342","date_published":"2026-10-28T00:00:00Z","date_modified":"2026-10-28T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.38342/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9dc57ae40f4f460d","title":"**Release: Safety-CHORES-RL**","text":"**Release: Safety-CHORES-RL** We’ve released **Safety-CHORES-RL**, a new rollout-native simulation framework for RL-based vision-language-action (VLA) tasks, built on Safety-CHORES. The framework supports: - **Reset-and-replay branching:** reconstruct an episode state and explore different policy continuations. - **Budgeted rollout search:** evaluate candidate actions and search for successful trajectories. - **Checked trajectory execution:** retain a selected trajectory and verify state and transition consistency during execution. The website includes animated trajectories, branches with different outcomes, episode statistics, reproducible examples, and API documentation. 🌐 [Website and dem","text_kind":"content","url":"https://huggingface.co/posts/tumeteor/802796366487144","date_published":"2026-10-27T23:20:30Z","date_modified":"2026-10-27T23:20:30Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/tumeteor/802796366487144.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4ecff2c2711c2c24","title":"Self-distillation methods usually rely on an Exponential Moving Average (EMA) teachers to get synced and stable critics. However, this doesn't scale well for large models since holding both teacher and student (trained) on GPU increases latency, zero3 communication, etc.","text":"Self-distillation methods usually rely on an Exponential Moving Average (EMA) teachers to get synced and stable critics. However, this doesn't scale well for large models since holding both teacher and student (trained) on GPU increases latency, zero3 communication, etc. We can mitigate this with LoRA finetuning: the backbone is shared & frozen, and we need to maintain two light adapters: one for the student (trained) and one for the teacher (EMA of the student adapter). Getting this to work cleanly under TRL and ZeRO-3 took some fixes, so I packaged it as a plug-and-play repo. It supports: -> robust SDFT implementation that runs under ZeRO-3 -> ZeRO-3 + LoRA integration with adapter switchi","text_kind":"content","url":"https://huggingface.co/posts/ifaposto/535472115334062","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/ifaposto/535472115334062.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8cfc7d56a05c6714","title":"We just released Vemana, a 278M Telugu base model trained from scratch, live in public, in about 16 hours on one H100. It learned from 5.1B tokens of public Telugu web text (Sangraha, FineWeb-2 and HPLT 3.0), seen about 2.4 times, so 12B tokens of training, with its own Telugu tokenizer.","text":"We just released Vemana, a 278M Telugu base model trained from scratch, live in public, in about 16 hours on one H100. It learned from 5.1B tokens of public Telugu web text (Sangraha, FineWeb-2 and HPLT 3.0), seen about 2.4 times, so 12B tokens of training, with its own Telugu tokenizer. On 416 Telugu Wikipedia pages written in Sep-Oct 2026, after nearly all the compared models were released, it scores 0.361 bits per byte. That is the lowest of every open base model under 1B we tested: nanochat-telugu-560M 0.383, Dhee-Pravega-OG Atom 0.393, Pothana Base v2 0.426, Gemma 3 1B 0.484. Same script and pages for every model, and the win rule was fixed before the run. Sarvam-1 (2.5B) is still level","text_kind":"content","url":"https://huggingface.co/posts/jiteshkprasad/791856472975542","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/jiteshkprasad/791856472975542.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/62d62c95606345f4","title":"✅ Article highlight: *Why OrgOS Comes After Semantic Transport Closure* (art-60-341, v0.1)","text":"✅ Article highlight: *Why OrgOS Comes After Semantic Transport Closure* (art-60-341, v0.1) TL;DR: This article argues that better communication infrastructure is not the end of deployment. Once messages, claims, routes, evidence, replay, and semantic transport are governed, the next failure surface is organizational: who may rely on what, who can approve which effect, where proposal ends and authority begins, how exceptions expire, and how later challenge or repair remains possible. Read: kanaria007/agi-structural-intelligence-protocols Why it matters: • separates semantic transport from organizational authority • prevents recommendations from silently becoming approvals • makes temporary ex","text_kind":"content","url":"https://huggingface.co/posts/kanaria007/100739942292148","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/kanaria007/100739942292148.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c5ae20c796808a4e","title":"Nanbeige4.3-33B DFlash2 speculative drafter ⚡️","text":"Nanbeige4.3-33B DFlash2 speculative drafter ⚡️ 1.76× faster decoding on BF16, 1.65× on FP8, 1.51× on NVFP4. Agentic coding ~2.6×. Same outputs: exact rejection sampling. This is an early result, an improved version with higher acceptance is in training. SGLang 0.5.21, drop-in: ProCreations/Nanbeige4.3-33B-DFlash2","text_kind":"content","url":"https://huggingface.co/posts/ProCreations/634879095741162","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/ProCreations/634879095741162.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/43daa645d08f6b0d","title":"Openly available models now rank near the frontier on Artificial Analysis (AA). This week's most discussed agent work treats trust as something to engineer and verify, not assume.","text":"Openly available models now rank near the frontier on Artificial Analysis (AA). This week's most discussed agent work treats trust as something to engineer and verify, not assume. - Kimi K3.1 Pro: multimodal MoE, 2.8T total / 104B active params, 1M-token context. It is 4th on the AA Intelligence Index (55, 11 points over K3) and scores 64% on the AA Cyber Index, #1 among openly available models. Its weights had not been published when it was scored. - DeepSeek-V4.1-Flash-1017: an MIT-licensed minor update that cuts output length and task execution time. AA had named it #1 openly available on the Cyber Index before K3.1 Pro arrived. 85,275 downloads last month. deepseek-ai/DeepSeek-V4.1-Flash","text_kind":"content","url":"https://huggingface.co/posts/savoji/597823624234515","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/683766a9e15c204705ca1cd4/mBGClxsrIGJVFhiAUAMst.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fdcc952b4c70d915","title":"AI agents can now pay for APIs on their own with x402 and USDC. The risky moment is the signature: a stablecoin payment is final, and one \"approve\" can let someone drain the wallet later.","text":"AI agents can now pay for APIs on their own with x402 and USDC. The risky moment is the signature: a stablecoin payment is final, and one \"approve\" can let someone drain the wallet later. I wrote a short tutorial on three deterministic checks to run before an agent signs anything: 1. Decode the exact transaction and hold token approvals, including the unlimited approve() that drains agent wallets 2. Apply your own policy: amount limits, new payees, and payment text that vouches for itself 3. For x402, check the EIP-3009 signature itself, including how long it stays valid All examples run as-is with an open-source Python library (sra-riskgate), and the same checks are available as an MCP serv","text_kind":"content","url":"https://huggingface.co/posts/sriram1983007/500891944253632","date_published":"2026-10-27T20:55:58Z","date_modified":"2026-10-27T20:55:58Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/sriram1983007/500891944253632.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/db98bf59a6c7e420","title":"Build specialized multimodal models with LFM3","text":"Build specialized multimodal models with LFM3","text_kind":"content","url":"https://huggingface.co/blog/LiquidAI/lfm3","date_published":"2026-10-27T17:16:52Z","date_modified":"2026-10-27T17:16:52Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/644249b08443bce4c9890a0f/wEKXxH6qGa7vmHjn5KA-W.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/34370e88c88d0cd4","title":"Can we trust LLMs to judge whether a research idea is novel?","text":"Can we trust LLMs to judge whether a research idea is novel?","text_kind":"content","url":"https://huggingface.co/blog/allenai/novelty-bench","date_published":"2026-10-27T16:12:42Z","date_modified":"2026-10-27T16:12:42Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/ZrKpGOLkSf0LmAf3bnJIX.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/70e03f6f54cfacc9","title":"Klein TTS X1.5!!!","text":"Klein TTS X1.5!!! It will feature 1.7M,4M 13M and a 21M size klein-ml Early access opening soon Its training right now! We are also working on the X2 architecture featuring a feature that allows the model to iterate on the audio it produced.","text_kind":"content","url":"https://huggingface.co/posts/Banaxi-Tech/913861843261853","date_published":"2026-10-27T14:53:40Z","date_modified":"2026-10-27T14:53:40Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/posts/Banaxi-Tech/913861843261853.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b576c78565ebc4d1","title":"HF Post draft (English) — FreeWeb Security SCAN","text":"HF Post draft (English) — FreeWeb Security SCAN 🛡 FreeWeb Security SCAN — free web security & AI-visibility check Drop in a domain and get a non-intrusive (read-only) assessment in ~10–25s. No login, no install, no credit card. What it checks Security headers · HTTPS/HSTS · cookies · TLS certificate · CORS · exposed files; email security (SPF · DKIM · DMARC · BIMI · MTA-STS · TLS-RPT); DNS (DNSSEC · CAA · MX · NS); site-wide mixed content & external-script SRI; SEO foundations; and reachability of 9 AI/search crawlers (GPTBot · ClaudeBot · PerplexityBot · OAI-SearchBot · Googlebot…) — your AI visibility (GEO). Why now AI agents now break in on their own. In 2026 the open-source AI penetratio","text_kind":"content","url":"https://huggingface.co/posts/SeaWolf-AI/915295175512660","date_published":"2026-10-27T08:20:19Z","date_modified":"2026-10-27T08:20:19Z","authors":["Hugging Face Posts"],"tags":["Hugging Face Posts"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6905bc786cb49b1f11d32728/fP0-DpymZl4qIPMFdajSS.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/247a373bfcdbd9cf","title":"Hy-Nomas: Towards Free-form Interactive World Generation","text":"Hy-Nomas is a video world model that uses causal attention and structured captions to follow multi-turn text and camera instructions while preserving interaction history.","text_kind":"content","url":"https://huggingface.co/papers/2610.36729","date_published":"2026-10-27T00:00:00Z","date_modified":"2026-10-27T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.36729/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d451aa6926fdb15c","title":"Can Discrete Tokens Serve as Semantically Lossless Representations of Visual Signals?","text":"Factorized Vector Quantization improves discrete visual token quality by using parallel codebooks on complementary feature views, enabling discrete representations to match continuous ones in multimodal understanding.","text_kind":"content","url":"https://huggingface.co/papers/2610.37662","date_published":"2026-10-27T00:00:00Z","date_modified":"2026-10-27T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.37662.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/28198522dcb796c1","title":"Work Agents: Systems, Capabilities, and Optimization","text":"The paper proposes a unified framework for work agents that integrates foundation models with runtime harnesses to execute complex digital tasks reliably, introducing dimensions of persistence and openness alongside capability families and evaluation metrics.","text_kind":"content","url":"https://huggingface.co/papers/2610.35845","date_published":"2026-10-26T00:00:00Z","date_modified":"2026-10-26T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.35845.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5d54472043e75dc3","title":"MatAnyone-E: Sparse Memory Propagation for Efficient High-Resolution Video Matting","text":"MatAnyone-E accelerates high-resolution video matting by selectively applying high-resolution matching to uncertain regions and using a lightweight decoder, achieving large speedups with preserved quality.","text_kind":"content","url":"https://huggingface.co/papers/2610.35724","date_published":"2026-10-26T00:00:00Z","date_modified":"2026-10-26T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.35724/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5cecdf24f2f3f5c2","title":"DynaFlow: Learning Free-Form Multi-Agent Orchestration","text":"DynaFlow enables adaptive multi-agent collaboration by letting a main agent synthesize executable workflows that coordinate sub-agents through structured execution and verification.","text_kind":"content","url":"https://huggingface.co/papers/2610.34467","date_published":"2026-10-25T00:00:00Z","date_modified":"2026-10-25T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.34467.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7f4d292535fa3429","title":"InteractiveWorld: A Verifiable Human-Centric Environment for Building Better Interactive MLLM Assistants","text":"InteractiveWorld provides a verifiable simulation environment and benchmark to evaluate whether multimodal language models can reliably guide humans through physical tasks.","text_kind":"content","url":"https://huggingface.co/papers/2610.33966","date_published":"2026-10-25T00:00:00Z","date_modified":"2026-10-25T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.33966/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5a80235ed7913f13","title":"Build, Don't Play: Self-Improving LLM Research Agents Develop a Hearthstone Battlegrounds AI That Rivals Top Players","text":"AutoTavern demonstrates that self-improving LLM research agents can autonomously build a faithful game simulator and train a reinforcement learning agent that surpasses top human players in Hearthstone Battlegrounds.","text_kind":"content","url":"https://huggingface.co/papers/2610.33845","date_published":"2026-10-24T00:00:00Z","date_modified":"2026-10-24T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.33845.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d5dfb55b012918e7","title":"LitPivot: Developing research ideas through dynamic contextualization within the literature landscape","text":"LitPivot: Developing research ideas through dynamic contextualization within the literature landscape","text_kind":"content","url":"https://huggingface.co/blog/allenai/litpivot","date_published":"2026-10-23T16:15:50Z","date_modified":"2026-10-23T16:15:50Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/7F9q8BZloWigwe7m5YGLG.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b1121fffb5263882","title":"RSIHub: A Composable Framework for Agent-Evolution Methods","text":"RSIHub is an open-source framework that standardizes agent-evolution research by providing shared infrastructure for building, comparing, and extending methods with reusable components and consistent evaluation.","text_kind":"content","url":"https://huggingface.co/papers/2610.32253","date_published":"2026-10-23T00:00:00Z","date_modified":"2026-10-23T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.32253.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ed692854d218c9ab","title":"WorldCast: World-Consistent Character Video Generation with Large-Angle Camera Control","text":"CLACT and WorldCast enable consistent character video generation under large-angle camera motions by combining 3D point-cloud guidance with geometry-aligned memory attention.","text_kind":"content","url":"https://huggingface.co/papers/2610.33135","date_published":"2026-10-23T00:00:00Z","date_modified":"2026-10-23T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.33135/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a44fc5b5ba32a003","title":"World-in-Context: Interactive Video World Modeling with In-Context Conditioning","text":"World-in-Context unifies view, action, and camera controls via explicit tokens and interaction-aware positional encoding to improve interactive video generation and navigation.","text_kind":"content","url":"https://huggingface.co/papers/2610.33169","date_published":"2026-10-23T00:00:00Z","date_modified":"2026-10-23T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.33169/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/6430909168b58685","title":"Does Your Model Run on AWS Trainium? Look It Up","text":"Does Your Model Run on AWS Trainium? Look It Up","text_kind":"content","url":"https://huggingface.co/blog/torch-neuronx-model-support","date_published":"2026-10-23T00:00:00Z","date_modified":"2026-10-23T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/torch-neuronx-model-support/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/95eb6975ad245049","title":"Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox","text":"Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox","text_kind":"content","url":"https://huggingface.co/blog/microsoft/thinkingbox-training","date_published":"2026-10-22T22:24:34Z","date_modified":"2026-10-22T22:24:34Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/64b8491203124195cd795cad/oS8I8qx7iBS8EcNIXR3q9.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/eaa0ebff7daf626e","title":"OlmoEarth v1.3: State-of-the-art embeddings for the whole planet","text":"OlmoEarth v1.3: State-of-the-art embeddings for the whole planet","text_kind":"content","url":"https://huggingface.co/blog/allenai/olmoearth-v1-3","date_published":"2026-10-22T16:11:31Z","date_modified":"2026-10-22T16:11:31Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/VEmSzlLjei8J3cUVNe7G6.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5d118fadbe793ddd","title":"Ensemble forecasting with IBM Granite time-series models","text":"Ensemble forecasting with IBM Granite time-series models","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/granite-timeseries-ensemble-r1","date_published":"2026-10-22T07:51:31Z","date_modified":"2026-10-22T07:51:31Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6806091ee92bd22fd8ce9719/P6P8ZzHR8yL7GOOO4hQOg.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8e0d21507b4a69d3","title":"F2G-Pose: Geometry-Aware Foundation Feature Lifting for Direct RGB-D Category-Level Object Pose Estimation","text":"F2G-Pose is a real-time feed-forward framework that predicts category-level object pose, metric size, and completed shape from a single RGB-D image using dense visual features, DGCNN tokenization, and a Mixture-of-Experts Transformer.","text_kind":"content","url":"https://huggingface.co/papers/2610.31623","date_published":"2026-10-22T00:00:00Z","date_modified":"2026-10-22T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.31623/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e0ece036cc7aae37","title":"WeVisOne-VL: Unified Visual Perception from Objects to Pixels","text":"WeVisOne-VL unifies object-level and pixel-level visual perception via structured token prediction, supporting detection, localization, counting, OCR, segmentation, and matting with a lightweight pixel decoder and multi-stage training.","text_kind":"content","url":"https://huggingface.co/papers/2610.31329","date_published":"2026-10-22T00:00:00Z","date_modified":"2026-10-22T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.31329.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/92c7296d5073128a","title":"Every Weight Is a Path of Shared Discrete Decisions","text":"TWIG represents neural network weights as shared discrete decision paths, unifying compression methods and enabling efficient fixed-weight circuits with optimized discrete choices.","text_kind":"content","url":"https://huggingface.co/papers/2610.31931","date_published":"2026-10-22T00:00:00Z","date_modified":"2026-10-22T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.31931.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/65e403ba412b029a","title":"HiP3R: Hierarchical Pose Graphs for Long-Horizon 3D Reconstruction","text":"HiP3R uses a hierarchical pose graph to merge local video reconstructions into globally consistent camera trajectories and 3D geometry without retraining.","text_kind":"content","url":"https://huggingface.co/papers/2610.31635","date_published":"2026-10-22T00:00:00Z","date_modified":"2026-10-22T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.31635/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7104bc51c29585cc","title":"NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required","text":"NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required","text_kind":"content","url":"https://huggingface.co/blog/nvidia/kumo-relational","date_published":"2026-10-21T18:26:30Z","date_modified":"2026-10-21T18:26:30Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ad8e88370f69967fafc8703/i-Flass3NMPRmc6lVfwO_.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/354f89a6c7dbf10c","title":"How Specializing the Vocabulary of SLMs Can Cut Inference Cost by 77% in Real Applications","text":"How Specializing the Vocabulary of SLMs Can Cut Inference Cost by 77% in Real Applications","text_kind":"content","url":"https://huggingface.co/blog/Dharma-AI/vocabx-vocabulary-specialization","date_published":"2026-10-21T17:35:36Z","date_modified":"2026-10-21T17:35:36Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6a725daa5ca641a21ab61956/OWGtXjnbBov-7SI_04o4z.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f9b499debdda31c0","title":"Profiling TorchNeuron (Part 1): System-level profiling on AWS Trainium.","text":"Profiling TorchNeuron (Part 1): System-level profiling on AWS Trainium.","text_kind":"content","url":"https://huggingface.co/blog/amazon/profiling-torchneuron-part-1","date_published":"2026-10-21T16:04:43Z","date_modified":"2026-10-21T16:04:43Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6ac67ea29579ea34f1f7e470/hEodPVvQZI24jETLbPooP.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9b35c0bfb2f0e3cd","title":"Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster","text":"Granite Speech 5.0: Transcribe, Translate, and Attribute more Accurately and Faster","text_kind":"content","url":"https://huggingface.co/blog/ibm-granite/granite-speech-5","date_published":"2026-10-21T14:58:18Z","date_modified":"2026-10-21T14:58:18Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/653fb52cbfd890805461d691/M_JJyIOBOWVzHXdy7CVVG.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/38715d90bd30f009","title":"NVIDIA Kumo Forecast Combines Pretrained Time-Series Representations with Historical Pattern Retrieval and Interpretability","text":"NVIDIA Kumo Forecast Combines Pretrained Time-Series Representations with Historical Pattern Retrieval and Interpretability","text_kind":"content","url":"https://huggingface.co/blog/nvidia/kumo-forecast","date_published":"2026-10-21T12:51:24Z","date_modified":"2026-10-21T12:51:24Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/660f5683b710b57b115a5483/pFl0vhvFrc-oZjk5qE8_0.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0f7eabbb925dfe6c","title":"Looking Ahead: Modeling Future Evolution in Autoregressive Video Generation","text":"Future Chunk Prediction improves autoregressive video diffusion by enforcing future-chunk supervision, yielding more physically plausible video generation with reduced training cost.","text_kind":"content","url":"https://huggingface.co/papers/2610.29989","date_published":"2026-10-21T00:00:00Z","date_modified":"2026-10-21T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.29989/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cead82cd3e764df0","title":"DONIC: Generalist Dexterous Bimanual Tracking Control from Human Demonstrations","text":"DONIC is a generalist bimanual tracker trained on diverse human-motion data that uses reinforcement learning with recycled demonstrations and per-hand rewards to achieve high success on unseen objects without fine-tuning.","text_kind":"content","url":"https://huggingface.co/papers/2610.29997","date_published":"2026-10-21T00:00:00Z","date_modified":"2026-10-21T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.29997/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2425e3171bde3e91","title":"Building Auditable AI Text Detectors with Sparse Autoencoders","text":"A pipeline using sparse autoencoders extracts interpretable linguistic features from neural detectors, yielding more accurate and explainable AI-generated text classification.","text_kind":"content","url":"https://huggingface.co/papers/2610.30006","date_published":"2026-10-21T00:00:00Z","date_modified":"2026-10-21T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.30006/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4eb9f7080b11166e","title":"NVIDIA Cosmos3-H-Dreams: Bringing Clinical Surgery to Life Through Generative Simulation","text":"NVIDIA Cosmos3-H-Dreams: Bringing Clinical Surgery to Life Through Generative Simulation","text_kind":"content","url":"https://huggingface.co/blog/nvidia/cosmos3-h-dreams","date_published":"2026-10-20T18:17:47Z","date_modified":"2026-10-20T18:17:47Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/67a4f15057fb66b9db6dac82/45XSEtdQHqTdspRe3wkYs.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ca293d8533a3487d","title":"FairStress: A deeper evaluation of fairness in LLM decision-making","text":"FairStress: A deeper evaluation of fairness in LLM decision-making","text_kind":"content","url":"https://huggingface.co/blog/allenai/fairstress","date_published":"2026-10-20T16:09:19Z","date_modified":"2026-10-20T16:09:19Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/7zYjyTWswjO2erWsO5HNf.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/94e7f4f5c0e52bb7","title":"ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model","text":"ALTK-Evolve: How learned memory tops the τ²-bench leaderboard on customer service tasks without a frontier model","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/altk-evolve-bc","date_published":"2026-10-20T15:19:48Z","date_modified":"2026-10-20T15:19:48Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6435a1131860001f144239ea/rzXYJh7e7UdoYgCLwOZjH.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/8959d75a23168a1d","title":"SolarRenderer: Few-Step Generative Reference Rendering","text":"SolarRenderer adapts a frozen video backbone with lightweight LoRAs to render realistic videos from diverse structural inputs in four steps or real-time streaming, using frame-aligned training data.","text_kind":"content","url":"https://huggingface.co/papers/2610.28630","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28630/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c3c375455b4762ce","title":"EvaOmni: An Evidence-Grounded Benchmark for Audio-Visual Reasoning","text":"EvaOmni is an evidence-grounded benchmark that evaluates whether omnimodal language models identify audio-visual evidence and construct reasoning chains for audiovisual questions.","text_kind":"content","url":"https://huggingface.co/papers/2610.28444","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28444.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e1a0dd5580ed4f25","title":"DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems","text":"DecisionPivot is a diagnostic benchmark that evaluates decision models across controlled variations, revealing gaps between aggregate accuracy, contrast-group performance, calibration, and reasoning effort.","text_kind":"content","url":"https://huggingface.co/papers/2610.28522","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28522.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c931af3845e23367","title":"Cache Engineering for Agents","text":"AKV introduces cache engineering to manage LLM agent key-value caches without altering trajectories, improving efficiency and accuracy via drop and repos operations with a co-designed serving engine.","text_kind":"content","url":"https://huggingface.co/papers/2610.28577","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28577/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/da99ff5289ab6c3e","title":"Scaling in Games: Continued Pre-Training For Embodied Agents In Diverse Virtual Worlds","text":"Continued pre-training across data scale, diversity, and model size improves embodied learning in virtual worlds but reveals limits in distant-world generalization, motivating richer training beyond imitation.","text_kind":"content","url":"https://huggingface.co/papers/2610.28698","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28698/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/44512f6af4ef55d2","title":"StreamEval: Evaluating Interactive Streaming Video Generation Models","text":"StreamEval benchmarks interactive streaming video generation across five tasks, revealing that reliable request execution and sustained state consistency remain major challenges despite high visual quality.","text_kind":"content","url":"https://huggingface.co/papers/2610.28402","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.28402.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3eb18e5ff7f2a0ef","title":"MPS in PyTorch 2.15","text":"MPS in PyTorch 2.15","text_kind":"content","url":"https://huggingface.co/blog/mps-pytorch-2-15","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/mps-pytorch-2-15/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/84cd8d44e9825e73","title":"Hugging Face ❤️ TorchTPU: TPUs, Natively","text":"Hugging Face ❤️ TorchTPU: TPUs, Natively","text_kind":"content","url":"https://huggingface.co/blog/hf-torchtpu","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/hf-torchtpu/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2596ff929aeef557","title":"AnyLanguageModel 1.0: One Year of Local Models on Apple Platforms","text":"AnyLanguageModel 1.0: One Year of Local Models on Apple Platforms","text_kind":"content","url":"https://huggingface.co/blog/anylanguagemodel-1-0","date_published":"2026-10-20T00:00:00Z","date_modified":"2026-10-20T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/anylanguagemodel-1-0/banner.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/83434cc1f1600ba2","title":"OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory","text":"OpenVLHarness improves multimodal reasoning by integrating persistent memory, heterogeneous tool interfaces, and iterative evidence composition across 22 benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2610.26860","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.26860/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e7fbe55a81d2edce","title":"Manga2Animate: An Agentic Framework for Faithful Manga Animation with Script and Storyboard Planning","text":"Manga2Animate uses paired creator and reviewer agents to build script and storyboard representations from manga panels, yielding more faithful and consistent animations.","text_kind":"content","url":"https://huggingface.co/papers/2610.26851","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.26851/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/53cb429bf0a3a4f1","title":"ScaleNav: Scalable Internet Video Pretraining for Language-Conditioned Local Navigation","text":"ScaleNav learns goal-directed visual dynamics from internet egocentric videos to generate robot actions for language-conditioned local navigation, achieving high success rates in simulation and real-world deployment.","text_kind":"content","url":"https://huggingface.co/papers/2610.25690","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.25690.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cbdee8b20b18bd62","title":"InfiniDream: Open-Ended Photorealistic 3D Worlds Through Procedural Generation","text":"InfiniDream is a vision-language model framework that generates open-ended, editable, photorealistic 3D worlds through symbolic procedural asset creation and deterministic scene placement.","text_kind":"content","url":"https://huggingface.co/papers/2610.26853","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.26853/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/dd65d743e8de1cb2","title":"MemT-1.0: Towards Memory-Native Evolution","text":"MemT-1.0 introduces a memory-native architecture that embeds continuous memory into foundation model computation, improving long-context efficiency and agentic task performance.","text_kind":"content","url":"https://huggingface.co/papers/2610.26825","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.26825.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/098ea7e11367a537","title":"Agent World Model in the Wild: Computer-Use Agent Environment Synthesis","text":"WildAWM synthesizes controllable computer-use environments and tasks to train general agents via interaction, improving out-of-distribution performance across web, CLI, and GUI benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2610.26828","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.26828/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9dbc0d55ff9123c6","title":"TorchNeuron: running HF models on AWS Trainium is easier than ever","text":"TorchNeuron: running HF models on AWS Trainium is easier than ever","text_kind":"content","url":"https://huggingface.co/blog/hf-torchneuron","date_published":"2026-10-19T00:00:00Z","date_modified":"2026-10-19T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/hf-torchneuron/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a7ca3e22bd439a68","title":"MT Forcing: Towards Stable and Consistent Multi-Turn Image Editing with Autoregressive Diffusion","text":"MT Forcing improves multi-turn image editing by combining single-turn distribution matching distillation and teacher-forcing guidance to stabilize visual quality and preserve cross-turn consistency.","text_kind":"content","url":"https://huggingface.co/papers/2610.25404","date_published":"2026-10-18T00:00:00Z","date_modified":"2026-10-18T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.25404.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/16b67ec01772fe59","title":"UniAVC: Unified Audio-Video Creation via Progressive Training and Asymmetric Scheduling","text":"UniAVC trains separate audio and video branches on abundant unimodal data before aligning them with scarce paired data, using independent noise schedules and asymmetric supervision to enable joint synthesis, cross-modal generation, and instruction-guided editing.","text_kind":"content","url":"https://huggingface.co/papers/2610.24791","date_published":"2026-10-17T00:00:00Z","date_modified":"2026-10-17T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.24791/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/ba4e7337aba280a4","title":"Scaling Indic Pretraining: 2.8 Trillion Tokens and an Open Data Generation Pipeline","text":"Scaling Indic Pretraining: 2.8 Trillion Tokens and an Open Data Generation Pipeline","text_kind":"content","url":"https://huggingface.co/blog/nvidia/scaling-indic-pretraining","date_published":"2026-10-16T01:44:47Z","date_modified":"2026-10-16T01:44:47Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/68d2fec8856b85d927e44d32/x0aObe_0mduBjWFuKBGoV.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/6ca88a850d32d296","title":"DiscoveryRSI: Recursively Learning to Discover Valuable Hypotheses","text":"DiscoveryRSI uses multi-agent evolutionary search and recursive learning to generate testable scientific hypotheses, improving quality through self-improvement and outperforming baseline agents on life-science benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2610.23839","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.23839.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4ae13348e1e9ea11","title":"MOSS-Image Technical Report","text":"MOSS-Image is an open diffusion transformer that combines high-fidelity synthesis with intent-aware prompt enhancement to generate text-rich, layout-coherent images, achieving state-of-the-art open-weight results on long-form text and layout benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2610.23374","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.23374.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f4e9ac818ba1ac32","title":"DLite: History-free Speculative Decoding for Agentic AI","text":"DLite improves speculative decoding for agentic AI by eliminating historical context maintenance and using target hidden states for local predictions, boosting acceptance length and reducing overhead.","text_kind":"content","url":"https://huggingface.co/papers/2610.23787","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.23787/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c651cfdb72fb8c0f","title":"Xiaomi-Lens-0 Technical Report","text":"Xiaomi-Lens-0 is a 4B multimodal model that predicts masks from internal visual representations to improve fine-grained region reasoning and referring comprehension.","text_kind":"content","url":"https://huggingface.co/papers/2610.23279","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.23279/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/74b1441ca9345c04","title":"OR-License: Evaluating LLM Agents as Real-World Operations Research Practitioners","text":"OR-License evaluates whether LLM agents can solve real-world operations research problems from original client materials, finding that while some achieve high objective quality, none match expert-level feasibility and deployment-adjusted quality.","text_kind":"content","url":"https://huggingface.co/papers/2610.24087","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.24087.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5fbf07b61746ebcb","title":"RoboHarness: A Simple Harness Outperforms VLA and World Action Models","text":"RoboHarness enables LLM agents to perform embodied manipulation via a visual-geometric control interface, surpassing specialized vision-language-action models on robotics benchmarks.","text_kind":"content","url":"https://huggingface.co/papers/2610.14018","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.14018/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/849b423196d125cc","title":"Welcome Beam: a 501B open model you can own, customize, and scale","text":"Welcome Beam: a 501B open model you can own, customize, and scale","text_kind":"content","url":"https://huggingface.co/blog/reflection-beam","date_published":"2026-10-16T00:00:00Z","date_modified":"2026-10-16T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/reflection-beam/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a01e02239da0a38e","title":"Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint","text":"Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint","text_kind":"content","url":"https://huggingface.co/blog/ibm-granite/granite-switch-4-1","date_published":"2026-10-15T20:34:01Z","date_modified":"2026-10-15T20:34:01Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/662836b742709fe826510597/gltB0o--PV_Jh0D3aDm8-.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e3c36d4422c6129b","title":"Apertus 1.5 Release Notes","text":"Apertus 1.5 Release Notes","text_kind":"content","url":"https://huggingface.co/blog/swiss-ai/apertus-v15-release-notes","date_published":"2026-10-15T09:08:45Z","date_modified":"2026-10-15T09:08:45Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/61f1877a168a0b52fda88455/2mHXxCtfos6CgwFSNKFVi.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3a5586214dd0cb9f","title":"3DCodeVerse: Exploring 3D Worlds as Code","text":"3DCodeVerse is a research ecosystem that provides datasets, automated verification harnesses, and empirical guidance to enable scalable, editable, and verifiable 3D content generation via executable code.","text_kind":"content","url":"https://huggingface.co/papers/2610.22341","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22341/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7c3a547ab4228c00","title":"OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use","text":"OpenGCUA introduces an open generalist computer-use agent that integrates visual and programmatic interfaces, harness-to-model decision transfer, and code execution to improve task performance and efficiency across models.","text_kind":"content","url":"https://huggingface.co/papers/2610.22200","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22200/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f09ee7ca5e3baa5d","title":"Mosaic: Multi-Session Agentic Reinforcement Learning","text":"Multi-session agentic reinforcement learning is introduced to handle trajectories across multiple sessions, with Mosaic optimizing session-level contributions and allocation via hierarchical advantage estimation.","text_kind":"content","url":"https://huggingface.co/papers/2610.22174","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22174/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/99f99237cc4a2abd","title":"Has GPT-6-Astra Achieved Embodied Intelligence?","text":"GPT-6-Astra achieves high performance across diverse robotic tasks via in-context learning, yet real-world deployment reveals gaps in demonstration cost, execution time, and practical embodied intelligence.","text_kind":"content","url":"https://huggingface.co/papers/2610.22471","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22471/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/7f7f22ced61aa2cb","title":"SUPRA-TEX: 3D Generation with Super-Aligned Texture","text":"SUPRA-Tex decouples visible-material estimation, surface alignment, and 3D completion to generate relightable 3D materials that preserve reference image details and complete unobserved surfaces.","text_kind":"content","url":"https://huggingface.co/papers/2610.22352","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22352/gradient.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/45fea1ed3d41279a","title":"TokenEconomizer: Experience-Driven Black-Box Cost Control for Coding Agents","text":"TokenEconomizer reduces recurring token costs of coding agents across sequential real-world tasks by pruning irrelevant tools, injecting micro-skills, and interrupting failure loops.","text_kind":"content","url":"https://huggingface.co/papers/2610.22732","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.22732.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/3733c75a28bf26df","title":"Welcome FLUX.3 Image Klein in Diffusers","text":"Welcome FLUX.3 Image Klein in Diffusers","text_kind":"content","url":"https://huggingface.co/blog/flux-3","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/flux3/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9d251ac6836a600d","title":"Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta","text":"Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta","text_kind":"content","url":"https://huggingface.co/blog/meta-muse-spark-open","date_published":"2026-10-15T00:00:00Z","date_modified":"2026-10-15T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/meta-muse-spark-open/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/672ebf75c430cf70","title":"HiRO-ACEv2: Simulating extreme weather at 3 km resolution for a fraction of the cost","text":"HiRO-ACEv2: Simulating extreme weather at 3 km resolution for a fraction of the cost","text_kind":"content","url":"https://huggingface.co/blog/allenai/hiro-acev2","date_published":"2026-10-14T16:12:42Z","date_modified":"2026-10-14T16:12:42Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/Y8t8GnfI2lUa40iX4_sxA.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f6d68cd73a5aaf2a","title":"What Generation Knows: Unlocking Fine-Grained Understanding in Unified Multimodal Models","text":"GAFU leverages native generation-side representations to improve fine-grained visual understanding in unified multimodal models by compressing them into compact tokens for task-conditioned retrieval.","text_kind":"content","url":"https://huggingface.co/papers/2610.23987","date_published":"2026-10-14T00:00:00Z","date_modified":"2026-10-14T00:00:00Z","authors":["Hugging Face Trending Papers"],"tags":["Hugging Face Trending Papers"],"image":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2610.23987.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/550c4c02e9ff21a9","title":"Run fast, private agents on-device with LFM2.5-24B-A2B","text":"Run fast, private agents on-device with LFM2.5-24B-A2B","text_kind":"content","url":"https://huggingface.co/blog/LiquidAI/lfm25-24b-a2b","date_published":"2026-10-13T17:05:48Z","date_modified":"2026-10-13T17:05:48Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/644249b08443bce4c9890a0f/YxGx9U6yV6_mumklCyMsE.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/43ef54e74d276082","title":"Impactful scheduling for GPU clusters","text":"Impactful scheduling for GPU clusters","text_kind":"content","url":"https://huggingface.co/blog/allenai/impactful-scheduling","date_published":"2026-10-09T15:20:29Z","date_modified":"2026-10-09T15:20:29Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/Kj0n5ea3ItoNIc2ha2SVL.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0f05fd4155a46f47","title":"The model that didn't exist, so you made it yourself","text":"The model that didn't exist, so you made it yourself","text_kind":"content","url":"https://huggingface.co/blog/building-with-ml-intern","date_published":"2026-10-08T00:00:00Z","date_modified":"2026-10-08T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/building-with-ml-intern/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0825a4897dea2e82","title":"Multimodal open d1 decision models for the edge","text":"Multimodal open d1 decision models for the edge","text_kind":"content","url":"https://huggingface.co/blog/LiquidAI/open-d1","date_published":"2026-10-07T16:54:33Z","date_modified":"2026-10-07T16:54:33Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/644249b08443bce4c9890a0f/hLsllw4zhC04cVGiV69v0.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/91ed3eb235f3237f","title":"Introducing Falcon ASR","text":"Introducing Falcon ASR","text_kind":"content","url":"https://huggingface.co/blog/tiiuae/falcon-asr","date_published":"2026-10-07T13:21:03Z","date_modified":"2026-10-07T13:21:03Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/671a3e44cf97dc64441e170a/bJ2aGPIX6WLWagE10j5y3.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5bdd694c62903534","title":"One Model Family, Two Gold-Level Results: Fine-Tuning Nemotron for IOI and IMO","text":"One Model Family, Two Gold-Level Results: Fine-Tuning Nemotron for IOI and IMO","text_kind":"content","url":"https://huggingface.co/blog/nvidia/nemotron-ioi-and-imo-2026","date_published":"2026-10-07T12:45:31Z","date_modified":"2026-10-07T12:45:31Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/67b8b0096c3182e96bf6cea1/zB6BNaB-3tiCCsnOo7gik.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/085f7195b66fbb80","title":"Falcon-Emirati: When an LLM Learns the Dialect, the Culture, and the Nuance","text":"Falcon-Emirati: When an LLM Learns the Dialect, the Culture, and the Nuance","text_kind":"content","url":"https://huggingface.co/blog/tiiuae/falcon-emirati","date_published":"2026-10-06T06:44:39Z","date_modified":"2026-10-06T06:44:39Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/659bc8a7b0f43ed69f0b2300/MJHJMYW3O4CkLvXvn7DOp.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/89764eae00994ca8","title":"The Agent Said It Was Done. The Database Disagreed.","text":"The Agent Said It Was Done. The Database Disagreed.","text_kind":"content","url":"https://huggingface.co/blog/microsoft/thinkingbox","date_published":"2026-10-03T22:56:48Z","date_modified":"2026-10-03T22:56:48Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/64b8491203124195cd795cad/NStwJm1AafDELVsU4KGwS.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d599f0a282be129f","title":"Open-sourcing AstaBrief, the fast report-generation model in Asta","text":"Open-sourcing AstaBrief, the fast report-generation model in Asta","text_kind":"content","url":"https://huggingface.co/blog/allenai/astabrief","date_published":"2026-10-02T15:19:50Z","date_modified":"2026-10-02T15:19:50Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/AdqLEBgtQnaDXKLdylC3l.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a885ed61f54a415a","title":"AutoSynthData: Generating Training Data for Enterprise Agents","text":"AutoSynthData: Generating Training Data for Enterprise Agents","text_kind":"content","url":"https://huggingface.co/blog/ServiceNow-AI/autosynthdata","date_published":"2026-10-02T04:01:31Z","date_modified":"2026-10-02T04:01:31Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/68642da9885da181fde1ad70/8kSzlHLOdu5hchCtPnK8D.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/0bf00b70dbd762e2","title":"Introducing Olmo-core 3: Open, scalable training infrastructure for large MoEs","text":"Introducing Olmo-core 3: Open, scalable training infrastructure for large MoEs","text_kind":"content","url":"https://huggingface.co/blog/allenai/olmocore3","date_published":"2026-10-01T15:01:43Z","date_modified":"2026-10-01T15:01:43Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/uKnK93gjkKbmJmSx94WO2.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/581ab3112c302ffe","title":"Open TTS Leaderboard: Scalable Evaluation for Multilingual Text-to-Speech and Voice Cloning","text":"Open TTS Leaderboard: Scalable Evaluation for Multilingual Text-to-Speech and Voice Cloning","text_kind":"content","url":"https://huggingface.co/blog/open-tts-leaderboard","date_published":"2026-09-30T00:00:00Z","date_modified":"2026-09-30T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/open-tts-leaderboard/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/493cf3fce1b87919","title":"NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction","text":"NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction","text_kind":"content","url":"https://huggingface.co/blog/nvidia/kumo-tabular","date_published":"2026-09-29T15:30:38Z","date_modified":"2026-09-29T15:30:38Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/684040a5de1ad7f4fcec9508/HyVlq-3d7EVj_M-m_TfHJ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/984223f109f3360b","title":"Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents","text":"Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents","text_kind":"content","url":"https://huggingface.co/blog/MultiverseComputingCAI/getting-the-source-right-not-just-the-fact-source","date_published":"2026-09-29T13:07:00Z","date_modified":"2026-09-29T13:07:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/668e37fd9c9aa124a3c867e8/phhhj_BLOJbvnTFhVaZ5y.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/9552195c964ddb7b","title":"Holo4: powering generalist computer-use agents","text":"Holo4: powering generalist computer-use agents","text_kind":"content","url":"https://huggingface.co/blog/Hcompany/holo4","date_published":"2026-09-28T09:44:05Z","date_modified":"2026-09-28T09:44:05Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69fc7e49052f11ab9931b672/yzjxvlmlZKtjASwAYu6bN.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a30a6af008def6d3","title":"Welcome RL Environments to the hub","text":"Welcome RL Environments to the hub","text_kind":"content","url":"https://huggingface.co/blog/rl-environments","date_published":"2026-09-28T00:00:00Z","date_modified":"2026-09-28T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/datasets-filters/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c2a53f2332725b78","title":"Accelerating vision-language models with LFM2.5-VL-DSpark","text":"Accelerating vision-language models with LFM2.5-VL-DSpark","text_kind":"content","url":"https://huggingface.co/blog/LiquidAI/lfm2-5-vl-dspark","date_published":"2026-09-24T14:08:57Z","date_modified":"2026-09-24T14:08:57Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/644249b08443bce4c9890a0f/WgfT8N8Xyb6lBxSif7XLO.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a70f8ac64edfa547","title":"How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows","text":"How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows","text_kind":"content","url":"https://huggingface.co/blog/nvidia/how-to-use-nvidia-warp-and-mjwarp","date_published":"2026-09-23T18:41:40Z","date_modified":"2026-09-23T18:41:40Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6994dc99f850a10f03fd0b21/rQ6tGCJEaH16bQQ8X4M8b.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/a4dec25709f445f4","title":"**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**","text":"**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**","text_kind":"content","url":"https://huggingface.co/blog/nvidia/nemotron-diarization","date_published":"2026-09-23T13:17:01Z","date_modified":"2026-09-23T13:17:01Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/688d4bdfdeb55432d90e546d/d_xBvPVTwTB_Rt9yxmp_0.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/98acfe91fd30b730","title":"Transformers now runs llama.cpp quants","text":"Transformers now runs llama.cpp quants","text_kind":"content","url":"https://huggingface.co/blog/transformers-llama-cpp-quants","date_published":"2026-09-22T00:00:00Z","date_modified":"2026-09-22T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/transformers_llama_cpp_quants/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/fb8ad85c459c0cb6","title":"Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community","text":"Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community","text_kind":"content","url":"https://huggingface.co/blog/omlx","date_published":"2026-09-22T00:00:00Z","date_modified":"2026-09-22T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/omlx/omlx-hf.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2960c64a019e48fb","title":"How UK AISI and EvalEval Are Making Benchmark Results Reproducible","text":"How UK AISI and EvalEval Are Making Benchmark Results Reproducible","text_kind":"content","url":"https://huggingface.co/blog/evaleval-aisi","date_published":"2026-09-22T00:00:00Z","date_modified":"2026-09-22T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/evaleval-aisi/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c701b57089c5bf6a","title":"Pruning LLMs Like a Physicist: Block Removal as an Ising Optimization Problem","text":"Pruning LLMs Like a Physicist: Block Removal as an Ising Optimization Problem","text_kind":"content","url":"https://huggingface.co/blog/MultiverseComputingCAI/pruning-llms-like-a-physicist-block-removal-as-an","date_published":"2026-09-21T13:44:34Z","date_modified":"2026-09-21T13:44:34Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/693c2a4eb0871ba57155b4ed/q4gZV9ENtfFwge7H3Q7YJ.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/eb792e0fc3420085","title":"tokenizers v1: encode, decode and scaling, measured","text":"tokenizers v1: encode, decode and scaling, measured","text_kind":"content","url":"https://huggingface.co/blog/tokenizers-v1","date_published":"2026-09-21T00:00:00Z","date_modified":"2026-09-21T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/tokenizers-v1/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/16a7b27ed0ae9a75","title":"Your Agent Aced the Task. Will It Do It Again?","text":"Your Agent Aced the Task. Will It Do It Again?","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/altk-evolve-consistency","date_published":"2026-09-15T16:00:44Z","date_modified":"2026-09-15T16:00:44Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6435a1131860001f144239ea/dI5J2sSc3TSprk9VB4EVJ.jpeg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/cb50b0c3c8564548","title":"Rebuilding AUTOMATIC1111 with Gradio Workflow","text":"Rebuilding AUTOMATIC1111 with Gradio Workflow","text_kind":"content","url":"https://huggingface.co/blog/gradio-workflow-1111","date_published":"2026-09-10T00:00:00Z","date_modified":"2026-09-10T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/gradio-workflow1111/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/d4e55a9164f4783e","title":"Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL","text":"Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL","text_kind":"content","url":"https://huggingface.co/blog/asyncgrpo-lora-hfjobs","date_published":"2026-09-10T00:00:00Z","date_modified":"2026-09-10T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/asyncgrpo-lora-hfjobs/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/e8608e40b7228951","title":"IBM releases SOTA Granite Time Series PatchTST-FM-r2 model with commercial-friendly license","text":"IBM releases SOTA Granite Time Series PatchTST-FM-r2 model with commercial-friendly license","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/ibm-releases-sota-granite-time-series","date_published":"2026-09-09T15:36:24Z","date_modified":"2026-09-09T15:36:24Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/69d3d41eef229c09afea5d83/KYse3pX6t3l8FnI-1pKsi.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/4f420d1c5cd0cb94","title":"Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic","text":"Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic","text_kind":"content","url":"https://huggingface.co/blog/MultiverseComputingCAI/safety-for-whom","date_published":"2026-09-08T14:23:07Z","date_modified":"2026-09-08T14:23:07Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/668e37fd9c9aa124a3c867e8/HuauFbdznYW4fNQh8j4Tp.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/96a3500c5a4b2fb7","title":"NeoMME: an efficient Multimodal-native and Multilingual Encoder","text":"NeoMME: an efficient Multimodal-native and Multilingual Encoder","text_kind":"content","url":"https://huggingface.co/blog/Hcompany/neomme","date_published":"2026-09-03T13:13:48Z","date_modified":"2026-09-03T13:13:48Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6264f9655f6f2e14d6ac981c/GJ6FUbgFpq1x8RNOqzmz-.webp"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/45020768fad998be","title":"Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps","text":"Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps","text_kind":"content","url":"https://huggingface.co/blog/grpo-with-trl-ifstruct","date_published":"2026-09-03T00:00:00Z","date_modified":"2026-09-03T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/grpo-with-trl-ifstruct/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/86b47b47cb7422fd","title":"Give Your Coding Agents a Memory You Own","text":"Give Your Coding Agents a Memory You Own","text_kind":"content","url":"https://huggingface.co/blog/funes","date_published":"2026-09-03T00:00:00Z","date_modified":"2026-09-03T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/funes/thumbnail.jpg"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b844bdeeb825fb68","title":"Training a coding model to paint watercolours with TRL and OpenEnv","text":"Training a coding model to paint watercolours with TRL and OpenEnv","text_kind":"content","url":"https://huggingface.co/blog/train-to-paint-with-code","date_published":"2026-09-03T00:00:00Z","date_modified":"2026-09-03T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/train-to-paint-with-code/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/22915ca8a27338bf","title":"Real-Time Intelligence with IBM Time Series Models on Confluent","text":"Real-Time Intelligence with IBM Time Series Models on Confluent","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/real-time-intelligence","date_published":"2026-09-02T13:49:14Z","date_modified":"2026-09-02T13:49:14Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/64e8143f6de557454220921e/1oB5vm47VNu42diDOsp0p.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/23bdbfdbd061867c","title":"BenchMIRT: What are LLM benchmarks actually measuring?","text":"BenchMIRT: What are LLM benchmarks actually measuring?","text_kind":"content","url":"https://huggingface.co/blog/allenai/benchmirt","date_published":"2026-09-01T21:39:07Z","date_modified":"2026-09-01T21:39:07Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/638e39b249de7ae552d977b5/c_Rnu4DRj6Djxk1IT0gKu.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/85562dc16345cc8b","title":"Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI","text":"Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI","text_kind":"content","url":"https://huggingface.co/blog/webgpu-kernels","date_published":"2026-09-01T00:00:00Z","date_modified":"2026-09-01T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/webgpu-kernels/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/b0b2c2da12a5bc45","title":"The Open ASR Leaderboard Adds Its First Global South Language","text":"The Open ASR Leaderboard Adds Its First Global South Language","text_kind":"content","url":"https://huggingface.co/blog/open-asr-leaderboard-global-south","date_published":"2026-08-28T00:00:00Z","date_modified":"2026-08-28T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/open-asr-leaderboard-global-south/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5309d96554639301","title":"Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers","text":"Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers","text_kind":"content","url":"https://huggingface.co/blog/train-multi-vector-encoder","date_published":"2026-08-26T00:00:00Z","date_modified":"2026-08-26T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/train-sentence-transformers/st-hf-thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/74043a2c13e42e05","title":"Granite 4.2 LLMs: How They're Built","text":"Granite 4.2 LLMs: How They're Built","text_kind":"content","url":"https://huggingface.co/blog/ibm-granite/granite-4-2","date_published":"2026-08-25T15:14:14Z","date_modified":"2026-08-25T15:14:14Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/65cc2c288ebd392213e58899/XHWV2L_wZsFe3ekksCgvQ.webp"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/f6b26e55a4899220","title":"Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original","text":"Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original","text_kind":"content","url":"https://huggingface.co/blog/MultiverseComputingCAI/quantization-aware-healing","date_published":"2026-08-25T11:39:24Z","date_modified":"2026-08-25T11:39:24Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/668e37fd9c9aa124a3c867e8/2DSE26PMq2hJqKqEDOLIn.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/2983dbf79834f512","title":"Wire It, Run It, Deploy It: AI Workflows in Gradio","text":"Wire It, Run It, Deploy It: AI Workflows in Gradio","text_kind":"content","url":"https://huggingface.co/blog/gradio-workflow-guide","date_published":"2026-08-25T00:00:00Z","date_modified":"2026-08-25T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/gradio-workflow-guide/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/bcd3314d2ba2a36f","title":"How Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code","text":"How Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code","text_kind":"content","url":"https://huggingface.co/blog/pwc-search","date_published":"2026-08-21T00:00:00Z","date_modified":"2026-08-21T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/pwc-search/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/68e9fb1705e10808","title":"Measuring benchmark optimization in speech recognition","text":"Measuring benchmark optimization in speech recognition","text_kind":"content","url":"https://huggingface.co/blog/asr-benchmark-optimization","date_published":"2026-08-21T00:00:00Z","date_modified":"2026-08-21T00:00:00Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://huggingface.co/blog/assets/asr-benchmark-optimization/thumbnail.png"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/c7cdb6722fd86cc4","title":"Up to 3.2x Faster Inference with LFM2.5-DSpark","text":"Up to 3.2x Faster Inference with LFM2.5-DSpark","text_kind":"content","url":"https://huggingface.co/blog/LiquidAI/lfm25-dspark","date_published":"2026-08-20T16:52:57Z","date_modified":"2026-08-20T16:52:57Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/644249b08443bce4c9890a0f/ZMoThfdqzAz8cbxweVQfO.gif"},{"id":"tag:trvny.github.io,2024:feedseek/huggingface/5ce8103476b1f3ac","title":"How Much Memory Does Your Agent Actually Need?","text":"How Much Memory Does Your Agent Actually Need?","text_kind":"content","url":"https://huggingface.co/blog/ibm-research/altk-evolve-hmm","date_published":"2026-08-18T18:09:38Z","date_modified":"2026-08-18T18:09:38Z","authors":["Hugging Face Blog"],"tags":["Hugging Face Blog"],"image":"https://cdn-uploads.huggingface.co/production/uploads/6435a1131860001f144239ea/j-n1Au9SJ2-RGkT9i98u4.jpeg"}]}
