π» ReposΓ1 mentions
Mudabench
βMuDABench is available at https://github.com/Zhanli-Li/MuDABench.β
paper: MuDABench Exposes RAG Limitations in Large-Scale Multi-Document Analytical QA β 2026-04-27
What the smart people are recommending. 7861 books, tools, and products endorsed by the thinkers absorb.md tracks. Ranked by how many times each has been recommended across compiled podcasts, papers, posts, and tweets.
βMuDABench is available at https://github.com/Zhanli-Li/MuDABench.β
βAudio samples are available at https://qiangchunyu.github.io/UniSonate/.β
βIn this paper, we formulate routing as a budget allocation problem and identify marginal gain... we propose RouteLMT (routing for LLM-based MT), an efficient in-model router... Extensive experiments dβ¦β
βProject page: https://muzhancun.github.io/preprints/DROL.β
βTTS-PRISM is open-source, with code and checkpoints at https://github.com/xiaomi-research/tts-prism.β
βCode is available at https://github.com/shuowl/llm-source-balancing.β
βCode is available at https://github.com/BU-DEPEND-Lab/SpecRLBench.β
βOverall, H-optimus-1 achieves the strongest survival prediction performance.β
βNotably, the compact distilled model H0-mini slightly outperforms its larger teacher model H-optimus-0, despite using fewer than 8% of the parameters and enabling significantly faster feature extractiβ¦β
βThe platform is publicly available at Energy-Arena.org.β
βWe propose Hyperparameter-Divergent Ensemble Training (HDET), a method that repurposes these replicas for simultaneous learning rate exploration at negligible communication overhead.β
βHDET is implemented as a drop-in replacement for PyTorch's OneCycleLR scheduler, requiring no changes to model architecture, optimizer, or data pipeline.β
βOur code, benchmark, and model are released at https://shiyi-zh0408.github.io/projectpages/Meta-CoT/β
βThe system uses instruction-tuned qwen3-embedding-0.6B embeddings...Qwen3 embeddings with 300-token chunk size achieved 94.6% accuracy on a clinical question-answering benchmark.β
βOur code and data are available at https://github.com/MAPS-research/SHaPEβ
βDV-World provides a realistic testbed to steer development toward the versatile expertise required in enterprise workflows. Our data and code are available at this project page.β
βIts closure leaves behind non-updatable benchmarks, irreproducible results, and ultimately a field at risk of perpetuating these issues by turning to closed-source LLMs.β
βOxyGent is publicly available at https://oxygent.jd.com/β
βOur code and data are available at https://github.com/bigspinai/bigspin-fluency-outcomesβ
βThe resulting corpus comprises Β± 35 billion tokens across the medical domain in about 100 million documents, freely available on Hugging Face.β
βthe most widely adopted multilingual base (Chatterbox, 23 languages) does not even tokenise Telugu or Tamilβ
βthe best open-source bases (Chatterbox, Indic Parler-TTS, IndicF5) trail them on measured phonological dimensionsβ
βFor intra-sentential code-mix we add a third branch (IndicF5 + native-script transliteration) that drops code-mix LLM-WER from 0.80-0.85 to 0.14-0.27β
βWe release R6 LoRA weights (Apache-2.0), inference code and router (MIT), and a Gradio demo.β
βWe release R6 LoRA weights (Apache-2.0), inference code and router (MIT), and a Gradio demo.β
βWe introduce SOB (The Structured Output Benchmark), a multi-source benchmark spanning three source modalities: native text, images, and audio conversations.β
βData and code are available at https://github.com/qzhangFDU/faithfulness-qa-dataset.β
βI've long objected to the 'surveillance advertising' trope, for it trivializes actual surveillance under force of government, such as this, brought to us by the folks at Palantir. Gift link.β
βWe test whether the causal inner product of \citet{park2024linear} -- defined by the unembedding covariance $Ξ£$ -- enables cross-lingual concept transport.β
βRemarkable @DIEZEIT story: a researcher brings a last letter from a communist executed by the Nazis to the daughter he never knew.β
βresolve the open question of Gaillard, Gerchinovitz, Huard, and Stoltz, \emph{``Uniform regret bounds over $\mathbb{R}^d$ for the sequential linear regression problem with the square loss''} (ALT 2019β¦β
βOur source code is publicly available at https://github.com/HySonLab/PRIMEβ
βWe introduce Minimum Specification Perturbation (MSP), the smallest number of changes.β
βWe introduce Perturb-and-Correct (P&C), a post-hoc method for constructing epistemically diverse predictors from a single pretrained network.β
βWe present Basis Selection with Importance (BSI), a principled low-rank compression framework that ranks and prunes bases by directly estimating the expected loss increase incurred when each basis is β¦β
βOur results establish Mollifier Layers as an efficient and scalable tool for physics-constrained learning.β
βIntroducing Super Broadband from T-Mobile for Business. Nationwide 5G integrated with Starlink. Discover more at superbroadband.com.β
βIn this paper, we develop two metrics for critically examining this assumption: Causal Importance of Reasoning (CIR)... and Sufficiency of Reasoning (SR)...β
βHere we introduce PluRel, a framework to synthesize multi-tabular relational databases from scratch.β
βwe're introducing the Open Molecules 2025 data set and Meta's universal model for atoms...By making open molecules and universal model available, we're enabling researchers to drive innovationβ
βThis model and data set combination enables exceptional speed and accuracy for modeling the world at the atomic scale, accelerating the discovery of new molecules and materials.β
βWe prove that SOAP, a recently proposed quasi-Newton method, efficiently approximates the Hessian preconditioner, enabling breakthrough performance in PINNsβ
βiOS 26.5 is officially out to the public... there were a good amount of features worth sharing with 26.5 and a bunch of quality of life updates.β
βWe formalize this setting as reinforcement learning with rich feedback and introduce Self-Distillation Policy Optimization (SDPO)β
βTo address these limitations, we introduce Obj-Disco, a framework that automatically decomposes an alignment reward signal into a sparse, weighted combination of human-interpretable natural language oβ¦β
βI encourage you to explore our full blog post for more details and together let's push the boundaries of AI research to solve the big scientific questions about human and machine intelligence.β
βWebsite is spyglass.org. Definitely one of my must reads uh whenever it comes to tech and AI.β
βthey put out this, um documentary. It's called The Thinking Game. I'm in the middle of it... Very good. It's with people... it has 260 million views on YouTube.β
βI have a Pixel Fold as my sort of Android backup device that I test things on, and I do um generally like it a lotβ
βmost people know the Transformers library, but there's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cppβ
βthere's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cpp. So it's a constellation of tools for AI builders to build with open modβ¦β
βthere's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cpp. So it's a constellation of tools for AI builders to build with open modβ¦β
βthere's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cpp. So it's a constellation of tools for AI builders to build with open modβ¦β
βthere's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cpp. So it's a constellation of tools for AI builders to build with open modβ¦β
βthere's a whole ecosystem around it from Diffusers, Sentence Transformers, TRL, PEFT, even LLaVA, and since last week Llama.cpp. So it's a constellation of tools for AI builders to build with open modβ¦β
βjust last week the Transformers team wrote a really, really cool blog post about MoEs and how they work. So if anybody watching is interested, that's a great way to get into it.β
βWe're building Mistral AI Studio, which is kind of like a platform you can customize as a company using our open source modelsβ
βMeta is the right partner for open-source AI development. Not just with their llama models they have a gamut of other infrastructure llama stack exeutor torch and many libraries.β
βSolo is a platform for physical AI inference. The basic philosophy of Solo is instead of trying to have a big model, we are able to have an ensemble of fine-tuned models on your devices locally.β
βIf people want to learn about Asana's AI teammates, where do they go? They go to asana.com.β
βfor the AI teammates launched, we have chosen Anthropics Opus 3.6 model. Uh, and that's what we're launching with right now. That's how it's powered. uh it uh did the best in our testing and analysisβ
βAll six checkpoints are released on the HuggingFace Hub at https://huggingface.co/PearlLeeStudio.β
βWe detail the SPINE framework and case studies at https://github.com/rminshen03/EAI_Privacy_Position.β
βCode is available at https://github.com/dl-m9/SIOP.git.β
βCode is available at https://github.com/Indigma-Innovations/federated-learning-ev-charging-demand.β
βConnect the dots: Build with built-in and custom MCPs in Studioβ
βWe introduce MuJoCo Playground, a fully open-source framework for robot learning built with MJX...the entire framework is freely available at playground.mujoco.orgβ
βHead to quince.com slash songexploder for free shipping on your order and 365-day returns.β
βI think you would really dig it. Mixtape comes out May 7th on console and PC. Check it out at mixtape.game.β
βEmma's newest novel, American Fantasy, is also about music fandom and identity. That story is set on a cruise ship.β
βI mean, the album. If I think of it as one complete album that I know better than any other.β
βWe argue that phi_first should be reported as a default low-cost baseline before invoking sampling-based uncertainty estimation.β
βFor reproducibility, we publish our code to https://github.com/UNL-CPN-Lab/Look-Once-Beam-Twice.β
βwe present a systematic taxonomy of jailbreak attacks and defenses and introduce Security Cube, a unified, multi-dimensional framework for comprehensive evaluation of these techniques.β
βThe whole thing is built from a centralized YAML file. So it should be pretty modular in case you want to try it on different content.β
βSome more vibe-coding fun - every math major's favorite party trick: the wobbly table theorem as an interactive 3D visualization. https://timvieira.github.io/table-theorem/β
βI generate the embeddings locally from PDF or markdown sources using an embedded model @nomic_ai, which supports large docsβ
βHere's something I built to explore the stuff I've written (blog posts and publications). It's still a bit of work in progress, but it's pretty fun, especially the "semantic" tab + sliders.β
βthese are from a great lecture given by Chase, the CEO of Crusoe, who's building a lot of these data center campuses. So I think he's a good source to pull from.β
βthe Magnetic Fields are putting out a special colored limited edition vinyl of their album, Love at the Bottom of the Sea, that's only going to be available at her store, Books Are Magic.β
βStephen Merritt himself was a guest on Song Exploder, talking about the making of the song, Andrew in Drag, way back in 2015. It's a great episode. Please check it out.β
βListen to Proxy with Yo-Wei Shaw wherever you get your podcasts.β
βBrilliant is an interactive learning platform built for thinkers and builders. It turns tough subjects, math, computer science, data analysis, even AI, into bite-sized, hands-on lessons that actually β¦β
βWe present Agent4MR, an agent-based framework that automatically generates and refines PyPulseq sequences using a structured, physics-aware validation report.β
βWe evaluated Agent4MR on a spin-echo EPI task across three state-of-the-art LLMs...Agent4MR...automatically generates and refines PyPulseq sequencesβ
βthanks to the tax cuts and jobs act uh thanks for rolling back regulations thanks to energy independence it was the greatest economy of our lifetimeβ
βTo get started, you need to download the Claude desktop app. Head over to claude.com/download. You'll see download options for Mac and Windows.β
βTo access Cowork specifically, you'll need a Claude Pro or Claude Max. Cowork is a premium feature.β
βI'm going to show you how to do this with BloFin. The link is in the description below, and you'll get up to five USDT just for signing up, plus up to 30 USDT in bonuses when you deposit.β
βYou can take this even further by connecting Claude Cowork to TradingView using webhooks. When your TradingView strategy triggers an alert, the webhook automatically notifies Claude Cowork, and it exeβ¦β
βI recommend using TRC-20 for USDT. It's fast, and the fees are very low.β
βhe's published a popular book that was a New York Times bestseller. The title of that book, is 'Lifespan: Why We Age And Why We Don't Have To.'β
βIf you'd like to try ROKA sunglasses or eyeglasses, you can go to roka.com... and enter the code Huberman, to save 20% off your first order.β
βIf you'd like to try InsideTracker, you can go to insidetracker.com/huberman to get 25% off any of InsideTracker's plans.β
βHe wrote a book called 'The Warrior Diet', which got very little attention at the time. But what he said was when he was in Israeli special forces, they rarely ate more than once per day.β
βI take a precursor... I take a gram of NMN every day... I take Metformin... resveratrol... I've been taking it for about 15 years now... a thousand milligrams.β
βwe take a gram of NMN every day... if you take NMN for the time period that I do... your NAD levels go up by about two fold or more.β
βI take Metformin... it's been found that looking at tens of thousands of veterans... those two type two diabetics live longer than people that don't even get type two diabetes. So it's a longevity druβ¦β
βbefore I had access to Metformin, I was taking berberine... it is effective at boosting energetics in the body, just like AMPK and Metformin does.β
βthere's another one called quercetin... we also showed back in 2003 that it activates sirtuins as well. But others have, 20 years later, found that it kills senescent cells.β