What the smart people are recommending. 8874 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.
βsomeone told me you know, Andrewing is cool, but there's a British guy who speaks really slowly so watch it at 1.5x. Uh his lectures are even better. That was Jeffrey Hint's lectures.β
βI I really thought thought that TV show Silicon Valley was pretty cool... it's actually pretty true uh it's not just like a humorous showβ
βCode and data are available at https://github.com/dieKarotte/Spatial-Omniβ
βthere also the movie on Stephven Hawking so like it's a lot of movies have been made... that's one of my one of my favorite moviesβ
βnot goodwill hunting Google hunting is also amazing.β
βwe propose AutoPDE, a code agent that maintains the solver strategy as an explicitly represented object throughout the solving processβ
βSpecifically, experiments with popular methods like ROME and FT reveal a substantial performance gap, up to 24%, between evaluations on directly edited knowledge and on entailed knowledge.β
βLong or short Google. Uh, just buy the S&P 500. That's my advice.β
βthere's an interview, 60 minutes interview with uh him and someone asked him like um why why didn't you give up... I don't ever give up. I would have to be dead or incapacitated.β
βI I don't have um Facebook. I don't have Messenger. Okay. I I don't have a Facebook account and I I kind of recommend this.β
βSpecifically, experiments with popular methods like ROME and FT reveal a substantial performance gap, up to 24%, between evaluations on directly edited knowledge and on entailed knowledge.β
βWe propose SDBN (Small Data Big Noise), a unified framework that brings adversarial training to PEFTβ
βWe also introduce two variants of the method that use discrete uncertainty sets: SDBN-h, which enumerates character-level edits and selects worst-case variants using gradientsβ
βSinch is already helping major brands send RCS messages around the world. And they're helping Lenny's Podcast listeners get registered first before the rush hits the US market. Learn more at getstarteβ¦β
βSign up for a free account today at Liveblocks.io/Lenny.β
βIf who have you have heard about slide the presentation framework um no one you should check it out it's cool.β
βand SDBN-p, which uses LLM-generated variants for robust optimization in generative tasksβ
βwe propose Causal Ensemble Agent (CEA), a novel framework that aggregates structural insights from statistical discovery experts across different graph levels via linear opinion pooling, and uses an Lβ¦β
βWe propose FlowTracer, an RL framework that traces answer-targeted reasoning flow on an attention-induced directed acyclic graphβ
βcompanies like Ander are companies like Anderoll is because they're making unmanned systems.β
βThe code is available at https://github.com/aisilab/arbiter.β
βIn this work, we propose the Unified Dexterous Hand Model (UDHM), which maps human and robot hand states into a shared 22-DoF semantic interface.β
βIn this work, we propose the Unified Dexterous Hand Model (UDHM), which maps human and robot hand states into a shared 22-DoF semantic interface.β
βUltimately, ParaEval provides a robust and efficient way to evaluate true underlying capability rather than surface-form familiarity.β
βWe validate this approach using a custom dataset of long Minecraft videos, where it produces substantially more consistent rollouts compared to existing baselines.β
βWe introduce WebChallenger, a web agent framework that addresses each gap through architecture design rather than model scale, built around PageMem: a structured page representation deterministically β¦β
βWe propose Spense, a practical hybrid sparse-dense format that splits each weight matrix into a 2:4 sparse region and a dense region. This design relaxes the effective sparsity constraint while remainβ¦β
βWe propose Spense, a practical hybrid sparse-dense format that splits each weight matrix into a 2:4 sparse region and a dense region.β
βthere's a there's a class of um or or or branch of literature that happened in Latin America called magic realism with Garcia Marcus and a bunch of others. Like reading those things continues to be suβ¦β
βllm.swift by @otabuzzman: a Swift port of this projectβ
βWe propose Prefilling-dLLM, a training-free prefill-decode disaggregation framework for dLLMs that partitions the prefix into N chunks, caches their KV representations once, and selects the top-K mostβ¦β
βWe introduce the representation-aware advantage estimation, which leverages RM hidden states and models them as auxiliary signals for better advantage estimation.β
βWe apply GraphAE to GRPO, GSPO and RLOO, and conduct extensive experiments on different models and benchmarks.β
βWe apply GraphAE to GRPO, GSPO and RLOO, and conduct extensive experiments on different models and benchmarks.β
βWe apply GraphAE to GRPO, GSPO and RLOO, and conduct extensive experiments on different models and benchmarks.β
βTo this end, we introduce LakeQA, a comprehensive benchmark for search-centric question answering over data lakes that jointly emphasizes searching and reasoning capabilities.β
βIn this paper, we present EinsteinArena, an agent-native platform for open distributed research and discovery.β
βWe introduce Knowledge-Grounded Counterfactual Reasoning (KG-CFR), a dual-stage architecture that enforces a strict separation of concerns between a private, retrieval-augmented planning buffer, and aβ¦β
βThis paper studies a third primitive: a parallel content-addressed memory over causal successor records. The proposed Parallel Causal Associative Field (PCAF) writes local records from a context windoβ¦β
βWe evaluate PCAF under full autoregressive pretraining on WikiText-103 and PG-19 using a distributed Google Cloud TPU v4-32 pod. At 303M parameters and context length T = 2048, PCAF-semantic reaches 3β¦β
βOur implementation is publicly available at https://github.com/IMSeonL/coset-ensemble-decoder.β
βWe release SWE-Marathon, evaluation code, and agent trajectories at https://swe-marathon.org/.β
βThe results reveal meaningful trade-offs between accuracy and resource efficiency, with the FRQI model achieving a 97 percent reduction in circuit depth while maintaining a near-perfect reconstructionβ¦β
βHowever, on real quantum hardware, these encodings can quickly lead to circuits with many gates, large circuit depth, and high qubit usage, which is a problem for Noisy Intermediate-Scale Quantum (NISβ¦β
βHowever, on real quantum hardware, these encodings can quickly lead to circuits with many gates, large circuit depth, and high qubit usage, which is a problem for Noisy Intermediate-Scale Quantum (NISβ¦β
βThe trajectories are generated using a recently derived Matsubara generalized Langevin equation, which produces the imaginary correlations by evolving the stochastic variables into the complex plane.β
βIn this paper, we develop a chip-integrated diamond photonic crystal cavity embedding an ensemble of nitrogen-vacancy (NV) centers.β
βHere we realize scalable and efficient homodyne detection by leveraging a large number of pixels in a charge-coupled-device (CCD) camera.β
βDespite the recent progress in generating large-scale quantum states, realizing quantum measurement at scale remains a major challenge.β
βThis work provides a scalable method for quantum measurement, paving the way for large-scale quantum information processing.β
βthe computational tool used: Qibo, an open-source framework for efficient quantum simulations and quantum hardware control.β
βwe propose EEVEE, the first multi-dataset test-time prompt learning framework for LLM agents, enabling test-time prompt learning under real-world task streams.β
βwe propose Target-SFT which constructs the training objective directly from the desired target distribution.β
βIn this work, we test the validity and robustness of this assumption based on the criterion $2 Ο_\varphi < |\langle \varphi \rangle| $ for classical field behaviour and applying it to aribtrary quaβ¦β
βthe folks at kaggle actually are are pushing on this notion of like um having personal evals and being able to sort of build a platform in which as new models are released and made available to the woβ¦β
βreleased the the sort of experimental variant of 2.0 pro which is the the most capable Frontier Model we have... Pro is where you're sort of not bound by costs in a lot of waysβ
βwe also previewed in December the multimodal live API which allows you to do this sort of collaborative realtime conversational video uh text interface with the modelsβ
βWe propose a Transformer-based network for cuffless BP estimation from PPG signal, leveraging self-attention to capture long-range dependencies across multiple cardiac cycles.β
βUnder calibration-based evaluation protocols on the large-scale PulseDB dataset, the proposed method achieves MAE of 4.56 mmHg for systolic BP and 2.62 mmHg for diastolic BP, reducing errors by 47% anβ¦β
βWe study the task of agnostically learning general (as opposed to homogeneous) ReLUs under the Gaussian distribution with respect to the squared loss.β
βif you're finding value in the show please take a moment to share it with friends write a review or reach out via our website cognitive revolutionβ
βagnostically learning general (as opposed to homogeneous) ReLUs under the Gaussian distributionβ
βif people haven't tried these tools then the best thing you can do for your career and for your friends and so on is to get your hands very dirtyβ
βMando starter pack is perfect for new customers it comes with a solid Stick deodorant cream tube deodorant two free products of your choice like a mini body wash and deodorant wipes and free shippingβ
βWe introduce the ItΓ΄ map, an any-step stochastic flow map that takes an intermediate state and Brownian path and predicts future states in a single pass.β
βSelf-distillation trains the model to retain this improvement when the context is not present.β
βwe have a company called do not pay it's this incredibly genius guy with this thing and it's it's an app that started out with get me out of a parking ticketβ
βthere's a there's this guy Peter Hubert who wrote this book called Orwell's revenge and he rewrites 1984 and he makes a key change in the technology which is he makes the telescreen two-wayβ
βthere's a great book you might have run into it's by Chuck cluster it's called what if we're wrong... it's amazing you already know you're going to love itβ
βNeil Ferguson is a really good guy by the way to read his writings on this he's historian who's quite quite good on this topicβ
βCode to reproduce the results is available at https://github.com/IlayMalinyak/mm_align_vs_pred.β
βthere's a Twitter account there's this guy who goes by the name Pliny the Elder p l i n y the Elder on X and his he's an independent guy and uh he jailbreaks all the new AIs on the same day that they'β¦β
βWe propose a scheme to realize tunable and strong nonreciprocal photon blockade (PB) in an asymmetric Fabry-PΓ©rot cavity.β
βThe proposed protocol requires only standard coherent laser sources and is compatible with current cavity QED experimental setups, offering a practical and scalable platform for nonreciprocal quantum β¦β
βWe introduce OncoTraj, a public benchmark of 813 EGFR-mutant NSCLC patients receiving first-line osimertinib, harmonized from three real-world clinical-genomic sourcesβ
βWe propose a specific adaptive CVaR schedule that gradually tightens the sampled tail used for optimization, together with a two-stage optimizer combining global exploration with Particle Swarm Optimiβ¦β
βApple Intelligence comes to Apple Vision Pro today with visionOS 2.4β
βWe propose a specific adaptive CVaR schedule that gradually tightens the sampled tail used for optimization, together with a two-stage optimizer combining global exploration with Particle Swarm Optimiβ¦β
βIn the reported QPU layout comparison, the proposed heavy-hex-native deep-chain layout achieves the best final objective value and CVaR-tail performance among the tested layouts.β
βWe also study ansatz depth and sequential growth strategies. Finally, we introduce two hardware-aware ansatz-layout modifications: a data-guided colored layout that assigns correlated variables to qubβ¦β
βLatent diffusion models (LDMs) provide efficient mappings from high-dimensional geological model space to a low-dimensional latent variable, reducing the dimensionality of the inverse problem while maβ¦β
βWe compare model-space and latent-space DA using the ensemble smoother with multiple data assimilation (ESMDA), and demonstrate a key trade-off: model-space updates achieve significant uncertainty redβ¦β
βMotivated by this, we explore rigorous Markov chain Monte Carlo (MCMC) and Sequential Monte Carlo (SMC) algorithms in the 3D-LDM latent space.β
βMotivated by this, we explore rigorous Markov chain Monte Carlo (MCMC) and Sequential Monte Carlo (SMC) algorithms in the 3D-LDM latent space.β
βAll models maintain geological realism due to the LDM parameterization. MCMC and SMC are consistent with one another and achieve lower data mismatch and more uncertainty reduction than latent-space ESβ¦β
βAll models maintain geological realism due to the LDM parameterization. MCMC and SMC are consistent with one another and achieve lower data mismatch and more uncertainty reduction than latent-space ESβ¦β
βWe present explicit controlled-SWAP-based unitary constructions that achieve exact uniformity with quadratic circuit size and linear depth (O(n)) on linear nearest neighbor topologies.β
βReid Hoffman pitches LinkedIn to himselfβ
β_Blog post by Reid Hoffman_ https://www.reidhoffman.org/my-five-years-on-the-openai-board/β
βWe introduce CIAware-Bench, a benchmark for measuring control intervention (CI) awareness across frontier models.β
βArmra colostrum is nature's original health hack. Packed with over 400 bioactive nutrients that fortify gut integrity, strengthen immunity, revitalize hair growth, fuel stamina, elevate focus.β
βCode and implementation details are available at https://github.com/zachmocz/temporal-ids-benchβ
βWe advocate leakage-free splits, explicit padding disclosure, and sequence-aware benchmarking as standard practice in future IDS research.β
βWe advocate leakage-free splits, explicit padding disclosure, and sequence-aware benchmarking as standard practice in future IDS research.β
βWe advocate leakage-free splits, explicit padding disclosure, and sequence-aware benchmarking as standard practice in future IDS research.β
βWe introduce GASLoC, a novel decentralized pre-training algorithm that generalizes the notion of communication acceleration to the recently popular "outer optimizer" to allow a practical gossip-based β¦β
βIn this paper we study the question of whether simple policy improvement schemes at test time alone, leaving stable supervised policy training intact, can be a competitive alternative which sidesteps β¦β
βTo this end, we propose QGF (Q-Guided Flow), an RL algorithm that performs policy optimization entirely at test time.β
βwe adapt a discrete-token, autoregressive trajectory generator from autonomous driving (SMART) and show it yields strong best-of-20 forecasting accuracy compared to baselines, while supporting fully hβ¦β
βwe propose GenAIR, a general framework that empowers sequential recommendation with Generative Archetype-grounded Item Representationsβ