Carlos's Debrief

⚡ Quick Summary

Clinical LLM safety vs. accuracy scaling: A new SaFE-Scale framework reveals that safety and accuracy follow different scaling laws in clinical LLMs — bigger models can be more accurate while being less safe on high-risk, evidence-contradicting errors that matter more than average benchmark scores.
AI red teaming in the agentic era: A new framework automates AI red-teaming workflows that previously took weeks, reducing the cycle to hours by dynamically assembling attacks, transforms, and scorers instead of hand-crafting each one.
Diffusion-based flow sampling: Flow Sampling introduces a training objective that regresses onto a denoising diffusion drift for sampling from unnormalized densities — enabling efficient data-free generative modeling when the target is defined by a known energy function.
Conversational symptom assessment at scale: SymptomAI, deployed via Fitbit, demonstrates end-to-end patient interviewing and differential diagnosis using conversational AI agents, moving beyond curated medical vignettes to everyday real-world symptom reporting.
Compositional vulnerability in coding agents: MOSAIC-Bench reveals that AI coding agents can pass per-prompt safety checks yet ship exploitable code when their tasks are decomposed into innocuous-looking engineering tickets — a structural vulnerability in multi-step AI software generation.

📰 News Headlines

RIPE NCC SSO token exposure: Over 1,000 third parties were given access to RIPE NCC's all-powerful single sign-on tokens by design — any logged-in user who clicked a single link could unknowingly leak their session to external parties.
Cloudflare reads ChatGPT React state on every keystroke: A technical analysis reveals that Cloudflare's infrastructure actively monitors React component state inside ChatGPT's frontend on every user keystroke, raising fresh questions about the boundaries of client-side surveillance in AI products.
Predictive video latents boost generation quality: A new Predictive Video VAE (PV-VAE) model achieves 52% faster convergence and a 34.42 FVD improvement by encoding only partial observations and predicting future frames — improving the temporal coherence of video generative models.
Autonomous research agents go multi-model adversarial: ARIS, an open-source research harness, coordinates long-horizon ML research by pitting an executor LLM against a reviewer from a different model family by default, catching plausible-but-unsupported claims before they propagate through the research pipeline.
Current AI agents average only 47.4% on real workspace tasks: Workspace-Bench, a new benchmark with 388 tasks across 20,476 real files, shows that the best AI agent harness reaches just 68.7% (human baseline: 80.7%), highlighting how far agent systems remain from reliable real-world file-based work.

🧠 LLMs 17

Safety and accuracy follow different scaling laws in clinical large language models [1]

Sebastian Wind, Tri-Thien Nguyen, Jeta Sopa, Mahshad Lotfinia, Sebastian Bickelhaup, Michael Uder, Harald Köstler, Gerhard Wellein, Sven Nebelung, Daniel Truhn, Andreas Maier, Soroosh Tayebi Arasteh

Clinical LLMs are often scaled by increasing model size, context length, retrieval complexity, or inference-time compute, with the implicit expectation that higher accuracy implies safer behavior. This assumption is incomplete in medicine, where a few…

cs.CLcs.AIcs.LG

OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty Trajectories [2]

Yuwen Du, Rui Ye, Shuo Tang, Keduan Huang, Xinyu Zhu, Yuzhu Cai, Siheng Chen

Deep search capabilities have become an indispensable competency for frontier Large Language Model (LLM) agents, yet their development remains dominated by industrial giants. The typical industry recipe involves a highly resource-intensive pipeline spanning pre-training, continual…

cs.AIcs.CL

Redefining AI Red Teaming in the Agentic Era: From Weeks to Hours [3]

Raja Sekhar Rao Dheekonda, Will Pearce, Nick Landers

AI systems are entering critical domains like healthcare, finance, and defense, yet remain vulnerable to adversarial attacks. While AI red teaming is a primary defense, current approaches force operators into manual, library-specific workflows. Operators spend…

cs.AIcs.CR

SymptomAI: Towards a Conversational AI Agent for Everyday Symptom Assessment [4]

Joseph Breda, Fadi Yousif, Beszel Hawkins, Marinela Cotoi, Miao Liu, Ray Luo, Po-Hsuan Cameron Chen, Mike Schaekermann, Samuel Schmidgall, Xin Liu, Girish Narayanswamy, Samuel Solomon, Maxwell A. Xu, Xiaoran Fan, Longfei Shangguan, Anran Wang, Bhavna Daryani, Buddy Herkenham, Cara Tan, Mark Malhotra, Shwetak Patel, John B. Hernandez, Quang Duong, Yun Liu, Zach Wasson, Dimitrios Antos, Bob Lou, Matthew Thompson, Jonathan Richina, Anupam Pathak, Nichole Young-Lin, Jake Sunshine, Daniel McDuff

Language models excel at diagnostic assessments on currated medical case-studies and vignettes, performing on par with, or better than, clinical professionals. However, existing studies focus on complex scenarios with rich context making it difficult to…

cs.AI

Physics-Grounded Multi-Agent Architecture for Traceable, Risk-Aware Human-AI Decision Support in Manufacturing [5]

Danny Hoang, Ryan Matthiessen, Christopher Miller, Nasir Mannan, Ruby ElKharboutly, David Gorsich, Matthew P. Castanier, Farhad Imani

High-precision CNC machining of free-form aerospace components requires bounded compensations informed by inspection, simulation, and process knowledge. Off-the-shelf large language model (LLM) assistants can generate text, but they do not reliably execute risk-constrained multi-step numerical…

cs.MAcs.AIcs.IR

An Agent-Oriented Pluggable Experience-RAG Skill for Experience-Driven Retrieval Strategy Orchestration [6]

Dutao Zhang, Tian Liao

Retrieval-augmented generation systems often assume that one fixed retrieval pipeline is sufficient across heterogeneous tasks, yet factoid question answering, multi-hop reasoning, and scientific verification exhibit different retrieval preferences. We present Experience-RAG Skill, an agent-oriented pluggable…

cs.AI

From Intent to Execution: Composing Agentic Workflows with Agent Recommendation [7]

Kishan Athrey, Ramin Pishehvar, Brian Riordan, Mahesh Viswanathan

Multi-Agent Systems (MAS) built using AI agents fulfill a variety of user intents that may be used to design and build a family of related applications. However, the creation of such MAS currently involves manual…

cs.AI

Flow Sampling: Learning to Sample from Unnormalized Densities via Denoising Conditional Processes [8]

Aaron Havens, Brian Karrer, Neta Shaul

Sampling from unnormalized densities is analogous to the generative modeling problem, but the target distribution is defined by a known energy function instead of data samples. Because evaluating the energy function is often costly, a…

cs.LGcs.AI

Feature-Augmented Transformers for Robust AI-Text Detection Across Domains and Generators [9]

Mohamed Mady, Johannes Reschke, Björn Schuller

AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detectors on HC3 PLUS and calibrate a…

cs.CLcs.AI

Label-Efficient School Detection from Aerial Imagery via Weakly Supervised Pretraining and Fine-Tuning [10]

Zakarya Elmimouni, Fares Fourati, Mohamed-Slim Alouini

Accurate school detection is essential for supporting education initiatives, including infrastructure planning and expanding internet connectivity to underserved areas. However, many regions around the world face challenges due to outdated, incomplete, or unavailable official records.…

cs.CVcs.AIcs.LG

FTPrimitiveBench: A Benchmark Suite For Logical Computation Under Hardware-Motivated and Biased Noise Models [11]

Shuwen Kan, Adrian Harkness, Zefan Du, Rod Rofougaran, Sean Garner, Chenxu Liu, Ying Mao, Samuel Stein

Fault-tolerant quantum computing requires understanding how error-correcting codes perform on diverse physical hardware. This is typically assessed via noisy stabilizer simulation of logical circuits at HPC scale, combined with a noise model that yields a…

quant-ph

Catastrophe-dispersion models in random and varying environments across generations [12]

Lucas R. de Lima, Fábio P. Machado

We study a class of branching processes in which the offspring distribution is not specified directly but is induced by a cycle of internal colony growth, catastrophic reduction and structured dispersal. The parameters governing growth,…

math.PRq-bio.PE

Audio-Visual Intelligence in Large Foundation Models [13]

You Qin, Kai Liu, Shengqiong Wu, Kai Wang, Shijian Deng, Yapeng Tian, Junbin Xiao, Yazhou Xing, Yinghao Ma, Bobo Li, Roger Zimmermann, Lei Cui, Furu Wei, Jiebo Luo, Hao Fei

Audio-Visual Intelligence (AVI) has emerged as a central frontier in artificial intelligence, bridging auditory and visual modalities to enable machines that can perceive, generate, and interact in the multimodal real world. In the era of…

cs.CV

UniCorrn: Unified Correspondence Transformer Across 2D and 3D [14]

Prajnan Goswami, Tianye Ding, Feng Liu, Huaizu Jiang

Visual correspondence across image-to-image (2D-2D), image-to-point cloud (2D-3D), and point cloud-to-point cloud (3D-3D) geometric matching forms the foundation for numerous 3D vision tasks. Despite sharing a similar problem structure, current methods use task-specific designs with…

cs.CV

Ergotropy Protection via Cavity Detuning in Collective Open Quantum Batteries [15]

Tariq Zeyad Jawad

This study investigates the performance and ergotropy protection of open collective quantum batteries subject to superradiant decay. By employing a passive spectral detuning strategy within an intermediate cavity, an optimal detuning value ($Δ^*$) is analytically…

quant-ph

Jordan Frame in Supergravity and Cosmology [16]

Renata Kallosh

The superconformal action can be gauge-fixed in a gauge where is leads to the Einstein frame supergravity defined by a \K potential $\mathcal{K}(z, \bar z)$, or in a gauge where it leads to a Jordan…

hep-th

Large Language Models are Universal Reasoners for Visual Generation [17]

Sucheng Ren, Chen Chen, Zhenbang Wang, Liangchen Song, Xiangxin Zhu, Alan Yuille, Liang-Chieh Chen, Jiasen Lu

Text-to-image generation has advanced rapidly with diffusion models, progressing from CLIP and T5 conditioning to unified systems where a single LLM backbone handles both visual understanding and generation. Despite the architectural unification, these systems frequently…

cs.CV

🤖 Machine Learning 7

A Closed-Form Adaptive-Landmark Kernel for Certified Point-Cloud and Graph Classification [18]

Sushovan Majhi, Atish Mitra, Žiga Virk, Pramita Bagchi

We introduce PALACE (Persistence Adaptive-Landmark Analytic Classification Engine), the data-adaptive companion to PLACE, paying a small cross-validation tier on three knobs (budget, radii, bandwidth; $\leq 5$ choices each). A cover-theoretic core (Lebesgue-number criterion on the…

cs.LGmath.AT

Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures [19]

Evangelos Ntavelis, Sean Wu, Mohamad Shahbazi, Fabio Maninchedda, Dmitry Kostiaev, Artem Sevastopolsky, Vittorio Megaro, Trevor Phillips, Alejandro Blumentals, Shridhar Ravikumar, Mehak Gupta, Reinhard Knothe, Jeronimo Bayer, Matthias Vestner, Simon Schaefer, Thomas Etterlin, Christian Zimmermann, Mathias Deschler, Peter Kaufmann, Stefan Brugger, Sebastian Martin, Brian Amberg, Tom Runia

We propose HeadsUp, a scalable feed-forward method for reconstructing high-quality 3D Gaussian heads from large-scale multi-camera setups. Our method employs an efficient encoder-decoder architecture that compresses input views into a compact latent representation. This latent…

cs.CVcs.LG

Conditional Diffusion Sampling [20]

Francisco M. Castro-Macías, Pablo Morales-Álvarez, Saifuddin Syed, Daniel Hernández-Lobato, Rafael Molina, José Miguel Hernández-Lobato

Sampling from unnormalized multimodal distributions with limited density evaluations remains a fundamental challenge in machine learning and natural sciences. Successful approaches construct a bridge between a tractable reference and the target distribution. Parallel Tempering (PT)…

stat.MLcs.LG

Enhanced 3D Brain Tumor Segmentation Using Assorted Precision Training [21]

Adwaitt Pandya, Ozioma C. Oguine, Harita Bhargava, Shrikant Zade

A brain tumor is a medical disorder faced by individuals of all demographics. Medically, it is described as the spread of non-essential cells close to or throughout the brain. Symptoms of this ailment include headaches,…

cs.CVcs.LG

Pretrained Model Representations as Acquisition Signals for Active Learning of MLIPs [22]

Eszter Varga-Umbrich, Shikha Surana, Paul Duckworth, Jules Tilly, Olivier Peltre, Zachary Weller-Davies

Training machine learning interatomic potentials (MLIPs) for reactive chemistry is often bottlenecked by the high cost of quantum chemical labels and the scarcity of transition state configurations in candidate pools. Active learning (AL) can mitigate…

cs.LGphysics.chem-ph

Transformers with Selective Access to Early Representations [23]

Skye Gunasekaran, Téa Wright, Rui-Jie Zhu, Jason Eshraghian

Several recent Transformer architectures expose later layers to representations computed in the earliest layers, motivated by the observation that low-level features can become harder to recover as the residual stream is repeatedly transformed through depth.…

cs.LGcs.CL

Integrating Feature Correlation in Differential Privacy with Applications in DP-ERM [24]

Tianyu Wang, Luhao Zhang, Rachel Cummings

Standard differential privacy imposes uniform privacy constraints across all features, overlooking the inherent distinction between sensitive and insensitive features in practice. In this paper, we introduce a relaxed definition of differential privacy that accounts for…

cs.LGstat.ML

🔐 Security & Crypto 11

Probabilistic-bit Guided CDCL for SAT Solving using Ising Consensus Assumptions [25]

Melki Bino

Boolean satisfiability (SAT) solvers are widely used in hardware verification, cryptanalysis, automatic test-pattern generation, and side-channel reasoning workflows. Modern conflict-driven clause-learning (CDCL) solvers are highly effective, but satisfiable instances may still require substantial conflict analysis…

cs.CRcs.LO

LIPPEN: A Lightweight In-Place Pointer Encryption Architecture for Pointer Integrity [26]

Erfan Iravani, Lalit Prasad Peri, Mohannad Ismail, Charitha Tumkur Siddalingaradhya, Changwoo Min, Elif Bilge Kavun, Wenjie Xiong

Memory-safety violations in C and C++ programs continue to enable sophisticated exploitation techniques such as control-flow hijacking and data-oriented attacks. Existing hardware defenses either rely on address space layout randomization (ASLR) or attach explicit metadata…

cs.CRcs.AR

Generating Proof-of-Vulnerability Tests to Help Enhance the Security of Complex Software [27]

Shravya Kanchi, Xiaoyan Zang, Ying Zhang, Danfeng Yao, Na Meng

Developers create modern software applications (Apps) on top of third-party libraries (Libs). When library vulnerabilities are reachable through application code, the applications can be vulnerable to software supply chain attacks. Prior work shows that developers…

cs.CRcs.SE

MOSAIC-Bench: Measuring Compositional Vulnerability Induction in Coding Agents [28]

Jonathan Steinberg, Oren Gal

Coding agents often pass per-prompt safety review yet ship exploitable code when their tasks are decomposed into routine engineering tickets. The challenge is structural: existing safety alignment evaluates overt requests in isolation, leaving models blind…

cs.CRcs.AIcs.SE

HELO Cryptography: A Lightweight Cryptographic System for Enhancing IoT Security in P2P Data Transmission [29]

Tahsin Ahmed, Arjita Saha, Arian Nuhan, Nafim Ahmed Bin Mohammad Noor, Md Faisal Ahmed, Muhammad Iqbal Hossain

The recent surge in security concerns for IoT devices highlights the increasing threat of cryptographic vulnerabilities. These weaknesses can lead to unauthorized access, data breaches, and manipulation of device functions, compromising the privacy and security…

cs.CR

A Deeper Dive into the Irreversibility of PolyProtect: Making Protected Face Templates Harder to Invert [30]

Vedrana Krivokuća Hahn, Jérémy Maceiras, Sébastien Marcel

This work presents a deeper analysis of the "irreversibility" property of PolyProtect, a biometric template protection method initially proposed for securing face embeddings. PolyProtect transforms embeddings into protected templates via multivariate polynomials, whose coefficients and…

cs.CVcs.CR

KVerus: Scalable and Resilient Formal Verification Proof Generation for Rust Code [31]

Yuwei Liu, Xinyi Wan, Yanhao Wang, Minghua Wang, Lin Huang, Tao Wei

Formal verification provides the highest assurance of software correctness and security, but its application to large-scale, evolving systems remains a major challenge. While large language models (LLMs) have shown promise in automating proof generation, they…

cs.SEcs.CR

Large homomorphisms on the homotopy lie coalgebra [32]

Andrew J. Soto Levins, Ryan Watson

We introduce and study a notion of large homomorphisms on the homotopy lie coalgebra; these homomorphisms are a variant of the large homomorphisms of Levin. As a consequence of our work, we establish new cases…

math.ACmath.AT

The Infinite Mutation Engine? Measuring Polymorphism in LLM-Generated Offensive Code [33]

Gabriel Hortea, Juan Tapiador

Malware authors have traditionally relied on polymorphic techniques to produce variants in the same malware family, complicating signature-based detection. Integrating generative AI into offensive toolchains enables attackers to synthesize structurally diverse payloads with identical behavior,…

cs.CR

Solving one-sided linear systems over symmetrized and supertropical semiring [34]

Sulaiman Alhussaini, Sergei Sergeev

One-sided linear systems of the form ``$Ax=b$'' are well-known and extensively studied over the tropical (max-plus) semiring and wide classes of related idempotent semirings. The usual approach is to first find the greatest solution to…

math.RA

Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems [35]

Yilun Zhao, Jinbiao Wei, Tingyu Song, Siyue Zhang, Chen Zhao, Arman Cohan

Reasoning-intensive retrieval aims to surface evidence that supports downstream reasoning rather than merely matching topical similarity. This capability is increasingly important for agentic search systems, where retrievers must provide complementary evidence across iterative search and…

cs.CLcs.IR

🔒 Zero Knowledge 2

Sequential vs. Simultaneous Entanglement Swapping under Optimal Link-Layer Control [36]

Priyam Srivastava, Akshat R. Sabavat, Siddharth Jain, Alan Scheller-Wolf, Sridhar Tayur, David Tipper, Prashant Krishnamurthy, Amy Babay, Kaushik P. Seshadreesan

Connection-less, packet-switched quantum network architectures distribute entanglement across multi-hop paths through sequential entanglement swapping, in which each node acts on purely local state information. The architectural advantages over the connection-oriented alternative -- simultaneous SWAP-ASAP --…

quant-phcs.NI

Can Transformers predict system collapse in dynamical systems? [37]

Zheng-Meng Zhai, Celso Grebogi, Ying-Cheng Lai

Transformer architectures have recently surged as promising solutions for nonlinear dynamical systems, proposed as foundation models capable of zero-shot dynamics reconstruction and forecasting. Despite this success, it remains unclear whether they can truly serve as…

nlin.CDphysics.comp-ph

⚛️ Quantum 3

Model order reduction for parametrized variational inequalities: application to crowd motion [38]

Giulia Sambataro, Virginie Ehrlacher

This work investigates model order reduction for time-dependent parametrized variational inequalities, with a focus on discrete contact problems. As a prototypical example, we consider an agent-based crowd model [Maury et al., 2011] in which agent…

math.NA

Entanglement transitions in translation-invariant tensor networks [39]

Yi-Cheng Wang, Samuel J. Garratt, Ehud Altman

We study the complexity of approximately contracting translation-invariant tensor networks. The computational cost of row-by-row tensor network contraction, which defines a discrete time evolution governed by a fixed transfer matrix, is associated with the entanglement…

quant-phcond-mat.stat-mech

Fast, accurate, high-resolution simulation of large-scale Fermi-Hubbard models on a digital quantum processor [40]

Gavin S. Hartnett, Khadijeh Sona Najafi, Aleksei Khindanov, Haoran Liao, Michael Schutzman, Michael R. Hush, Michael J. Biercuk, Yuval Baum

We report experimental digital quantum simulation of the one-dimensional Fermi-Hubbard model on a superconducting quantum processor at a scale beyond the reach of exact statevector simulation and challenging for state-of-the-art tensor-network methods. We encode this…

quant-ph

🛡️ AI Safety 1

How to pick your football team [41]

Bhargav Narayanan

Team captains Alice and Bob divide up $2m$ footballers, each reduced to a real-valued score, into two teams of $m$ footballers each. On each turn, one captain plays picker, and the other chooser: the picker…

math.CO

🌐 Web Findings

🦞 Lobste.rs 2

1000 third parties could have stolen RIPE NCC session tokens - by design [42]

security

The RIPE NCC made its all-powerful single sign-on tokens available to over 1000 third parties. From a single link click, any logged-in RIPE NCC user would leak …

LOBSTERS

ChatGPT Won't Let You Type Until Cloudflare Reads Your React State. I Decrypted the Program That Does It [42]

privacy

Edit April 2, 2026: I've been getting inbound interest from researchers wanting to run their own queries. The MCP integration I use for my own research lets you analyze live…

LOBSTERS

🤗 HuggingFace Papers 4

Video Generation with Predictive Latents [44]

daily curated papers

Video Variational Autoencoder (VAE) enables latent video generative modeling by mapping the visual world into compact spatiotemporal latent spaces, improving training efficiency and stability. While existing video VAEs achieve commendable…

HUGGINGFACE_PAPERS

ARIS: Autonomous Research via Adversarial Multi-Agent Collaboration [44]

daily curated papers

This report describes ARIS (Auto-Research-in-sleep), an open-source research harness for autonomous research, including its architecture, assurance mechanisms, and early deployment experience. The performance of agent systems built on LLMs depends…

HUGGINGFACE_PAPERS

Skills-Coach: A Self-Evolving Skill Optimizer via Training-Free GRPO [44]

daily curated papers

We introduce Skills-Coach, a novel automated framework designed to significantly enhance the self-evolution of skills within Large Language Model (LLM)-based agents. Addressing the current fragmentation of the skill ecosystem, Skills-Coach…

HUGGINGFACE_PAPERS

Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies [44]

daily curated papers

Workspace learning requires AI agents to identify, reason over, exploit, and update explicit and implicit dependencies among heterogeneous files in a worker's workspace, enabling them to complete both routine and…

HUGGINGFACE_PAPERS

🔬 Google Scholar 6

Early identification of breakthrough technologies: Insights from science-driven innovations [48]

machine learning breakthroughs

… Building on this premise, this paper presents a framework for identifying breakthrough … potential to trigger technological breakthroughs. Next, a machine learning-based link prediction …

GOOGLE_SCHOLAR

Catalyst breakthroughs in methane dry reforming: Employing machine learning for future advancements [48]

machine learning breakthroughs

Rising levels of atmospheric carbon dioxide (CO 2 ) and methane (CH 4 ) have sparked the interest of researchers in resolving this issue. Various technologies have been utilized such…

GOOGLE_SCHOLAR

Large language models (LLM) in computational social science: prospects, current state, and challenges [48]

large language models LLM

… The advent of large language models (LLMs) has marked a new … LLM usage. We further present the challenges associated with data bias, privacy, and the integration of these…

GOOGLE_SCHOLAR

Artificial intelligence and machine learning in cybersecurity: a deep dive into state-of-the-art techniques and future paradigms [48]

AI security cybersecurity

… in adversarial AI, automated threat intelligence, and AI-driven security orchestration, this … AI’s role in cybersecurity. Figure 1 shows the key areas where Artificial intelligence (AI) and …

GOOGLE_SCHOLAR

Securing the future: exploring post-quantum cryptography for authentication and user privacy in IoT devices [48]

cryptography post-quantum

Cluster Computing - With the emergence of quantum computers, traditional cryptographic methods are vulnerable to attacks, emphasizing the need for post-quantum cryptography to secure devices and...

GOOGLE_SCHOLAR

Quantum machine learning: A comprehensive review of integrating AI with quantum computing for computational advancements [48]

quantum computing algorithms

… in quantum-enhanced classical ML to native quantum algorithms and hybrid quantum-… It varies from applications in optimization, drug discovery, and quantum-secured communications, …

GOOGLE_SCHOLAR

⚡ Quick Summary

📰 News Headlines

🧠 LLMs 17

🤖 Machine Learning 7

🔐 Security & Crypto 11

🔒 Zero Knowledge 2

⚛️ Quantum 3

🛡️ AI Safety 1

🌐 Web Findings

🦞 Lobste.rs 2

🤗 HuggingFace Papers 4

🔬 Google Scholar 6

🔗 All Sources