✨Benchmarking Knowledge-Extraction Attack and Defense on Retrieval-Augmented Generation

📝 Summary:
This paper introduces the first systematic benchmark for evaluating knowledge-extraction attacks and defenses on Retrieval-Augmented Generation systems. It standardizes testing across diverse models and strategies to enable comparable evaluation and help build privacy-preserving RAG.

🔹 Publication Date: Published on Feb 10

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.09319
• PDF: https://arxiv.org/pdf/2602.09319

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#RAG #KnowledgeExtraction #Cybersecurity #AIPrivacy #Benchmarking

✨Blind to the Human Touch: Overlap Bias in LLM-Based Summary Evaluation

📝 Summary:
LLM judges show bias, increasingly preferring AI-generated summaries over human ones as similarity to human references decreases. This widespread bias across models suggests LLM-as-a-judge needs more sophisticated evaluation beyond simple comparison.

🔹 Publication Date: Published on Feb 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.07673
• PDF: https://arxiv.org/pdf/2602.07673

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#LLM #AIbias #AIEvaluation #NLP #AIethics

✨Data Darwinism Part I: Unlocking the Value of Scientific Data for Pre-training

📝 Summary:
Data Darwinism introduces a ten-level taxonomy for data-model co-evolution. Advanced processing of scientific text, like generative refinement, significantly improves foundation model performance on domain-aligned tasks. This systematic approach unlocks latent data value.

🔹 Publication Date: Published on Feb 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.07824
• PDF: https://arxiv.org/pdf/2602.07824
• Github: https://github.com/GAIR-NLP/Data-Darwinism

🔹 Models citing this paper:
• https://huggingface.co/GAIR/daVinci-origin-3B
• https://huggingface.co/GAIR/daVinci-origin-7B

✨ Datasets citing this paper:
• https://huggingface.co/datasets/GAIR/Darwin-Science

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#DataScience #FoundationModels #Pretraining #GenerativeAI #ScientificData

✨Nanbeige4.1-3B: A Small General Model that Reasons, Aligns, and Acts

📝 Summary:
Nanbeige4.1-3B is a 3B-parameter model excelling in agentic behavior, code generation, and reasoning. It outperforms larger models through advanced reward modeling and training, demonstrating broad competence for a small language model.

🔹 Publication Date: Published on Feb 13

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.13367
• PDF: https://arxiv.org/pdf/2602.13367
• Project Page: https://huggingface.co/Nanbeige/Nanbeige4.1-3B

🔹 Models citing this paper:
• https://huggingface.co/Nanbeige/Nanbeige4.1-3B

✨ Spaces citing this paper:
• https://huggingface.co/spaces/PioTio/AIMan

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#LLM #AI #SmallLanguageModels #AgenticAI #CodeGeneration

✨DeepImageSearch: Benchmarking Multimodal Agents for Context-Aware Image Retrieval in Visual Histories

📝 Summary:
DeepImageSearch introduces an agentic image retrieval paradigm that enables multi-step reasoning over visual histories, moving beyond isolated semantic matching. It uses contextual cues for autonomous exploration. The DISBench benchmark shows current models struggle, proving agentic reasoning is ...

🔹 Publication Date: Published on Feb 11

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.10809
• PDF: https://arxiv.org/pdf/2602.10809
• Github: https://github.com/RUC-NLPIR/DeepImageSearch

✨ Spaces citing this paper:
• https://huggingface.co/spaces/RUC-NLPIR/DISBench-Leaderboard

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#ImageRetrieval #AgenticAI #MultimodalAI #ComputerVision #AIResearch

✨AutoDev: Automated AI-Driven Development

📝 Summary:
AutoDev is an automated AI framework that uses autonomous agents to perform diverse software engineering tasks like coding, testing, and git operations in a secure Docker environment. It achieved high performance on HumanEval, significantly advancing AI-driven development.

🔹 Publication Date: Published on Mar 13, 2024

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2403.08299
• PDF: https://arxiv.org/pdf/2403.08299
• Github: https://github.com/vxcontrol/pentagi

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #SoftwareEngineering #AutomatedDevelopment #AutonomousAgents #GenAI

✨Can I Have Your Order? Monte-Carlo Tree Search for Slot Filling Ordering in Diffusion Language Models

📝 Summary:
McDiffuSE uses Monte Carlo Tree Search to optimize slot infilling order in Masked Diffusion Models, enhancing reasoning performance. It achieved significant gains, revealing non-sequential generation and larger exploration are key to overcoming model biases.

🔹 Publication Date: Published on Feb 13

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.12586
• PDF: https://arxiv.org/pdf/2602.12586

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#MonteCarloTreeSearch #DiffusionModels #NLP #LanguageModels #AI

✨LM-Lexicon: Improving Definition Modeling via Harmonizing Semantic Experts

📝 Summary:
LM-Lexicon improves definition modeling using data clustering and a sparse mixture-of-experts architecture. It trains specialized semantic experts, achieving substantial improvements in definition quality and higher BLEU scores. This advances efficient language models for semantic applications.

🔹 Publication Date: Published on Feb 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.14060
• PDF: https://arxiv.org/pdf/2602.14060
• Project Page: https://lm-lexicon.github.io
• Github: https://github.com/jacklanda/LMLexicon

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨DHPLT: large-scale multilingual diachronic corpora and word representations for semantic change modelling

📝 Summary:
In this resource paper, we present DHPLT, an open collection of diachronic corpora in 41 diverse languages. DHPLT is based on the web-crawled HPLT datasets; we use web crawl timestamps as the approxim...

🔹 Publication Date: Published on Feb 12

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.11968
• PDF: https://arxiv.org/pdf/2602.11968
• Project Page: https://data.hplt-project.org/three/diachronic/
• Github: https://github.com/ltgoslo/scdisc_hplt

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨SPILLage: Agentic Oversharing on the Web

📝 Summary:
Web agents inadvertently disclose user information through both content and behavioral traces, with behavioral oversharing being more prevalent than content oversharing, and this issue persists despit...

🔹 Publication Date: Published on Feb 13

🔹 Paper Links:
• arXiv Page: https://www.arxiv.org/abs/2602.13516
• PDF: https://arxiv.org/pdf/2602.13516
• Github: https://github.com/jrohsc/SPILLage

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨STATe-of-Thoughts: Structured Action Templates for Tree-of-Thoughts

📝 Summary:
STATe-of-Thoughts is an interpretable method that replaces stochastic sampling with discrete textual interventions to search over reasoning patterns. This produces more diverse, high-quality, and explainable text, offering better control and insight into the generation process.

🔹 Publication Date: Published on Feb 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.14265
• PDF: https://arxiv.org/pdf/2602.14265
• Github: https://github.com/zbambergerNLP/state-of-thoughts

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨EditCtrl: Disentangled Local and Global Control for Real-Time Generative Video Editing

📝 Summary:
Efficient video inpainting framework that focuses computation on masked regions while maintaining global context consistency through a lightweight embedder. AI-generated summary High-fidelity generati...

🔹 Publication Date: Published on Feb 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.15031
• PDF: https://arxiv.org/pdf/2602.15031
• Project Page: https://yehonathanlitman.github.io/edit_ctrl/
• Github: https://github.com/yehonathanlitman/EditCtrl

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories

📝 Summary:
AnchorWeave addresses long-term video generation consistency by replacing global 3D scene reconstruction with multiple local geometric memories and a multi-anchor weaving controller to reconcile cross...

🔹 Publication Date: Published on Feb 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.14941
• PDF: https://arxiv.org/pdf/2602.14941
• Project Page: https://zunwang1.github.io/AnchorWeave
• Github: https://github.com/wz0919/AnchorWeave

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Learning to Configure Agentic AI Systems

📝 Summary:
Current AI agent configurations are static and inefficient. This paper introduces ARC, an RL-based system that learns dynamic, per-query configurations. ARC significantly boosts accuracy while cutting costs, proving dynamic configuration is superior.

🔹 Publication Date: Published on Feb 12

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.11574
• PDF: https://arxiv.org/pdf/2602.11574
• Github: https://github.com/somsagar07/Context_Optimization

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis

📝 Summary:
CellMaster uses LLM-encoded knowledge for zero-shot cell-type annotation in single-cell RNA sequencing, improving accuracy over existing tools through interpretable rationales without pre-training. AI...

🔹 Publication Date: Published on Feb 12

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.13346
• PDF: https://arxiv.org/pdf/2602.13346

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨VisPhyWorld: Probing Physical Reasoning via Code-Driven Video Reconstruction

📝 Summary:
VisPhyWorld evaluates MLLMs physical reasoning by requiring them to generate executable simulator code from visual observations. This makes their inferred world representations inspectable and falsifiable, unlike recognition-based methods. Experiments show MLLMs understand scenes but struggle wit...

🔹 Publication Date: Published on Feb 9

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.13294
• PDF: https://arxiv.org/pdf/2602.13294
• Project Page: https://tiger-ai-lab.github.io/VisPhyWorld/
• Github: https://github.com/TIGER-AI-Lab/VisPhyWorld

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Found-RL: foundation model-enhanced reinforcement learning for autonomous driving

📝 Summary:
Found-RL enhances reinforcement learning for autonomous driving using vision-language models, overcoming VLM latency with asynchronous inference. It distills VLM knowledge into a lightweight RL policy through supervision and reward shaping, achieving near VLM performance at real-time speeds.

🔹 Publication Date: Published on Feb 11

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.10458
• PDF: https://arxiv.org/pdf/2602.10458
• Project Page: https://ys-qu.github.io/found-rl-website/
• Github: https://github.com/ys-qu/found-rl

🔹 Models citing this paper:
• https://huggingface.co/ys-qu/found-rl_vlms

✨ Datasets citing this paper:
• https://huggingface.co/datasets/ys-qu/found-rl_dataset

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨GLM-5: from Vibe Coding to Agentic Engineering

📝 Summary:
GLM-5 is a new foundation model designed for agentic engineering. It reduces costs with DSA and improves alignment and efficiency using asynchronous reinforcement learning. GLM-5 achieves state-of-the-art performance in real-world coding and end-to-end software engineering tasks.

🔹 Publication Date: Published on Feb 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.15763
• PDF: https://arxiv.org/pdf/2602.15763
• Github: https://github.com/zai-org/GLM-5

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨On Surprising Effectiveness of Masking Updates in Adaptive Optimizers

📝 Summary:
Random parameter update masking achieves superior optimization for large language models by inducing curvature-dependent regularization, with a momentum-aligned variant delivering significant performa...

🔹 Publication Date: Published on Feb 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.15322
• PDF: https://arxiv.org/pdf/2602.15322

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Does Socialization Emerge in AI Agent Society? A Case Study of Moltbook

📝 Summary:
Large language model agents in networked environments exhibit dynamic stability without true social convergence, maintaining individual diversity while lacking collective influence structures. AI-gene...

🔹 Publication Date: Published on Feb 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.14299
• PDF: https://arxiv.org/pdf/2602.14299
• Project Page: https://github.com/MingLiiii/Moltbook_Socialization
• Github: https://github.com/MingLiiii/Moltbook_Socialization

✨ Datasets citing this paper:
• https://huggingface.co/datasets/AIcell/moltbook-data

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Visual Persuasion: What Influences Decisions of Vision-Language Models?

📝 Summary:
Visual-language models' decision-making preferences are studied through controlled image choice tasks with systematic input perturbations, revealing visual vulnerabilities and safety concerns. AI-gene...

🔹 Publication Date: Published on Feb 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.15278
• PDF: https://arxiv.org/pdf/2602.15278
• Project Page: https://visual-persuasion-website.vercel.app/

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research