✨OmniRad: A Radiological Foundation Model for Multi-Task Medical Image Analysis

📝 Summary:
OmniRad is a self-supervised radiological foundation model pretrained on 1.2 million medical images. It improves classification F1 by 2.05 percent and achieves better segmentation through representation reuse and cross-task transferability.

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04547
• PDF: https://arxiv.org/pdf/2602.04547
• Github: https://github.com/unica-visual-intelligence-lab/OmniRad

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

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

#MedicalAI #FoundationModels #Radiology #SelfSupervisedLearning #MedicalImaging

✨LongVPO: From Anchored Cues to Self-Reasoning for Long-Form Video Preference Optimization

📝 Summary:
LongVPO is a two-stage DPO framework for short-context VLMs to understand long videos. It uses synthetic preference data from anchored clips and recursive captioning for multi-segment reasoning. LongVPO achieves state-of-the-art with minimal human annotation.

🔹 Publication Date: Published on Feb 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.02341
• PDF: https://arxiv.org/pdf/2602.02341
• Github: https://github.com/MCG-NJU/LongVPO

🔹 Models citing this paper:
• https://huggingface.co/MCG-NJU/LongVPO-Stage2-InternVL3-8B
• https://huggingface.co/MCG-NJU/LongVPO-Stage1-InternVL3-8B

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

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

#VideoUnderstanding #MachineLearning #VLMs #DeepLearning #AIResearch

✨SpatiaLab: Can Vision-Language Models Perform Spatial Reasoning in the Wild?

📝 Summary:
SpatiaLab introduces a comprehensive benchmark to evaluate vision language model spatial reasoning in realistic scenarios. Experiments show a significant performance gap between current models and humans, revealing major limitations in tasks like depth and 3D geometry. This highlights challenges ...

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03916
• PDF: https://arxiv.org/pdf/2602.03916
• Project Page: https://spatialab-reasoning.github.io/
• Github: https://github.com/SpatiaLab-Reasoning/SpatiaLab

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

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

#VisionLanguageModels #SpatialReasoning #ComputerVision #AIResearch #DeepLearning

✨Self-Rewarding Sequential Monte Carlo for Masked Diffusion Language Models

📝 Summary:
This work introduces Self-Rewarding Sequential Monte Carlo SMC to improve masked diffusion language model sampling. SMC uses multiple parallel diffusion processes and trajectory-level confidence as a self-rewarding signal, guiding generation to high-quality samples and boosting performance withou...

🔹 Publication Date: Published on Feb 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.01849
• PDF: https://arxiv.org/pdf/2602.01849
• Project Page: https://algolzw.github.io/sr-smc/
• Github: https://github.com/Algolzw/self-rewarding-smc

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

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

#DiffusionModels #SequentialMonteCarlo #LanguageModels #GenerativeAI #MachineLearning

✨CL-bench: A Benchmark for Context Learning

📝 Summary:
Current LMs struggle with context learning, requiring new knowledge and reasoning beyond pre-training. The CL-bench, a new real-world benchmark, reveals models solve only 17.2 percent of tasks, showing a critical bottleneck for complex real-world applications.

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03587
• PDF: https://arxiv.org/pdf/2602.03587
• Project Page: https://www.clbench.com
• Github: https://github.com/Tencent-Hunyuan/CL-bench

✨ Datasets citing this paper:
• https://huggingface.co/datasets/tencent/CL-bench

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

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

#ContextLearning #LanguageModels #AIBenchmark #NLP #AIResearch

✨Proxy Compression for Language Modeling

📝 Summary:
Proxy compression trains language models on both raw bytes and compressed views. This enables efficient training on compressed inputs while offering a robust, end-to-end raw-byte inference. It improves training efficiency and eventually matches tokenizer performance.

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04289
• PDF: https://arxiv.org/pdf/2602.04289
• Github: https://github.com/LZhengisme/proxy-compression

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

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

#LanguageModels #Compression #MachineLearning #AI #Efficiency

✨3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation

📝 Summary:
3DiMo enables view-agnostic human motion control in video generation by training a motion encoder alongside a pretrained video generator to distill driving frames into compact motion tokens that align...

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03796
• PDF: https://arxiv.org/pdf/2602.03796
• Github: https://hjrphoebus.github.io/3DiMo/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Skin Tokens: A Learned Compact Representation for Unified Autoregressive Rigging

📝 Summary:
Generative 3D models face challenges in animation rigging, which this work addresses by introducing SkinTokens—a learned discrete representation for skinning weights—and TokenRig, a unified autoregres...

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04805
• PDF: https://arxiv.org/pdf/2602.04805
• Project Page: https://zjp-shadow.github.io/works/SkinTokens/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨FASA: Frequency-aware Sparse Attention

📝 Summary:
FASA addresses LLM KV cache memory for long contexts by dynamically predicting token importance. It leverages functional sparsity in RoPEs frequency chunks to identify critical tokens for focused attention. This significantly reduces memory and computation while maintaining high performance.

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03152
• PDF: https://arxiv.org/pdf/2602.03152

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

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

#LLM #SparseAttention #MemoryEfficiency #DeepLearning #NLP

✨AutoFigure: Generating and Refining Publication-Ready Scientific Illustrations

📝 Summary:
AutoFigure is an agentic AI framework that automatically generates publication-ready scientific illustrations from long-form text. It uses extensive thinking and validation to ensure structural soundness and aesthetic appeal. Supported by FigureBench, a large new benchmark, AutoFigure surpasses b...

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03828
• PDF: https://arxiv.org/pdf/2602.03828
• Github: https://github.com/ResearAI/AutoFigure-Edit

✨ Datasets citing this paper:
• https://huggingface.co/datasets/WestlakeNLP/FigureBench

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

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

#AI #GenerativeAI #ScientificIllustrations #ResearchTools #AcademicPublishing

✨D-CORE: Incentivizing Task Decomposition in Large Reasoning Models for Complex Tool Use

📝 Summary:
D-CORE is a two-stage training framework improving large reasoning models' task decomposition and reasoning. It overcomes Lazy Reasoning using self-distillation and diversity-aware reinforcement learning. D-CORE achieves superior tool-use performance, setting new state-of-the-art results even wit...

🔹 Publication Date: Published on Feb 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.02160
• PDF: https://arxiv.org/pdf/2602.02160
• Github: https://github.com/alibaba/EfficientAI

🔹 Models citing this paper:
• https://huggingface.co/bowiehsu/D-CORE-8B

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

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

#LLM #TaskDecomposition #ToolUse #ReinforcementLearning #AIResearch

✨No One-Size-Fits-All: Building Systems For Translation to Bashkir, Kazakh, Kyrgyz, Tatar and Chuvash Using Synthetic And Original Data

📝 Summary:
This paper explores machine translation for five Turkic languages using nllb-200 LoRA fine-tuning on synthetic data and prompt-based methods. It achieved varied chrF++ scores for different language pairs and releases the dataset and model weights.

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04442
• PDF: https://arxiv.org/pdf/2602.04442

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨SAFE: Stable Alignment Finetuning with Entropy-Aware Predictive Control for RLHF

📝 Summary:
A new reinforcement learning algorithm for language model alignment that improves stability and performance over PPO through enhanced KL divergence control and adaptive reward management. AI-generated...

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04651
• PDF: https://arxiv.org/pdf/2602.04651

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨SkeletonGaussian: Editable 4D Generation through Gaussian Skeletonization

📝 Summary:
SkeletonGaussian enables editable 4D generation by decomposing motion into rigid skeleton-driven and non-rigid fine-grained components using hexplane-based refinement. AI-generated summary 4D generati...

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04271
• PDF: https://arxiv.org/pdf/2602.04271
• Project Page: https://wusar.github.io/projects/skeletongaussian/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Reward-free Alignment for Conflicting Objectives

📝 Summary:
This paper introduces RACO, a reward-free alignment framework for LLMs facing multiple conflicting objectives. It uses a novel clipped conflict-averse gradient descent to resolve gradient conflicts directly from pairwise preferences. Experiments show RACO consistently achieves superior Pareto tra...

🔹 Publication Date: Published on Feb 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.02495
• PDF: https://arxiv.org/pdf/2602.02495

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨FOTBCD: A Large-Scale Building Change Detection Benchmark from French Orthophotos and Topographic Data

📝 Summary:
A large-scale building change detection dataset named FOTBCD is introduced, covering 28 French departments with high-resolution imagery and comprehensive annotations for both binary and instance-level...

🔹 Publication Date: Published on Jan 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.22596
• PDF: https://arxiv.org/pdf/2601.22596
• Github: https://github.com/abdelpy/FOTBCD-datasets

✨ Datasets citing this paper:
• https://huggingface.co/datasets/retgenai/FOTBCD-Binary

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨"I May Not Have Articulated Myself Clearly": Diagnosing Dynamic Instability in LLM Reasoning at Inference Time

📝 Summary:
An instability signal from LLM token log probabilities and entropy predicts reasoning failures. This signal, combining distributional shift and uncertainty, reliably forecasts wrong answers. Early instability can be corrective, but late instability more often leads to failure, indicating timing i...

🔹 Publication Date: Published on Feb 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.02863
• PDF: https://arxiv.org/pdf/2602.02863

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent

📝 Summary:
AgentArk distills multi-agent reasoning into a single LLM to overcome the high computational cost of multi-agent systems. This framework enables a single agent to achieve multi-agent intelligence, offering efficient yet powerful reasoning, self-correction, and robustness across diverse tasks.

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.03955
• PDF: https://arxiv.org/pdf/2602.03955
• Github: https://github.com/AIFrontierLab/AgentArk

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨HalluHard: A Hard Multi-Turn Hallucination Benchmark

📝 Summary:
Large language models continue to generate plausible but ungrounded factual claims in multi-turn dialogue, with hallucinations remaining significant even when utilizing web search for verification acr...

🔹 Publication Date: Published on Feb 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.01031
• PDF: https://arxiv.org/pdf/2602.01031

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Trust The Typical

📝 Summary:
Trust The Typical T3 frames LLM safety as an out-of-distribution detection problem, learning what is safe in semantic space. It achieves state-of-the-art performance without harmful example training, drastically reducing false positives and generalizing across languages with low overhead.

🔹 Publication Date: Published on Feb 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.04581
• PDF: https://arxiv.org/pdf/2602.04581

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Learning to Repair Lean Proofs from Compiler Feedback

📝 Summary:
A new dataset, APRIL, pairs erroneous Lean proofs with compiler feedback, corrected proofs, and natural language diagnoses. Training language models on APRIL substantially improves proof repair accuracy and feedback-conditioned reasoning, outperforming existing baselines.

🔹 Publication Date: Published on Feb 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.02990
• PDF: https://arxiv.org/pdf/2602.02990

✨ Datasets citing this paper:
• https://huggingface.co/datasets/uw-math-ai/APRIL

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research