✨KV Packet: Recomputation-Free Context-Independent KV Caching for LLMs

📝 Summary:
KV Packet is a cache reuse framework that eliminates recomputation overhead in large language models by treating cached documents as immutable packets with trainable soft-token adapters. AI-generated ...

🔹 Publication Date: Published on Apr 14

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.13226
• PDF: https://arxiv.org/pdf/2604.13226
• Github: https://github.com/ChuangtaoChen-TUM/KVPacket

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework

📝 Summary:
A unified generator-discriminator framework for autonomous driving motion planning that improves stability and performance through diffusion-based trajectory generation and reinforcement learning opti...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15308
• PDF: https://arxiv.org/pdf/2604.15308
• Project Page: https://hgao-cv.github.io/RAD-2/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨HiVLA: A Visual-Grounded-Centric Hierarchical Embodied Manipulation System

📝 Summary:
HiVLA presents a hierarchical vision-language-action framework that decouples semantic planning from motor control using a diffusion transformer action expert with cascaded cross-attention for improve...

🔹 Publication Date: Published on Apr 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14125
• PDF: https://arxiv.org/pdf/2604.14125
• Project Page: https://tianshuoy.github.io/HiVLA-page/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens

📝 Summary:
GlobalSplat introduces a global scene representation framework that achieves compact, consistent 3D Gaussian splatting with reduced computational overhead and improved inference speed. AI-generated su...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15284
• PDF: https://arxiv.org/pdf/2604.15284
• Project Page: https://r-itk.github.io/globalsplat/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Model Capability Dominates: Inference-Time Optimization Lessons from AIMO 3

📝 Summary:
Majority voting improves mathematical reasoning but is limited by correlated errors; diverse reasoning strategies and model capability are more impactful than prompt engineering. AI-generated summary ...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.27844
• PDF: https://arxiv.org/pdf/2603.27844
• Project Page: https://www.kaggle.com/code/natnitarach/aimo-3-model-capability-dominate
• Github: https://github.com/nat-nischw/model-capability-dominates-lessons-aimo3

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨SuperLocalMemory V3.3: The Living Brain -- Biologically-Inspired Forgetting, Cognitive Quantization, and Multi-Channel Retrieval for Zero-LLM Agent Memory Systems

📝 Summary:
A new local-first agent memory system implements comprehensive cognitive memory processes with enhanced retrieval and forgetting mechanisms, achieving superior performance in zero-LLM settings. AI-gen...

🔹 Publication Date: Published on Apr 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.04514
• PDF: https://arxiv.org/pdf/2604.04514
• Project Page: https://superlocalmemory.com/
• Github: https://github.com/qualixar/superlocalmemory

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨TRACER: Trace-Based Adaptive Cost-Efficient Routing for LLM Classification

📝 Summary:
TRACER trains ML surrogates using LLM classification production traces. These cost-efficient surrogates activate only if they agree with the original LLM above a threshold, saving significant costs. TRACER also provides interpretability for its routing decisions and achieves high coverage.

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14531
• PDF: https://arxiv.org/pdf/2604.14531

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

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

#LLMs #MachineLearning #CostEfficiency #AI #Interpretability

✨OneHOI: Unifying Human-Object Interaction Generation and Editing

📝 Summary:
OneHOI is a unified diffusion transformer framework that consolidates human-object interaction generation and editing into a single conditional denoising process. It uses structured interaction representations to overcome limitations of prior approaches, achieving state-of-the-art results across ...

🔹 Publication Date: Published on Apr 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14062
• PDF: https://arxiv.org/pdf/2604.14062
• Project Page: https://jiuntian.github.io/OneHOI/
• Github: https://github.com/jiuntian/OneHOI

✨ Datasets citing this paper:
• https://huggingface.co/datasets/jiuntian/hoiedit44k
• https://huggingface.co/datasets/jiuntian/IEBench

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Towards Autonomous Mechanistic Reasoning in Virtual Cells

📝 Summary:
Large language models are enhanced for biological research through a multi-agent framework that generates and validates mechanistic explanations using structured formalism and verified datasets. AI-ge...

🔹 Publication Date: Published on Apr 14

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.11661
• PDF: https://arxiv.org/pdf/2604.11661
• Project Page: https://valencelabs.substack.com/p/towards-reasoning-in-virtual-cells
• Github: https://github.com/valence-labs/VCR-Agent

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Switch-KD: Visual-Switch Knowledge Distillation for Vision-Language Models

📝 Summary:
Vision-language models face deployment challenges due to their large size, but knowledge distillation can improve efficiency while maintaining performance through a novel visual-switch framework that ...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14629
• PDF: https://arxiv.org/pdf/2604.14629
• Project Page: https://haoyi199815.github.io/Switch-KD/
• Github: https://github.com/haoyi199815/Switch-KD

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Representations Before Pixels: Semantics-Guided Hierarchical Video Prediction

📝 Summary:
Re2Pix is a hierarchical video prediction framework that improves future video generation by first predicting semantic representations and then using them to guide photorealistic visual synthesis, add...

🔹 Publication Date: Published on Apr 13

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.11707
• PDF: https://arxiv.org/pdf/2604.11707
• Github: https://github.com/Sta8is/Re2Pix

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Don't Retrieve, Navigate: Distilling Enterprise Knowledge into Navigable Agent Skills for QA and RAG

📝 Summary:
Corpus2Skill structures document corpora into hierarchical skill directories for LLM agents. This allows agents to navigate, reason about information, and combine evidence more effectively than traditional RAG. It significantly outperforms other RAG methods on an enterprise benchmark.

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14572
• PDF: https://arxiv.org/pdf/2604.14572
• Github: https://github.com/dukesun99/Corpus2Skill

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography

📝 Summary:
RadAgent, a tool-using AI agent, enhances chest CT report generation through interpretable step-by-step reasoning traces that improve clinical accuracy, robustness, and faithfulness compared to existi...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15231
• PDF: https://arxiv.org/pdf/2604.15231
• Project Page: https://rad-agent.github.io

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Boosting Visual Instruction Tuning with Self-Supervised Guidance

📝 Summary:
Multimodal models struggle with visual reasoning due to under-utilizing visual information during instruction tuning. This paper proposes augmenting instruction tuning with visually grounded self-supervised tasks expressed as natural language. This simple method significantly improves performance...

🔹 Publication Date: Published on Apr 14

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.12966
• PDF: https://arxiv.org/pdf/2604.12966
• Github: https://github.com/sirkosophia/V-GIFT

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨An Optimal Transport-driven Approach for Cultivating Latent Space in Online Incremental Learning

📝 Summary:
An online mixture model learning framework based on optimal transport theory addresses challenges in incremental learning with distributional shifts by enabling dynamic centroid updates and improving ...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2211.16780
• PDF: https://arxiv.org/pdf/2211.16780

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Three-Phase Transformer

📝 Summary:
The Three-Phase Transformer introduces a structural prior for decoder-only Transformers through channel partitioning and phase-respecting operations that stabilize training and improve convergence. AI...

🔹 Publication Date: Published on Apr 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14430
• PDF: https://arxiv.org/pdf/2604.14430
• Github: https://github.com/achelousace/three-phase-transformer

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Beyond Prompts: Unconditional 3D Inversion for Out-of-Distribution Shapes

📝 Summary:
State-of-the-art text-to-3D generative models suffer from latent sink traps where they lose sensitivity to text prompts, but a robust framework can overcome this by decoupling geometric representation...

🔹 Publication Date: Published on Apr 16

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14914
• PDF: https://arxiv.org/pdf/2604.14914
• Project Page: https://daidedou.sorpi.fr/publication/beyondprompts

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Reinforcement Learning via Value Gradient Flow

📝 Summary:
Value Gradient Flow presents a scalable approach to behavior-regularized reinforcement learning by formulating it as an optimal transport problem solved through discrete gradient flow, enabling adapti...

🔹 Publication Date: Published on Apr 15

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14265
• PDF: https://arxiv.org/pdf/2604.14265
• Project Page: https://ryanxhr.github.io/vgf/
• Github: https://github.com/ryanxhr/vgf

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Envisioning the Future, One Step at a Time

📝 Summary:
Autoregressive diffusion models predict open-set future scene dynamics by modeling sparse point trajectories, enabling fast and scalable multi-modal motion prediction with physical plausibility. AI-ge...

🔹 Publication Date: Published on Apr 10

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.09527
• PDF: https://arxiv.org/pdf/2604.09527
• Project Page: https://compvis.github.io/myriad
• Github: https://github.com/compvis/myriad

🔹 Models citing this paper:
• https://huggingface.co/CompVis/myriad

✨ Datasets citing this paper:
• https://huggingface.co/datasets/CompVis/owm-95
• https://huggingface.co/datasets/CompVis/myriad-physics

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨VideoFlexTok: Flexible-Length Coarse-to-Fine Video Tokenization

📝 Summary:
VideoFlexTok enables efficient video representation through variable-length token sequences that capture abstract information first, followed by fine-grained details, allowing for reduced computationa...

🔹 Publication Date: Published on Apr 14

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.12887
• PDF: https://arxiv.org/pdf/2604.12887
• Github: https://github.com/apple/ml-videoflextok

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨PersonaVLM: Long-Term Personalized Multimodal LLMs

📝 Summary:
PersonaVLM introduces a framework for long-term personalized multimodal LLMs. It remembers interactions, reasons multi-turn using retrieved memories, and aligns responses with evolving user personality. This novel method significantly outperforms baselines and GPT-4o on a new evaluation benchmark.

🔹 Publication Date: Published on Mar 20

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.13074
• PDF: https://arxiv.org/pdf/2604.13074
• Project Page: https://personavlm.github.io/
• Github: https://github.com/MiG-NJU/PersonaVLM

🔹 Models citing this paper:
• https://huggingface.co/ClareNie/PersonaVLM

✨ Datasets citing this paper:
• https://huggingface.co/datasets/ClareNie/Persona-MME
• https://huggingface.co/datasets/ClareNie/PersonaVLM-Dataset

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

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

#LLM #MultimodalAI #PersonalizedAI #AIResearch #MemoryAI