✨Few Tokens Matter: Entropy Guided Attacks on Vision-Language Models

📝 Summary:
Targeting high-entropy tokens in vision-language models causes significant semantic degradation with reduced budgets. This attack strategy reveals critical transferable safety risks across different VLM architectures.

🔹 Publication Date: Published on Dec 26, 2025

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.21815
• PDF: https://arxiv.org/pdf/2512.21815

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

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

#VisionLanguageModels #AdversarialAI #AIsecurity #MachineLearning #DeepLearning

✨Multi-Agent Software Development through Cross-Team Collaboration

📝 Summary:
Existing multi-agent LLM software development yields a single solution, missing better alternatives. We introduce Cross-Team Collaboration CTC, a framework where multiple agent teams propose and communicate diverse decisions. This significantly improves software quality and generalizes well.

🔹 Publication Date: Published on Jun 13, 2024

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2406.08979
• PDF: https://arxiv.org/pdf/2406.08979
• Github: https://github.com/OpenBMB/ChatDev

✨ Spaces citing this paper:
• https://huggingface.co/spaces/shanghengdu/LLM-Agent-Optimization-PaperList

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

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

#MultiAgentSystems #LLMAgents #SoftwareDevelopment #AICollaboration #AIResearch

✨CoV: Chain-of-View Prompting for Spatial Reasoning

📝 Summary:
Chain-of-View CoV prompting enhances spatial reasoning in 3D embodied question answering for vision-language models. It actively explores environments by selecting question-aligned views and iteratively adjusting camera positions to gather context, leading to significant performance gains across ...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05172
• PDF: https://arxiv.org/pdf/2601.05172

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

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

#SpatialReasoning #VisionLanguageModels #EmbodiedAI #Prompting #AI

✨One Sample to Rule Them All: Extreme Data Efficiency in RL Scaling

📝 Summary:
This paper demonstrates extreme data efficiency in RL for LLMs. A single, carefully designed training sample, called polymath learning, significantly enhances multidisciplinary reasoning, outperforming traditional methods that rely on large datasets. The findings suggest sample quality and design...

🔹 Publication Date: Published on Jan 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.03111
• PDF: https://arxiv.org/pdf/2601.03111

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

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

#ReinforcementLearning #LLMs #DataEfficiency #AI #DeepLearning

✨LEMAS: Large A 150K-Hour Large-scale Extensible Multilingual Audio Suite with Generative Speech Models

📝 Summary:
LEMAS introduces the largest open-source 150K-hour multilingual speech dataset with word-level timestamps. Models trained on this dataset, LEMAS-TTS and LEMAS-Edit, achieve high-quality zero-shot speech synthesis and seamless speech editing.

🔹 Publication Date: Published on Jan 4

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.04233
• PDF: https://arxiv.org/pdf/2601.04233
• Project Page: https://huggingface.co/spaces/LEMAS-Project/LEMAS-Edit

🔹 Models citing this paper:
• https://huggingface.co/LEMAS-Project/LEMAS-TTS

✨ Datasets citing this paper:
• https://huggingface.co/datasets/LEMAS-Project/LEMAS-Dataset-train
• https://huggingface.co/datasets/LEMAS-Project/LEMAS-Dataset-eval

✨ Spaces citing this paper:
• https://huggingface.co/spaces/LEMAS-Project/LEMAS-TTS
• https://huggingface.co/spaces/LEMAS-Project/LEMAS-Edit
• https://huggingface.co/spaces/Kaiden423/LEMAS-TTS

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Multi-Scale Local Speculative Decoding for Image Generation

📝 Summary:
Multi-Scale Local Speculative Decoding accelerates autoregressive image generation through multi-resolution drafting and spatially informed verification while maintaining semantic quality and perceptu...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05149
• PDF: https://arxiv.org/pdf/2601.05149
• Project Page: https://qualcomm-ai-research.github.io/mulo-sd-webpage/
• Github: https://qualcomm-ai-research.github.io/mulo-sd-webpage

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Scaling Behavior Cloning Improves Causal Reasoning: An Open Model for Real-Time Video Game Playing

📝 Summary:
Behavior cloning demonstrates improved performance and causal reasoning through scaling model size and training data, achieving human-level gameplay in 3D video games. AI-generated summary Behavior cl...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.04575
• PDF: https://arxiv.org/pdf/2601.04575
• Project Page: https://elefant-ai.github.io/open-p2p/
• Github: https://github.com/elefant-ai/open-p2p

🔹 Models citing this paper:
• https://huggingface.co/elefantai/open-p2p

✨ Datasets citing this paper:
• https://huggingface.co/datasets/elefantai/p2p-toy-examples
• https://huggingface.co/datasets/elefantai/p2p-full-data

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Scaling Large-Language-Model-based Multi-Agent Collaboration

📝 Summary:
This paper introduces MacNet for multi-agent collaboration using DAGs for reasoning, outperforming baselines and scaling to many agents. It unveils a collaborative scaling law where emergent abilities appear much earlier than neural emergence.

🔹 Publication Date: Published on Jun 11, 2024

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2406.07155
• PDF: https://arxiv.org/pdf/2406.07155
• Project Page: https://github.com/OpenBMB/ChatDev/tree/macnet
• Github: https://github.com/OpenBMB/ChatDev/tree/macnet

✨ Spaces citing this paper:
• https://huggingface.co/spaces/shanghengdu/LLM-Agent-Optimization-PaperList

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨PyramidalWan: On Making Pretrained Video Model Pyramidal for Efficient Inference

📝 Summary:
Pyramidal diffusion models offer efficient inference by varying resolution based on noise. This paper presents a low-cost finetuning pipeline to convert pretrained diffusion models into pyramidal ones, maintaining output quality. They also explore step distillation for enhanced efficiency.

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.04792
• PDF: https://arxiv.org/pdf/2601.04792
• Project Page: https://qualcomm-ai-research.github.io/PyramidalWan

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ReHyAt: Recurrent Hybrid Attention for Video Diffusion Transformers

📝 Summary:
ReHyAt introduces a recurrent hybrid attention mechanism that combines softmax and linear attention benefits, enabling efficient video generation with reduced computational costs and improved scalabil...

🔹 Publication Date: Published on Jan 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.04342
• PDF: https://arxiv.org/pdf/2601.04342
• Project Page: https://qualcomm-ai-research.github.io/rehyat

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Learning User Preferences Through Interaction for Long-Term Collaboration

📝 Summary:
MultiSessionCollab benchmark evaluates agents' ability to learn and adapt to user preferences through persistent memory systems that enhance long-term collaboration quality. AI-generated summary As co...

🔹 Publication Date: Published on Jan 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.02702
• PDF: https://arxiv.org/pdf/2601.02702

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

👩‍💻 FREE 2026 IT Learning Kits Giveaway

🔥Whether you're preparing for #Cisco #AWS #PMP #Python #Excel #Google #Microsoft #AI or any other in-demand certification – SPOTO has got you covered!

🎁 Explore Our FREE Study Resources
·IT Certs E-book : https://bit.ly/3YvSMHL
·IT exams skill Test : https://bit.ly/4r4VHnd
·Python, ITIL, PMP, Excel, Cyber Security, cloud, SQL Courses : https://bit.ly/4qNWl8r
·Free AI online preparation material and support tools : https://bit.ly/4qKiKTN

🔗 Need IT Certs Exam Help？ contact: wa.link/dm4kyz
📲 Join IT Study Group for insider tips & expert support:
https://chat.whatsapp.com/BEQ9WrfLnpg1SgzGQw69oM

✨GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

📝 Summary:
GRPO in multi-reward RL suffers from reward normalization collapse, hindering training. GDPO resolves this by decoupling individual reward normalization, improving stability and accuracy. GDPO consistently outperforms GRPO across various reasoning tasks.

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05242
• PDF: https://arxiv.org/pdf/2601.05242
• Project Page: https://nvlabs.github.io/GDPO/
• Github: https://github.com/NVlabs/GDPO

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

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

#ReinforcementLearning #MultiRewardRL #PolicyOptimization #MachineLearning #AI

✨Learnable Multipliers: Freeing the Scale of Language Model Matrix Layers

📝 Summary:
Learnable multipliers address suboptimal weight norms caused by weight decay in large language models. They free the scale of weight matrices using learnable scalar, then per-row and per-column multipliers, outperforming baselines and improving performance with reduced overhead.

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.04890
• PDF: https://arxiv.org/pdf/2601.04890
• Project Page: https://tiiuae.github.io/Falcon-H1/
• Github: https://github.com/tiiuae/falcon-h1

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

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

#LLM #DeepLearning #MachineLearning #AI #Optimization

✨RL-AWB: Deep Reinforcement Learning for Auto White Balance Correction in Low-Light Night-time Scenes

📝 Summary:
RL-AWB is a novel framework for nighttime auto white balance. It combines statistical methods with deep reinforcement learning, mimicking expert tuning to improve color constancy in low-light scenes. The method shows superior generalization across various lighting conditions and includes a new mu...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05249
• PDF: https://arxiv.org/pdf/2601.05249
• Project Page: https://ntuneillee.github.io/research/rl-awb/
• Github: https://github.com/BrianChen1120/RL-AWB

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

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

#ReinforcementLearning #DeepLearning #ComputerVision #ImageProcessing #AWB

✨Token-Level LLM Collaboration via FusionRoute

📝 Summary:
FusionRoute is a token-level multi-LLM collaboration framework that uses a lightweight router to select optimal experts and add complementary logits, outperforming existing methods in diverse tasks wh...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05106
• PDF: https://arxiv.org/pdf/2601.05106

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨RelayLLM: Efficient Reasoning via Collaborative Decoding

📝 Summary:
RelayLLM enables efficient collaborative reasoning between small and large language models through token-level dynamic invocation, achieving high accuracy with minimal computational overhead. AI-gener...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05167
• PDF: https://arxiv.org/pdf/2601.05167
• Github: https://github.com/Chengsong-Huang/RelayLLM

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice

📝 Summary:
VideoAuto-R1 framework employs a reason-when-necessary strategy for video understanding, using a Thinking Once, Answering Twice training paradigm with verifiable rewards and confidence-based reasoning...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05175
• PDF: https://arxiv.org/pdf/2601.05175
• Project Page: https://ivul-kaust.github.io/projects/videoauto-r1/
• Github: https://github.com/IVUL-KAUST/VideoAuto-R1/

✨ Spaces citing this paper:
• https://huggingface.co/spaces/sming256/VideoAuto-R1_Demo

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨RoboVIP: Multi-View Video Generation with Visual Identity Prompting Augments Robot Manipulation

📝 Summary:
Collecting diverse robot manipulation data is challenging. This paper introduces visual identity prompting, using exemplar images to guide diffusion models for generating multi-view, temporally coherent data. This augmented data improves robot policy performance in both simulation and real-world ...

🔹 Publication Date: Published on Jan 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05241
• PDF: https://arxiv.org/pdf/2601.05241
• Project Page: https://robovip.github.io/RoboVIP/
• Github: https://robovip.github.io/RoboVIP/

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

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

#Robotics #AI #GenerativeAI #ComputerVision #MachineLearning