🤖🧠 HunyuanWorld-Mirror: Tencent’s Breakthrough in Universal 3D Reconstruction

🗓️ 03 Nov 2025
📚 AI News & Trends

The race toward achieving universal 3D understanding has reached a significant milestone with Tencent’s HunyuanWorld-Mirror, a cutting-edge open-source model designed to revolutionize 3D reconstruction. In an era dominated by visual intelligence and immersive digital experiences, this new model stands out by offering a feed-forward, geometry-aware framework that can predict multiple 3D outputs in a single ...

#HunyuanWorld #Tencent #3DReconstruction #UniversalAI #GeometryAware #OpenSourceAI

✨ Title: SemCoT: Accelerating Chain-of-Thought Reasoning through Semantically-Aligned Implicit Tokens

📝 Summary:
Chain-of-Thought CoT reasoning is verbose. SemCoT accelerates implicit CoT by ensuring semantic alignment of reasoning steps and speeding up individual implicit token generation. It uses a contrastive sentence transformer and an efficient, lightweight reasoning generator, outperforming state-of-t...

🔹 Publication Date: Published on Oct 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.24940
• PDF: https://arxiv.org/pdf/2510.24940
• Github: https://github.com/YinhanHe123/SemCoT/

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

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

✨ Title: Rank-GRPO: Training LLM-based Conversational Recommender Systems with Reinforcement Learning

📝 Summary:
ConvRec-R1, a two-stage framework, enhances LLM-based conversational recommender systems. It uses behavioral cloning for quality data and introduces Rank-GRPO, an RL method tailored for rank-style outputs. This improves recommendation quality, convergence, Recall, and NDCG.

🔹 Publication Date: Published on Oct 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.20150
• PDF: https://arxiv.org/pdf/2510.20150
• Github: https://github.com/yaochenzhu/Rank-GRPO

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

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

✨ Title: MisSynth: Improving MISSCI Logical Fallacies Classification with Synthetic Data

📝 Summary:
Misinformation is difficult to classify. MisSynth uses RAG to create synthetic fallacy data for LLM fine-tuning. This pipeline substantially improves LLM accuracy in identifying scientific misinformation fallacies, with over 35% F1-score gains.

🔹 Publication Date: Published on Oct 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.26345
• PDF: https://arxiv.org/pdf/2510.26345
• Github: https://github.com/mxpoliakov/MisSynth

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

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

✨ Title: Monopoly Deal: A Benchmark Environment for Bounded One-Sided Response Games

📝 Summary:
A new game structure, Bounded One-Sided Response Games BORGs, involves actions briefly transferring control to an opponent to satisfy a condition. A modified Monopoly Deal is used as a benchmark, and standard CFR effectively learns strategies.

🔹 Publication Date: Published on Oct 29

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.25080
• PDF: https://arxiv.org/pdf/2510.25080

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

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

✨ Title: Beyond Objects: Contextual Synthetic Data Generation for Fine-Grained Classification

📝 Summary:
BOB is a T2I model fine-tuning strategy for synthetic data generation in low-shot fine-grained classification. It extracts class-agnostic attributes to condition fine-tuning, then marginalizes them out during generation. This mitigates overfitting and achieves state-of-the-art results.

🔹 Publication Date: Published on Oct 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.24078
• PDF: https://arxiv.org/pdf/2510.24078
• Github: https://github.com/princetonvisualai/BeyondObjects

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

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

✨ Title: Every Activation Boosted: Scaling General Reasoner to 1 Trillion Open Language Foundation

📝 Summary:
Ling 2.0 introduces reasoning-oriented language models, scaling to 1 trillion parameters using sparse Mixture-of-Experts. It leverages activated computation to boost reasoning efficiency and capability up to 7-fold compared to dense models. This demonstrates sparse activation enables scalable, ef...

🔹 Publication Date: Published on Oct 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.22115
• PDF: https://arxiv.org/pdf/2510.22115

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

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

✨ Title: Towards Universal Video Retrieval: Generalizing Video Embedding via Synthesized Multimodal Pyramid Curriculum

📝 Summary:
This paper presents a co-designed framework for universal video retrieval. It introduces the UVRB benchmark, synthesizes multimodal data, and devises a Modality Pyramid curriculum for the General Video Embedder GVE. GVE achieves state-of-the-art zero-shot generalization, highlighting limitations ...

🔹 Publication Date: Published on Oct 31

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.27571
• PDF: https://arxiv.org/pdf/2510.27571
• Project Page: https://gzn00417.github.io/GVE/

🔹 Models citing this paper:
• https://huggingface.co/Alibaba-NLP/GVE-3B
• https://huggingface.co/Alibaba-NLP/GVE-7B

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

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

✨ Title: Generalizing Test-time Compute-optimal Scaling as an Optimizable Graph

📝 Summary:
This paper optimizes multi-LLM collaboration graphs for TTS, finding compute-optimal designs. It proposes Agent-REINFORCE, an LLM-agent framework using textual feedback to efficiently find them. Outperforms baselines, balancing accuracy and latency.

🔹 Publication Date: Published on Oct 29

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.00086
• PDF: https://arxiv.org/pdf/2511.00086

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

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

✨ Title: Towards Robust Mathematical Reasoning

📝 Summary:
The paper presents IMO-Bench, advanced mathematical reasoning benchmarks at the International Mathematical Olympiad level. These include short answer and proof writing evaluations. Gemini Deep Think achieved gold-level IMO performance, significantly outperforming other models on IMO-Bench.

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01846
• PDF: https://arxiv.org/pdf/2511.01846
• Project Page: https://imobench.github.io/

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

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

✨ Title: UniREditBench: A Unified Reasoning-based Image Editing Benchmark

📝 Summary:
UniREditBench is a new benchmark for reasoning-based image editing. It covers diverse scenarios including multi-object interactions and game-worlds, using multimodal evaluation to assess generative models. This helps improve their performance on complex editing tasks.

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01295
• PDF: https://arxiv.org/pdf/2511.01295

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

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

✨ Title: LongCat-Flash-Omni Technical Report

📝 Summary:
LongCat-Flash-Omni is a 560B parameter open-source omni-modal model excelling at low-latency real-time audio-visual interaction. It employs a progressive training strategy and achieves state-of-the-art performance across diverse multimodal benchmarks.

🔹 Publication Date: Published on Oct 31

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.00279
• PDF: https://arxiv.org/pdf/2511.00279

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

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

✨ Title: TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning

📝 Summary:
TIR-Bench introduces a comprehensive benchmark for evaluating agentic thinking-with-images in AI. It features 13 tasks requiring novel tool use for image processing. The benchmark is universally challenging, demanding genuine thinking-with-images capabilities.

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01833
• PDF: https://arxiv.org/pdf/2511.01833
• Github: https://github.com/agents-x-project/TIR-Bench

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

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

✨ Title: Unified Diffusion VLA: Vision-Language-Action Model via Joint Discrete Denoising Diffusion Process

📝 Summary:
This paper introduces Unified Diffusion VLA UD-VLA, a vision-language-action model that jointly optimizes image generation and action prediction. It uses a Joint Discrete Denoising Diffusion Process JD3P for intrinsic synergy across modalities. UD-VLA achieves state-of-the-art results on multiple...

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01718
• PDF: https://arxiv.org/pdf/2511.01718
• Project Page: https://irpn-eai.github.io/UD-VLA.github.io/
• Github: https://github.com/OpenHelix-Team/UD-VLA

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

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

✨ Title: The Underappreciated Power of Vision Models for Graph Structural Understanding

📝 Summary:
Vision models show surprising power for graph understanding, matching GNNs on benchmarks and outperforming them on global structural perception. Our new GraphAbstract benchmark reveals vision models excel at holistic graph properties and scale-invariant reasoning, suggesting their use for graph f...

🔹 Publication Date: Published on Oct 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.24788
• PDF: https://arxiv.org/pdf/2510.24788

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

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

✨ Title: ROVER: Benchmarking Reciprocal Cross-Modal Reasoning for Omnimodal Generation

📝 Summary:
ROVER is a new benchmark evaluating reciprocal cross-modal reasoning in unified multimodal models. It tests how models use one modality to guide or verify outputs in another, in both verbal and visual generation tasks. Experiments show cross-modal reasoning is vital for visual generation, but mod...

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01163
• PDF: https://arxiv.org/pdf/2511.01163
• Github: https://roverbench.github.io/

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

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

✨ Title: Trove: A Flexible Toolkit for Dense Retrieval

📝 Summary:
Trove is an open-source toolkit for dense retrieval that simplifies research. It offers efficient on-the-fly data management, reducing memory use and allowing flexible dataset experiments. Trove is highly customizable and provides a unified, scalable pipeline for evaluation.

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01857
• PDF: https://arxiv.org/pdf/2511.01857
• Project Page: https://ir-trove.dev/
• Github: https://github.com/BatsResearch/trove

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

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

✨ Title: Data-Efficient RLVR via Off-Policy Influence Guidance

📝 Summary:
This paper proposes CROPI a new method for efficient data selection in Reinforcement Learning with Verifiable Rewards RLVR. It uses off-policy influence estimation and sparse random projection to identify the most valuable data points. CROPI significantly accelerates training achieving 2.66x spee...

🔹 Publication Date: Published on Oct 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.26491
• PDF: https://arxiv.org/pdf/2510.26491

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

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

✨ Title: How Far Are Surgeons from Surgical World Models? A Pilot Study on Zero-shot Surgical Video Generation with Expert Assessment

📝 Summary:
This study introduces SurgVeo and the Surgical Plausibility Pyramid to evaluate video generation models in surgery. Experts found Veo-3 visually convincing but lacking in actual surgical understanding. This highlights a critical gap between visual mimicry and causal knowledge in surgical AI.

🔹 Publication Date: Published on Nov 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.01775
• PDF: https://arxiv.org/pdf/2511.01775

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

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

✨ Title: UME-R1: Exploring Reasoning-Driven Generative Multimodal Embeddings

📝 Summary:
UME-R1 introduces generative multimodal embeddings, unifying embedding tasks within a generative paradigm. Its two-stage MLLM training creates reasoning-driven embeddings that significantly outperform conventional discriminative methods, offering a foundation for new interpretability.

🔹 Publication Date: Published on Nov 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.00405
• PDF: https://arxiv.org/pdf/2511.00405
• Github: https://github.com/DeepLearnXMU/UME-R1

🔹 Models citing this paper:
• https://huggingface.co/zhibinlan/UME-R1-2B
• https://huggingface.co/zhibinlan/UME-R1-7B

✨ Datasets citing this paper:
• https://huggingface.co/datasets/zhibinlan/UME-sft-train
• https://huggingface.co/datasets/zhibinlan/UME-rl-train

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

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

✨ Title: World Simulation with Video Foundation Models for Physical AI

📝 Summary:
Cosmos-Predict2.5 is a new world foundation model for physical AI, unifying Text, Image, and Video2World generation with enhanced quality and control for robotics. It works with Cosmos-Transfer2.5 for Sim2Real translation. Both are open-source to accelerate embodied intelligence research.

🔹 Publication Date: Published on Oct 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.00062
• PDF: https://arxiv.org/pdf/2511.00062
• Github: https://github.com/nvidia-cosmos/cosmos-transfer2.5

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

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