🔹 Title: Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens

🔹 Publication Date: Published on Aug 2

🔹 Abstract: CoT reasoning in LLMs is found to be limited by the distribution discrepancy between training and test data, suggesting it is not a robust form of reasoning. AI-generated summary Chain-of-Thought (CoT) prompting has been shown to improve Large Language Model (LLM) performance on various tasks. With this approach, LLMs appear to produce human-like reasoning steps before providing answers (a.k.a., CoT reasoning), which often leads to the perception that they engage in deliberate inferential processes. However, some initial findings suggest that CoT reasoning may be more superficial than it appears, motivating us to explore further. In this paper, we study CoT reasoning via a data distribution lens and investigate if CoT reasoning reflects a structured inductive bias learned from in-distribution data, allowing the model to conditionally generate reasoning paths that approximate those seen during training. Thus, its effectiveness is fundamentally bounded by the degree of distribution discrepancy between the training data and the test queries. With this lens, we dissect CoT reasoning via three dimensions: task, length, and format. To investigate each dimension, we design DataAlchemy , an isolated and controlled environment to train LLMs from scratch and systematically probe them under various distribution conditions. Our results reveal that CoT reasoning is a brittle mirage that vanishes when it is pushed beyond training distributions. This work offers a deeper understanding of why and when CoT reasoning fails, emphasizing the ongoing challenge of achieving genuine and generalizable reasoning .

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.01191

• PDF: https://arxiv.org/pdf/2508.01191

• Github: https://github.com/ChengshuaiZhao0/DataAlchemy

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Selective Contrastive Learning for Weakly Supervised Affordance Grounding

🔹 Publication Date: Published on Aug 11

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.07877
• PDF: https://arxiv.org/pdf/2508.07877
• Github: https://github.com/hynnsk/SelectiveCL

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: EgoTwin: Dreaming Body and View in First Person

🔹 Publication Date: Published on Aug 18

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.13013
• PDF: https://arxiv.org/pdf/2508.13013

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Enhanced Arabic Text Retrieval with Attentive Relevance Scoring

🔹 Publication Date: Published on Jul 31

🔹 Abstract: An enhanced Dense Passage Retrieval framework for Arabic uses a novel Attentive Relevance Scoring mechanism to improve retrieval performance and ranking accuracy. AI-generated summary Arabic poses a particular challenge for natural language processing (NLP) and information retrieval (IR) due to its complex morphology, optional diacritics and the coexistence of Modern Standard Arabic (MSA) and various dialects. Despite the growing global significance of Arabic, it is still underrepresented in NLP research and benchmark resources. In this paper, we present an enhanced Dense Passage Retrieval (DPR) framework developed specifically for Arabic. At the core of our approach is a novel Attentive Relevance Scoring (ARS) that replaces standard interaction mechanisms with an adaptive scoring function that more effectively models the semantic relevance between questions and passages. Our method integrates pre-trained Arabic language models and architectural refinements to improve retrieval performance and significantly increase ranking accuracy when answering Arabic questions. The code is made publicly available at https://github.com/Bekhouche/APR{GitHub}.

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2507.23404

• PDF: https://arxiv.org/pdf/2507.23404

• Github: https://github.com/Bekhouche/APR

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: ODYSSEY: Open-World Quadrupeds Exploration and Manipulation for Long-Horizon Tasks

🔹 Publication Date: Published on Aug 11

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.08240
• PDF: https://arxiv.org/pdf/2508.08240
• Project Page: https://kaijwang.github.io/odyssey.github.io/

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Learnable SMPLify: A Neural Solution for Optimization-Free Human Pose Inverse Kinematics

🔹 Publication Date: Published on Aug 19

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.13562
• PDF: https://arxiv.org/pdf/2508.13562
• Github: https://github.com/Charrrrrlie/Learnable-SMPLify

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: CARFT: Boosting LLM Reasoning via Contrastive Learning with Annotated Chain-of-Thought-based Reinforced Fine-Tuning

🔹 Publication Date: Published on Aug 21

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.15868
• PDF: https://arxiv.org/pdf/2508.15868

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Sketch3DVE: Sketch-based 3D-Aware Scene Video Editing

🔹 Publication Date: Published on Aug 19

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.13797
• PDF: https://arxiv.org/pdf/2508.13797
• Github: https://github.com/IGLICT/Sketch3DVE

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: RotaTouille: Rotation Equivariant Deep Learning for Contours

🔹 Publication Date: Published on Aug 22

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.16359
• PDF: https://arxiv.org/pdf/2508.16359
• Github: https://github.com/odinhg/rotation-equivariant-contour-learning

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Hop, Skip, and Overthink: Diagnosing Why Reasoning Models Fumble during Multi-Hop Analysis

🔹 Publication Date: Published on Aug 6

🔹 Abstract: Research investigates reasoning failures in language models for multi-hop question answering, introducing a framework to categorize errors and improve model fidelity. AI-generated summary The emergence of reasoning models and their integration into practical AI chat bots has led to breakthroughs in solving advanced math, deep search, and extractive question answering problems that requires a complex and multi-step thought process. Yet, a complete understanding of why these models hallucinate more than general purpose language models is missing. In this investigative study, we systematicallyexplore reasoning failures of contemporary language models on multi-hop question answering tasks. We introduce a novel, nuanced error categorization framework that examines failures across three critical dimensions: the diversity and uniqueness of source documents involved ("hops"), completeness in capturing relevant information ("coverage"), and cognitive inefficiency ("overthinking"). Through rigorous hu-man annotation, supported by complementary automated metrics, our exploration uncovers intricate error patterns often hidden by accuracy-centric evaluations. This investigative approach provides deeper insights into the cognitive limitations of current models and offers actionable guidance toward enhancing reasoning fidelity , transparency , and robustness in future language modeling efforts.

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.04699

• PDF: https://arxiv.org/pdf/2508.04699

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: PosterGen: Aesthetic-Aware Paper-to-Poster Generation via Multi-Agent LLMs

🔹 Publication Date: Published on Aug 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.17188
• PDF: https://arxiv.org/pdf/2508.17188

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.18265
• PDF: https://arxiv.org/pdf/2508.18265
• Github: https://github.com/OpenGVLab/InternVL

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: UQ: Assessing Language Models on Unsolved Questions

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.17580
• PDF: https://arxiv.org/pdf/2508.17580
• Project Page: https://huggingface.co/datasets/uq-project/uq

🔹 Datasets citing this paper:
• https://huggingface.co/datasets/uq-project/uq

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: ST-Raptor: LLM-Powered Semi-Structured Table Question Answering

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.18190
• PDF: https://arxiv.org/pdf/2508.18190
• Github: https://github.com/weAIDB/ST-Raptor

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: SpotEdit: Evaluating Visually-Guided Image Editing Methods

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.18159
• PDF: https://arxiv.org/pdf/2508.18159
• Github: https://github.com/SaraGhazanfari/SpotEdit

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Neither Valid nor Reliable? Investigating the Use of LLMs as Judges

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.18076
• PDF: https://arxiv.org/pdf/2508.18076

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for General LLM Reasoning

🔹 Publication Date: Published on Aug 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.16949
• PDF: https://arxiv.org/pdf/2508.16949

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Visual-CoG: Stage-Aware Reinforcement Learning with Chain of Guidance for Text-to-Image Generation

🔹 Publication Date: Published on Aug 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.18032
• PDF: https://arxiv.org/pdf/2508.18032

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: T2I-ReasonBench: Benchmarking Reasoning-Informed Text-to-Image Generation

🔹 Publication Date: Published on Aug 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.17472
• PDF: https://arxiv.org/pdf/2508.17472

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: TaDiCodec: Text-aware Diffusion Speech Tokenizer for Speech Language Modeling

🔹 Publication Date: Published on Aug 22

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.16790
• PDF: https://arxiv.org/pdf/2508.16790

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

🔹 Title: Explain Before You Answer: A Survey on Compositional Visual Reasoning

🔹 Publication Date: Published on Aug 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.17298
• PDF: https://arxiv.org/pdf/2508.17298
• Project Page: https://github.com/pokerme7777/Compositional-Visual-Reasoning-Survey
• Github: https://github.com/pokerme7777/Compositional-Visual-Reasoning-Survey

🔹 Datasets citing this paper:
No datasets found

🔹 Spaces citing this paper:
No spaces found
==================================

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT