ML Research Hub – Telegram
ML Research Hub
@DataScienceT
32.9K subscribers
4.45K photos
273 videos
23 files
4.81K links
Advancing research in Machine Learning – practical insights, tools, and techniques for researchers.

Admin: @HusseinSheikho || @Hussein_Sheikho
Download Telegram
About
Blog
Apps
Platform
Join
ML Research Hub
32.9K subscribers
ML Research Hub
🔹 Title: Multilingual Routing in Mixture-of-Experts

🔹 Publication Date: Published on Oct 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.04694
• PDF: https://arxiv.org/pdf/2510.04694

🔹 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
186 views
Explore Data Science
ML Research Hub
🔹 Title: TaTToo: Tool-Grounded Thinking PRM for Test-Time Scaling in Tabular Reasoning

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/pdf/2510.06217
• PDF: https://arxiv.org/pdf/2510.06217

🔹 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
165 views
Explore Data Science
ML Research Hub
🔹 Title: EgoNight: Towards Egocentric Vision Understanding at Night with a Challenging Benchmark

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06218
• PDF: https://arxiv.org/pdf/2510.06218

🔹 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
135 views
Explore Data Science
ML Research Hub
🔹 Title: Deforming Videos to Masks: Flow Matching for Referring Video Segmentation

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06139
• PDF: https://arxiv.org/pdf/2510.06139

🔹 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
130 views
Explore Data Science
ML Research Hub
🔹 Title: AInstein: Assessing the Feasibility of AI-Generated Approaches to Research Problems

🔹 Publication Date: Published on Oct 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05432
• PDF: https://arxiv.org/pdf/2510.05432

🔹 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
145 views
Explore Data Science
ML Research Hub
🔹 Title: Fast-dLLM v2: Efficient Block-Diffusion LLM

🔹 Publication Date: Published on Sep 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2509.26328
• PDF: https://arxiv.org/pdf/2509.26328

🔹 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
156 views
Explore Data Science
ML Research Hub
🔹 Title: Margin Adaptive DPO: Leveraging Reward Model for Granular Control in Preference Optimization

🔹 Publication Date: Published on Oct 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05342
• PDF: https://arxiv.org/pdf/2510.05342
• Github: https://github.com/sirano1004/Margin-Apative-Direct-Preference-Optimization

🔹 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
164 views
Explore Data Science
ML Research Hub
🔹 Title: Demystifying deep search: a holistic evaluation with hint-free multi-hop questions and factorised metrics

🔹 Publication Date: Published on Oct 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05137
• PDF: https://arxiv.org/pdf/2510.05137

🔹 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
160 views
Explore Data Science
ML Research Hub
🔹 Title: HoloScene: Simulation-Ready Interactive 3D Worlds from a Single Video

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05560
• PDF: https://arxiv.org/pdf/2510.05560
• Project Page: https://xiahongchi.github.io/HoloScene/
• Github: https://github.com/xiahongchi/HoloScene

🔹 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
131 views
Explore Data Science
ML Research Hub
🔹 Title: LightCache: Memory-Efficient, Training-Free Acceleration for Video Generation

🔹 Publication Date: Published on Oct 6

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05367
• PDF: https://arxiv.org/pdf/2510.05367
• Github: https://github.com/NKUShaw/LightCache

🔹 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
144 views
Explore Data Science
ML Research Hub
🔹 Title: Scaling Code-Assisted Chain-of-Thoughts and Instructions for Model Reasoning

🔹 Publication Date: Published on Oct 5

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.04081
• PDF: https://arxiv.org/pdf/2510.04081

🔹 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
142 views
Explore Data Science
ML Research Hub
🔹 Title: A Contextual Quality Reward Model for Reliable and Efficient Best-of-N Sampling

🔹 Publication Date: Published on Oct 5

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.04087
• PDF: https://arxiv.org/pdf/2510.04087

🔹 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
184 views
Explore Data Science
ML Research Hub
🔹 Title: DRIFT: Learning from Abundant User Dissatisfaction in Real-World Preference Learning

🔹 Publication Date: Published on Sep 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.02341
• PDF: https://arxiv.org/pdf/2510.02341

🔹 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
180 views
Explore Data Science
ML Research Hub
🔹 Title: Discrete Diffusion Models with MLLMs for Unified Medical Multimodal Generation

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06131
• PDF: https://arxiv.org/pdf/2510.06131
• Project Page: https://github.com/UCSC-VLAA/MeDiM
• Github: https://github.com/UCSC-VLAA/MeDiM

🔹 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
178 views
Explore Data Science
ML Research Hub
🔹 Title: VeriGuard: Enhancing LLM Agent Safety via Verified Code Generation

🔹 Publication Date: Published on Oct 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05156
• PDF: https://arxiv.org/pdf/2510.05156

🔹 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
162 views
Explore Data Science
ML Research Hub
🔹 Title: ASPO: Asymmetric Importance Sampling Policy Optimization

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06062
• PDF: https://arxiv.org/pdf/2510.06062
• Github: https://github.com/wizard-III/Archer2.0

🔹 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
130 views
Explore Data Science
ML Research Hub
🔹 Title: Mixing Mechanisms: How Language Models Retrieve Bound Entities In-Context

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06182
• PDF: https://arxiv.org/pdf/2510.06182
• Project Page: https://yoav.ml/blog/2025/mixing-mechs/
• Github: https://github.com/yoavgur/mixing-mechs

🔹 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
170 views
Explore Data Science
ML Research Hub
🔹 Title: CARE: Cognitive-reasoning Augmented Reinforcement for Emotional Support Conversation

🔹 Publication Date: Published on Sep 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.05122
• PDF: https://arxiv.org/pdf/2510.05122
• Project Page: https://github.com/aliyun/qwen-dianjin

🔹 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
167 views
Explore Data Science
ML Research Hub
🚀 Release day: Qwen launched Qwen3-Omni — the first native end-to-end *omni-modal AI*

The model processes text, images, audio, and video in a single model.

On benchmarks, it looks like all modalities work with equal quality.

⚡️ Features
- First place in 22 out of 36 audio and multimodal benchmarks
- Support for 119 text languages,
- Minimal latency — 211 ms
- Audio processing up to 30 minutes long
- Allows flexible customization via system prompts
- Built-in tool calling

🌟 Open-source releases
The company released three versions:
- Qwen3-Omni-30B-A3B-Instruct
- Qwen3-Omni-30B-A3B-Thinking
- Qwen3-Omni-30B-A3B-Captioner

👉 You can try it here:

💬 Chat: https://chat.qwen.ai/?models=qwen3-omni-flash

👨‍💻 GitHub: https://github.com/QwenLM/Qwen3-Omni

🤗 Hugging Face: https://huggingface.co/collections/Qwen/qwen3-omni-68d100a86cd0906843ceccbe

🤖 ModelScope: https://modelscope.cn/collections/Qwen3-Omni-867aef131e7d4f

🎬 Demo: https://huggingface.co/spaces/Qwen/Qwen3-Omni-Demo

#qwen #opensource #llm #ml
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
166 views
ML Research Hub
🔹 Title: Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?

🔹 Publication Date: Published on Oct 7

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2510.06036
• PDF: https://arxiv.org/pdf/2510.06036

🔹 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
139 views
Explore Data Science