✨OmniLottie: Generating Vector Animations via Parameterized Lottie Tokens

📝 Summary:
OmniLottie framework generates high-quality vector animations from multi-modal instructions using a specialized Lottie tokenizer and pretrained vision-language models. AI-generated summary Omni Lottie...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02138
• PDF: https://arxiv.org/pdf/2603.02138
• Project Page: https://openvglab.github.io/OmniLottie/
• Github: https://github.com/OpenVGLab/OmniLottie

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨LaSER: Internalizing Explicit Reasoning into Latent Space for Dense Retrieval

📝 Summary:
LaSER introduces a self-distillation framework that embeds explicit reasoning into dense retrievers' latent space through dual-view training and multi-grained alignment, enabling efficient reasoning w...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01425
• PDF: https://arxiv.org/pdf/2603.01425

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨When Does RL Help Medical VLMs? Disentangling Vision, SFT, and RL Gains

📝 Summary:
Reinforcement learning enhances medical vision-language model performance primarily by sharpening output distributions when models already have sufficient reasoning support, with supervised fine-tunin...

🔹 Publication Date: Published on Mar 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01301
• PDF: https://arxiv.org/pdf/2603.01301
• Project Page: https://medbridgerl.github.io/
• Github: https://github.com/armenjeddi/medbridgerl

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨RAISE: Requirement-Adaptive Evolutionary Refinement for Training-Free Text-to-Image Alignment

📝 Summary:
RAISE is a training-free, requirement-driven evolutionary framework that adaptively improves text-to-image generation by dynamically allocating computational resources based on prompt complexity throu...

🔹 Publication Date: Published on Feb 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.00483
• PDF: https://arxiv.org/pdf/2603.00483
• Github: https://github.com/LiyaoJiang1998/RAISE

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨CharacterFlywheel: Scaling Iterative Improvement of Engaging and Steerable LLMs in Production

📝 Summary:
CharacterFlywheel is an iterative optimization process that enhances large language models for social chat applications through multiple generations of refinement, achieving significant improvements i...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01973
• PDF: https://arxiv.org/pdf/2603.01973

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Agentic Code Reasoning

📝 Summary:
LLM agents can perform code reasoning tasks like patch verification, fault localization, and code QA with improved accuracy through structured semi-formal reasoning that requires explicit premises and...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01896
• PDF: https://arxiv.org/pdf/2603.01896

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨FireRed-OCR Technical Report

📝 Summary:
FireRed-OCR transforms general vision-language models into specialized OCR systems through structured data synthesis and progressive training strategies. AI-generated summary We present FireRed-OCR, a...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01840
• PDF: https://arxiv.org/pdf/2603.01840
• Github: https://github.com/FireRedTeam/FireRed-OCR

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨MMR-Life: Piecing Together Real-life Scenes for Multimodal Multi-image Reasoning

📝 Summary:
MMR-Life is a new benchmark assessing multimodal large language models reasoning across real-life scenarios using diverse multi-image questions. It features 2,646 questions on 19,108 real-world images covering seven reasoning types. Top models like GPT-5 only achieve 58 percent accuracy, showing ...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02024
• PDF: https://arxiv.org/pdf/2603.02024
• Project Page: https://mmr-life-bench.github.io/
• Github: https://github.com/BugMakerzzz/MMR-Life

✨ Datasets citing this paper:
• https://huggingface.co/datasets/Septzzz/MMR-Life

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨CoVe: Training Interactive Tool-Use Agents via Constraint-Guided Verification

📝 Summary:
CoVe is a post-training data synthesis framework that generates high-quality training trajectories for interactive tool-use agents by incorporating task constraints as verification mechanisms, achievi...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01940
• PDF: https://arxiv.org/pdf/2603.01940
• Project Page: https://cove-agent.github.io

🔹 Models citing this paper:
• https://huggingface.co/Zichen1024/CoVe-4B

✨ Datasets citing this paper:
• https://huggingface.co/datasets/Zichen1024/CoVe-12k

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Learn Hard Problems During RL with Reference Guided Fine-tuning

📝 Summary:
Reference-Guided Fine-Tuning (ReGFT) addresses reward sparsity in reinforcement learning for mathematical reasoning by using human-written solutions to create guided training trajectories that improve...

🔹 Publication Date: Published on Mar 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01223
• PDF: https://arxiv.org/pdf/2603.01223

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Tool Verification for Test-Time Reinforcement Learning

📝 Summary:
Test-time reinforcement learning with tool verification addresses consensus bias in large reasoning models by using external validation to improve reward estimation and model stability. AI-generated s...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02203
• PDF: https://arxiv.org/pdf/2603.02203

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ArtLLM: Generating Articulated Assets via 3D LLM

📝 Summary:
ArtLLM generates articulated 3D assets from meshes using a 3D multimodal large language model that predicts part layouts and joints while synthesizing high-fidelity geometries. AI-generated summary Cr...

🔹 Publication Date: Published on Mar 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01142
• PDF: https://arxiv.org/pdf/2603.01142

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨MicroVerse: A Preliminary Exploration Toward a Micro-World Simulation

📝 Summary:
Current video generation models struggle with microscale simulation tasks, prompting the development of MicroVerse, a specialized video generation model trained on expert-verified simulation data to a...

🔹 Publication Date: Published on Feb 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.00585
• PDF: https://arxiv.org/pdf/2603.00585

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Recursive Think-Answer Process for LLMs and VLMs

📝 Summary:
Recursive Think-Answer Process enables iterative reasoning cycles that improve accuracy and reduce self-reflective errors in language and vision-language models through confidence-based reinforcement ...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02099
• PDF: https://arxiv.org/pdf/2603.02099
• Project Page: https://litcoderr.github.io/rtap_page/

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨From Scale to Speed: Adaptive Test-Time Scaling for Image Editing

📝 Summary:
Image-CoT methods are extended to image editing with ADE-CoT, which improves efficiency and performance through adaptive resource allocation, edit-specific verification, and opportunistic stopping mec...

🔹 Publication Date: Published on Feb 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.00141
• PDF: https://arxiv.org/pdf/2603.00141

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨CHIMERA: Compact Synthetic Data for Generalizable LLM Reasoning

📝 Summary:
A synthetic reasoning dataset called CHIMERA is introduced to overcome data-centric challenges in training large language models for cross-domain reasoning, achieving performance comparable to much la...

🔹 Publication Date: Published on Mar 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.00889
• PDF: https://arxiv.org/pdf/2603.00889
• Project Page: https://huggingface.co/datasets/TianHongZXY/CHIMERA

✨ Datasets citing this paper:
• https://huggingface.co/datasets/TianHongZXY/CHIMERA

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨SeeThrough3D: Occlusion Aware 3D Control in Text-to-Image Generation

📝 Summary:
SeeThrough3D generates 3D layout-conditioned scenes with explicit occlusion modeling using translucent 3D boxes and visual tokens derived from rendered representations. AI-generated summary We identif...

🔹 Publication Date: Published on Feb 26

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2602.23359
• PDF: https://arxiv.org/pdf/2602.23359

🔹 Models citing this paper:
• https://huggingface.co/va1bhavagrawa1/seethrough3d

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

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

#AI #DataScience #MachineLearning #HuggingFace #Research

✨WorldStereo: Bridging Camera-Guided Video Generation and Scene Reconstruction via 3D Geometric Memories

📝 Summary:
WorldStereo integrates camera-guided video generation and 3D reconstruction using geometric memory modules. These provide camera control and structural priors for multi-view consistent videos, enabling high-quality 3D scene reconstruction.

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02049
• PDF: https://arxiv.org/pdf/2603.02049
• Project Page: https://3d.hunyuan.tencent.com/sceneTo3D
• Github: https://github.com/FuchengSu/WorldStereo

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

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

#VideoGeneration #3DReconstruction #ComputerVision #DeepLearning #NeuralRendering

✨Reasoning Core: A Scalable Procedural Data Generation Suite for Symbolic Pre-training and Post-Training

📝 Summary:
Reasoning Core is a new scalable system that procedurally generates verifiable symbolic reasoning data across diverse formal domains. Mixing this data into pre-training improves language model reasoning abilities while preserving language modeling quality.

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.02208
• PDF: https://arxiv.org/pdf/2603.02208
• Project Page: https://github.com/sileod/reasoning_core/
• Github: https://github.com/sileod/reasoning_core

✨ Datasets citing this paper:
• https://huggingface.co/datasets/reasoning-core/symbolic-pretraining-pile
• https://huggingface.co/datasets/reasoning-core/symbolic-reasoning-env

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

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

#AI #LLM #SymbolicReasoning #DataGeneration #MachineLearning

✨Spectral Attention Steering for Prompt Highlighting

📝 Summary:
SEKA and AdaSEKA introduce training-free attention steering by editing key embeddings using spectral decomposition. This amplifies attention for specific tokens, outperforming baselines with less memory and latency, compatible with optimized attention.

🔹 Publication Date: Published on Mar 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01281
• PDF: https://arxiv.org/pdf/2603.01281
• Github: https://github.com/waylonli/SEKA

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

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

#AttentionMechanisms #NLP #DeepLearning #MachineLearning #AI

✨OpenAutoNLU: Open Source AutoML Library for NLU

📝 Summary:
OpenAutoNLU is an open-source AutoML library for NLU tasks like text classification and named entity recognition. Its key innovation is data-aware training selection requiring no manual configuration. It also offers integrated diagnostics, out-of-distribution detection, and LLM features through a...

🔹 Publication Date: Published on Mar 2

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2603.01824
• PDF: https://arxiv.org/pdf/2603.01824
• Project Page: https://openautonlu.dev
• Github: https://github.com/mts-ai/OpenAutoNLU

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

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

#AutoML #NLU #LLM #OpenSource #MachineLearning