Adapting Vision-Language Models for E-commerce Understanding at Scale

ThinkRouter: Efficient Reasoning via Routing Thinking between Latent and Discrete Spaces

ScalSelect: Scalable Training-Free Multimodal Data Selection for Efficient Visual Instruction Tuning

ABot-N0: Technical Report on the VLA Foundation Model for Versatile Embodied Navigation

Pretraining A Large Language Model using Distributed GPUs: A Memory-Efficient Decentralized Paradigm

Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Use

Multimodal Fact-Level Attribution for Verifiable Reasoning

MolmoSpaces: A Large-Scale Open Ecosystem for Robot Navigation and Manipulation

Voxtral Realtime

GameDevBench: Evaluating Agentic Capabilities Through Game Development

MOSS-Audio-Tokenizer: Scaling Audio Tokenizers for Future Audio Foundation Models

Use A2A to connect agents across different frameworks and teams

How Do Decoder-Only LLMs Perceive Users? Rethinking Attention Masking for User Representation Learning

Neural Additive Experts: Context-Gated Experts for Controllable Model Additivity

MetaphorStar: Image Metaphor Understanding and Reasoning with End-to-End Visual Reinforcement Learning

LiveMedBench: A Contamination-Free Medical Benchmark for LLMs with Automated Rubric Evaluation

Latent Thoughts Tuning: Bridging Context and Reasoning with Fused Information in Latent Tokens

Andrew Ng on Vibe Coding

Stemphonic: All-at-once Flexible Multi-stem Music Generation

Proposes method for multi-stem music generation with flexible instrument control. Relevant to ML research but outside core AI agent/LLM focus.

The Devil Behind Moltbook: Anthropic Safety is Always Vanishing in Self-Evolving AI Societies

χ_{0}: Resource-Aware Robust Manipulation via Taming Distributional Inconsistencies

StealthRL: Reinforcement Learning Paraphrase Attacks for Multi-Detector Evasion of AI-Text Detectors

Why are diffusion LLMs so fast?

PISCO: Precise Video Instance Insertion with Sparse Control

Dreaming in Code for Curriculum Learning in Open-Ended Worlds

Graph-Enhanced Deep Reinforcement Learning for Multi-Objective Unrelated Parallel Machine Scheduling

Applies deep reinforcement learning with graph neural networks to optimize parallel machine scheduling. Relevant to ML research but not directly related to LLMs or AI agents.

MemFly: On-the-Fly Memory Optimization via Information Bottleneck

From Features to Actions: Explainability in Traditional and Agentic AI Systems

The Biggest Mistake AI Beginners Make

The Boring Way to Learn AI (That Actually Works)

Sparse Video Generation Propels Real-World Beyond-the-View Vision-Language Navigation

Unveiling Implicit Advantage Symmetry: Why GRPO Struggles with Exploration and Difficulty Adaptation

Reasoning Cache: Continual Improvement Over Long Horizons via Short-Horizon RL

An image is worth NxN words | Diffusion Transformers (ViT, DiT, MMDiT)

TIC-VLA: A Think-in-Control Vision-Language-Action Model for Robot Navigation in Dynamic Environments

ECHO-2: A Large-Scale Distributed Rollout Framework for Cost-Efficient Reinforcement Learning

Learn to equip AI agents with reusable skills

Is vibe coding real coding?

Unlock data from your files with Agentic Document Extraction

Document AI: How Agents add the brain to OCR's eyes

The Physics of Diffusion Models

Text Diffusion: A new LLM paradigm

I asked them to show me their RAG pipeline...

Transformers & Diffusion LLMs: What's the connection?

Text diffusion: A new paradigm for LLMs

The physics behind diffusion models

Reverse-engineering GGUF | Post-Training Quantization

Training models with only 4 bits | Fully-Quantized Training

The myth of 1-bit LLMs | Quantization-Aware Training

How LLMs survive in low precision | Quantization Fundamentals