WorldDiT: A Unified Architecture for World and Action Modeling
WorldDiT is a pareto frontier robot world-action model architecture without a VLM backbone
Everything Bagel
Bagel Labs is a distributed machine learning research lab.
Live Last read · last published · next check
Latest posts
Introducing Paris 2.0
A Decentralized Diffusion Model for Video Generation
Heterogeneous Decentralized Diffusion Models
Training free merge of heterogeneous objective diffusion ensemble
Stability–Quality Paradox in Decentralized Diffusion Models
Why the most stable samplers in decentralized diffusion produce the worst generations
Introducing Paris
World's First Decentralized Trained Open-Weight Diffusion Model
Tiny Tool Use
Open source library for teaching LLMs tool-use
Return on Experience (RoE)
A single benchmark that predicts Reinforcement Learning's future.
ZKLoRA
Zero-Knowledge Verification of LoRA training on SOTA Models in 1-2 Seconds
Train Fast, But Think Slow
Inference Time Compute vs Training Time Compute for AI Reasoning