DiffusionBlocks: Blockwise Training for Generative Models via Score-Based Diffusion Paper • 2506.14202 • Published Jun 17 • 3
ALE-Bench: A Benchmark for Long-Horizon Objective-Driven Algorithm Engineering Paper • 2506.09050 • Published Jun 10 • 7
Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization Paper • 2502.19261 • Published Feb 26 • 7