[Survey] Post-Training of Modern LLMs

Modern LLM post-training, from preference learning to reinforcement learning with verifiable rewards.   ·   llm   rl  

[Survey] One-step Generation in the Post Diffusion Era

Training-based routes toward one-step generation for diffusion and flow models, covering distillation, Consistency Models, CTM, MeanFlow, DMD, and Drifting Models.   ·   generative  

[Paper Review] Designing Diffusion Models in Real World

Elucidating the Design Space of Diffusion-Based Generative Models, focusing on the rationale behind its engineering details.   ·   generative  

[Survey] Understanding Diffusion Models in Two Perspectives

DDPM and score-based SDEs as two routes to reverse diffusion, contrasting their shared structure and modeling differences.   ·   generative