https://huggingface.co/papers/2512.05106
AI Summary: The paper introduces a Phase-Preserving Diffusion (φ-PD) technique that enhances diffusion model performance by preserving spatial structure through phase retention, improving tasks like image/video re-rendering and simulation-to-real transfer without extra computational cost or architecture changes; notably, φ-PD boosts driving planner performance in the CARLA simulator by 50%.
https://huggingface.co/papers/2511.13720
AI Summary: A new research paper proposes a Transformer-based generative model called JiT that directly predicts clean image data—unlike current denoising diffusion models that predict noise—leveraging manifold assumptions for improved high-dimensional image generation without tokenizers or pre-training.
https://huggingface.co/papers/2511.08923
AI Summary: TiDAR is a hybrid diffusion–autoregressive model that improves throughput while maintaining autoregressive-level quality by using diffusion-based drafting and AR sampling in a single forward pass; it outperforms speculative decoding and diffusion models in efficiency and quality.
https://huggingface.co/papers/2601.20802
AI Summary: Introduces Self-Distillation Policy Optimization (SDPO) for RL with rich feedback by converting tokenized feedback into a dense learning signal, improving sample efficiency and accuracy vs scalar-reward methods, and outperforming baselines in challenging tasks.
https://towardsdatascience.com/deep-reinforcement-learning-the-actor-critic-method/
AI Summary: Explains Actor–Critic RL with separate actor/critic networks, showing improved learning stability and efficiency (example: drone landing), and discusses practical pitfalls like TD target detachment and reward engineering.