🏢 Peking University

This survey presents a methodology-centered taxonomy of LLM agent systems, linking design principles to emergent behaviors and identifying future research directions.

Bottleneck Sampling: Accelerate diffusion models without retraining by cleverly using low-resolution priors for efficient inference!

ActVLP: Enhancing VLMs through visual-linguistic guidance for superior action-based decision-making in interactive environments.

MagicComp: Dual-Phase Refinement Enables Training-Free Compositional Video Generation

WideRange4D: A new benchmark & reconstruction method for high-quality 4D scenes with wide-range movements, pushing the boundaries of 4D reconstruction.

BlobCtrl: Precisely edit images at the element level with a unified, flexible framework, bridging the gap between generation and editing.

Being-0: A humanoid robot agent achieves complex tasks by integrating a vision-language model with modular skills, enhancing efficiency and real-time performance.

UniFace: a novel UMM tailored for fine-grained face understanding and generation.

SBD: Self-supervised skill discovery from unsegmented videos!

WISE: Evaluates world knowledge in text-to-image generation.

TinyR1-32B-Preview: A novel branch-merge distillation approach that significantly enhances model accuracy and reduces computational costs for LLMs.

LONGCODEU: A new benchmark to challenge & enhance long code understanding in language models for software engineering!

SoRFT enhances LLMs for issue resolving via subtask-oriented reinforced fine-tuning, outperforming other open-source models.

DREAM ENGINE: Text-image interleaved control made easy, unifying text and visual cues for creative image generation.

HermesFlow seamlessly bridges the understanding-generation gap in MLLMs using a novel Pair-DPO framework and self-play optimization on homologous data, achieving significant performance improvements.

Diffusion-Sharpening enhances diffusion model fine-tuning by optimizing sampling trajectories, achieving faster convergence and high inference efficiency without extra NFEs, leading to improved alignm…

HealthGPT: A novel medical vision-language model unifying comprehension and generation via heterogeneous knowledge adaptation, achieving superior performance on various medical tasks.

Next-Block Prediction (NBP) revolutionizes video generation by using a semi-autoregressive model that predicts blocks of video content simultaneously, resulting in significantly faster inference.

Magic141 generates one-minute video clips in under a minute by cleverly factorizing the generation task and employing optimization techniques.

InferenceGuard ensures almost-sure safe LLM responses at inference time by framing safe generation as a constrained Markov Decision Process in the LLM’s latent space, achieving high safety rates witho…