Paper Reviews by AI

LVAS-Agent: Multi-agent system conquers long-video audio synthesis with collaborative dubbing, script, design, & more!

LCT: Fine-tunes single-shot video diffusion models for coherent multi-shot video generation without extra parameters!

Light-R1: Trains long COT models from scratch using curriculum SFT, DPO, and RL, achieving SOTA performance and strong generalization with limited resources.

LHM: Animatable 3D avatars from a single image in seconds.

GROVE: Pre-training on large-scale data for grounded video caption generation.

KUDA unifies dynamics learning and visual prompting with keypoints for open-vocabulary robot manipulation.

KArAt: Can Learnable Attention Beat Standard Attention in Vision Transformers?

GroundingSuite: A new benchmark that measures complex multi-granular pixel grounding to overcome current dataset limitations and push forward vision-language understanding.

GoT: Reasoning guides vivid image generation and editing!

SPIRE: Adds speech to text-only LLMs, maintaining text performance via discretized speech and continued pre-training.

FlowTok: Seamlessly flows across text and image tokens!

ETCH: Equivariantly fitting bodies to clothed humans through tightness for better pose and shape accuracy.

Distilling diffusion models?💡 This paper shows you how to retain base model diversity while keeping the distilled model’s speed!

COSTA*: A cost-effective agent that smartly navigates AI tools to edit images with high quality and low cost, balancing user preferences!

CINEMA: MLLM-guided coherent multi-subject video generation for consistent and controllable content creation.

Navigating millions of models is hard. This paper charts Hugging Face, revealing model relationships and attribute predictions.

ARPG: Randomly generate high-quality images by parallel decoding, outperforming existing methods in efficiency, memory, and quality.

4D LangSplat learns 4D language fields for dynamic scenes using multimodal large language models, enabling time-sensitive open-vocabulary queries.

TPDiff accelerates video diffusion by progressively increasing frame rates during diffusion, optimizing computational efficiency with a novel stage-wise training strategy.

New ‘Silent Branding Attack’ poisons text-to-image models, embedding brand logos without text prompts, raising ethical issues for image generation tools.