CoReDiT reduces self-attention FLOPs in DiTs by up to 55% via linear-time spatial coherence pruning and neighbor-based reconstruction, delivering 1.33x-1.72x speedups with maintained quality.
hub
Gans trained by a two time-scale update rule converge to a local nash equilib- rium.Advances in neural information processing systems, 30
15 Pith papers cite this work. Polarity classification is still indexing.
15
Pith papers citing it
hub tools
citation-role summary
method 2
background 1
citation-polarity summary
representative citing papers
C-MET transfers emotions from speech to facial video by learning cross-modal semantic vectors with pretrained audio and disentangled expression encoders, yielding 14% higher emotion accuracy on MEAD and CREMA-D even for unseen emotions.
Banana100 dataset shows that none of 21 popular NR-IQA metrics consistently rate images degraded by 100 iterative edits lower than clean originals.
IQPIR uses NR-IQA-derived quality scores to condition a Transformer and dual-branch codebook for perceptually superior real-world image restoration.
Face2Scene uses facial restoration as an oracle to derive degradation codes that condition a diffusion model for restoring the entire degraded scene.
Flow matching models follow a two-stage process of navigation across data modes then refinement to nearest samples, revealed by exact computation of the oracle marginal velocity field.
LIFT decomposes distillation into coarse linear alignment then fine refinement while PLACE adds error-based local adaptation, allowing stable training of 1.3M-parameter students (1.6% teacher size) to FID 15.73 across diffusion and flow models.
GOR-IS removes objects from 3D Gaussian Splatting reconstructions by performing inpainting in an intrinsic decomposition space that explicitly models light transport for consistent global lighting and non-Lambertian surfaces.
TICoE achieves more precise and faithful concept erasure in text-to-image models by collaborating text and image data through a convex manifold and hierarchical learning, outperforming prior methods.
BVE framework enables text-guided 3D editing beyond voxel limits by combining self-constructed data, lightweight semantic injection, and annotation-free masking to preserve local invariance.
RDVQ enables joint rate-distortion optimization for vector-quantized generative image compression via differentiable codebook distribution relaxation and an autoregressive entropy model.
O2MAG generates high-fidelity text-guided anomalies from a single image without training by manipulating self-attention in diffusion models with anomaly masks and dual enhancements.
DeCo decouples high- and low-frequency generation in pixel diffusion via a DiT plus lightweight decoder and a frequency-aware flow-matching loss, reaching FID 1.62 at 256x256 and 2.22 at 512x512 on ImageNet while closing the gap to latent diffusion methods.
A two-stage method predicts an intermediate Canny map for structure then renders the image conditioned on appearance and structure, paired with a 100k text-aware dataset, to improve detail preservation in subject-driven generation.
GaussianZoom enables high-fidelity extreme zoom-in 3D rendering from low-res inputs via an iterative framework combining geometry-consistent modeling, depth-based super-resolution, VLM detail synthesis, and an expandable continuous Level-of-Detail hierarchy.
citing papers explorer
-
CoReDiT: Spatial Coherence-Guided Token Pruning and Reconstruction for Efficient Diffusion Transformers
CoReDiT reduces self-attention FLOPs in DiTs by up to 55% via linear-time spatial coherence pruning and neighbor-based reconstruction, delivering 1.33x-1.72x speedups with maintained quality.
-
Cross-Modal Emotion Transfer for Emotion Editing in Talking Face Video
C-MET transfers emotions from speech to facial video by learning cross-modal semantic vectors with pretrained audio and disentangled expression encoders, yielding 14% higher emotion accuracy on MEAD and CREMA-D even for unseen emotions.
-
Banana100: Breaking NR-IQA Metrics by 100 Iterative Image Replications with Nano Banana Pro
Banana100 dataset shows that none of 21 popular NR-IQA metrics consistently rate images degraded by 100 iterative edits lower than clean originals.
-
Beyond Ground-Truth: Leveraging Image Quality Priors for Real-World Image Restoration
IQPIR uses NR-IQA-derived quality scores to condition a Transformer and dual-branch codebook for perceptually superior real-world image restoration.
-
Face2Scene: Using Facial Degradation as an Oracle for Diffusion-Based Scene Restoration
Face2Scene uses facial restoration as an oracle to derive degradation codes that condition a diffusion model for restoring the entire degraded scene.
-
From Navigation to Refinement: Revealing the Two-Stage Nature of Flow-based Diffusion Models through Oracle Velocity
Flow matching models follow a two-stage process of navigation across data modes then refinement to nearest samples, revealed by exact computation of the oracle marginal velocity field.
-
LIFT and PLACE: A Simple, Stable, and Effective Knowledge Distillation Framework for Lightweight Diffusion Models
LIFT decomposes distillation into coarse linear alignment then fine refinement while PLACE adds error-based local adaptation, allowing stable training of 1.3M-parameter students (1.6% teacher size) to FID 15.73 across diffusion and flow models.
-
GOR-IS: 3D Gaussian Object Removal in the Intrinsic Space
GOR-IS removes objects from 3D Gaussian Splatting reconstructions by performing inpainting in an intrinsic decomposition space that explicitly models light transport for consistent global lighting and non-Lambertian surfaces.
-
Beyond Text Prompts: Precise Concept Erasure through Text-Image Collaboration
TICoE achieves more precise and faithful concept erasure in text-to-image models by collaborating text and image data through a convex manifold and hierarchical learning, outperforming prior methods.
-
Beyond Voxel 3D Editing: Learning from 3D Masks and Self-Constructed Data
BVE framework enables text-guided 3D editing beyond voxel limits by combining self-constructed data, lightweight semantic injection, and annotation-free masking to preserve local invariance.
-
Differentiable Vector Quantization for Rate-Distortion Optimization of Generative Image Compression
RDVQ enables joint rate-distortion optimization for vector-quantized generative image compression via differentiable codebook distribution relaxation and an autoregressive entropy model.
-
One-to-More: High-Fidelity Training-Free Anomaly Generation with Attention Control
O2MAG generates high-fidelity text-guided anomalies from a single image without training by manipulating self-attention in diffusion models with anomaly masks and dual enhancements.
-
DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation
DeCo decouples high- and low-frequency generation in pixel diffusion via a DiT plus lightweight decoder and a frequency-aware flow-matching loss, reaching FID 1.62 at 256x256 and 2.22 at 512x512 on ImageNet while closing the gap to latent diffusion methods.
-
Decomposing Subject-Driven Image Generation via Intermediate Structural Prediction
A two-stage method predicts an intermediate Canny map for structure then renders the image conditioned on appearance and structure, paired with a 100k text-aware dataset, to improve detail preservation in subject-driven generation.
-
GaussianZoom: Progressive Zoom-in Generative 3D Gaussian Splatting with Geometric and Semantic Guidance
GaussianZoom enables high-fidelity extreme zoom-in 3D rendering from low-res inputs via an iterative framework combining geometry-consistent modeling, depth-based super-resolution, VLM detail synthesis, and an expandable continuous Level-of-Detail hierarchy.