Search:
Match:
4 results
Research Paper#AI Acceleration, Diffusion Models, Transformer Networks🔬 ResearchAnalyzed: Jan 3, 2026 15:47

CorGi: Accelerating Diffusion Transformers with Caching

Published:Dec 30, 2025 12:55
1 min read
ArXiv

Analysis

This paper addresses the computational cost of Diffusion Transformers (DiT) in visual generation, a significant bottleneck. By introducing CorGi, a training-free method that caches and reuses transformer block outputs, the authors offer a practical solution to speed up inference without sacrificing quality. The focus on redundant computation and the use of contribution-guided caching are key innovations.
Reference

“CorGi and CorGi+ achieve up to 2.0x speedup on average, while preserving high generation quality.”

PermalinkArXiv
Paper#AI/Generative Models/Attention Mechanisms🔬 ResearchAnalyzed: Jan 3, 2026 15:54

RainFusion2.0: Hardware-Efficient Sparse Attention for Video and Image Generation

Published:Dec 30, 2025 08:55
1 min read
ArXiv

Analysis

This paper addresses the computational bottlenecks of Diffusion Transformer (DiT) models in video and image generation, particularly the high cost of attention mechanisms. It proposes RainFusion2.0, a novel sparse attention mechanism designed for efficiency and hardware generality. The key innovation lies in its online adaptive approach, low overhead, and spatiotemporal awareness, making it suitable for various hardware platforms beyond GPUs. The paper's significance lies in its potential to accelerate generative models and broaden their applicability across different devices.
Reference

“RainFusion2.0 can achieve 80% sparsity while achieving an end-to-end speedup of 1.5~1.8x without compromising video quality.”

PermalinkArXiv
Research#Diffusion🔬 ResearchAnalyzed: Jan 10, 2026 10:53

SDAR-VL: Stable and Efficient Block-wise Diffusion for Vision-Language Advancements

Published:Dec 16, 2025 04:12
1 min read
ArXiv

Analysis

This research paper introduces SDAR-VL, focusing on improving the efficiency and stability of diffusion models in the domain of vision-language understanding. The study's focus on block-wise diffusion suggests a potential for significant performance gains and broader applicability.
Reference

“The paper focuses on Stable and Efficient Block-wise Diffusion.”

PermalinkArXiv
Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 08:15

BAMBO: Construct Ability and Efficiency LLM Pareto Set via Bayesian Adaptive Multi-objective Block-wise Optimization

Published:Dec 10, 2025 15:32
1 min read
ArXiv

Analysis

This article introduces BAMBO, a method for optimizing Large Language Models (LLMs) to achieve a Pareto set balancing ability and efficiency. The approach uses Bayesian optimization and block-wise optimization, suggesting a focus on computational efficiency and model performance trade-offs. The source being ArXiv indicates this is a research paper.
Reference

“”

PermalinkArXiv