Post-Training Pruning for Diffusion Transformers
Diffusion Transformers (DiTs) have demonstrated impressive performance in image generation but suffer from substantial computational overhead and resource consumption. Post-training pruning offers a promising solution; however, due to DiTs' unique architectural design and parameter distribution, traditional pruning methods are inapplicable, leading to significant performance degradation. Specifically, prior methods developed for LLMs, which derive metrics through a series of approximations, amplify the relative contribution of weights in the saliency metric. In addition, weights in DiTs exhibi
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- LinkedLinked via unknownquant-kit →
- LinkedLinked via unknownDiffuse-XL →
- LinkedLinked via unknownApparently you can skip entire transformer blocks at load time with minimal performance impact →
- FuzzySimilar title/name (fuzzy) · 59%CompVis/stable-diffusion-v1-4 →
“Fuzzy title match (0.73): “Post-Training Pruning for Diffusion Transformers” ≈ “CompVis/stable-diffusion-v1-4””
- PossiblePossibly related (embedding) · 51%bghira/SimpleTuner →
- FuzzySimilar title/name (fuzzy) · 87%lucidrains/x-transformers →
“Fuzzy title match (0.94): “Post-Training Pruning for Diffusion Transformers” ≈ “lucidrains/x-transformers””
- FuzzySimilar title/name (fuzzy) · 84%huggingface/transformers →
“Fuzzy title match (0.92): “Post-Training Pruning for Diffusion Transformers” ≈ “huggingface/transformers””