Unverified50% confidenceSolutionExact time
The study proposes correcting CFG in the early flow steps: reducing or reweighting the CFG influence during the noise stage, then gradually restoring standard CFG; this method requires no retraining of the model and adds no inference cost, being completed entirely within the sampling stage.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe 'noise stage' or 'early flow steps' of flow matching refers to the first few steps of integration where t is close to 0 and the sample is still close to pure noise; at this stage the image structure information has not yet taken shape, and this is exactly the time window in which CFG behavior becomes anomalous.71% similarUnverifiedClassifier-free guidance (CFG) runs both a conditional branch and an unconditional branch simultaneously, amplifying the conditional signal by subtracting the unconditional direction from the conditional direction, trading a doubling of inference cost for quality improvement.70% similarUnverified负向提示词在Stable Diffusion等开源模型中通过Classifier-Free Guidance（CFG）机制实现，CFG Scale通常设置在7-12之间69% similarUnverifiedUsing excessively high CFG values during low-step inference amplifies the correction magnitude at each step, leading to oversaturated colors, distorted contrast, and even artifacts; when using an accelerated LoRA, the CFG value often needs to be reduced from 7.5 to 2–4.68% similarUnverifiedCFG=1 means no interpolation amplification is applied at all, causing the generation process to degrade into pure unconditional diffusion, rendering the prompt ineffective as a guide.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046700API
curl https://kongchang.com/api/v1/knowledge/claims/1046700MCP
get_claim(id=1046700)