Unverified50% confidenceSolutionExact time
Diffusion-LM的可控生成通过引入超参数λ平衡流畅性与控制强度，并在每个扩散步执行多次梯度更新提升控制效果
1
Sources
50%
Confidence
Long-term
Relevance
9/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDiffusion Policy 将扩散模型引入机器人控制，通过迭代去噪生成动作序列，能处理动作空间的多峰分布69% similarUnverifiedDALL·E 2和Stable Diffusion均基于扩散模型原理69% similarVerifiedStable Diffusion是开源模型，可以本地部署，支持LoRA微调、ControlNet精确控制等高级功能68% similarUnverified在五项分类器引导任务上，Diffusion-LM的控制成功率全面高于PPLM和FUDGE基线，同时保持较好的流畅度68% similarVerifiedStable Diffusion is built on the Latent Diffusion Model (LDM) architecture67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/843372API
curl https://kongchang.com/api/v1/knowledge/claims/843372MCP
get_claim(id=843372)