Unverified60% confidenceFactExact time
The paper "Diffusion-LM Improves Controllable Text Generation" transfers diffusion models to discrete text for controllable generation and is widely recognized as the seminal work in the direction of diffusion language models.
2
Sources
60%
Confidence
Long-term
Relevance
9/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe underlying architecture of mainstream text-to-image models such as Stable Diffusion, Midjourney, and DALL-E typically combines a diffusion model with a CLIP text encoder.74% similarUnverifiedContinuous diffusion language models (dLLMs) are a class of models that apply diffusion models to text generation, generating text in a continuous latent space through a denoising process, and their latent-space-to-text mapping is learned rather than random.73% similarUnverifiedSplicing autoregressive text generation together with diffusion-based iterative denoising introduces a fundamental asymmetry in computational paradigms.72% similarUnverifiedContinuous Diffusion Language Models (CDLM) map text into a continuous embedding space to perform the diffusion process and then decode back to discrete text, rather than performing diffusion directly on discrete tokens.72% similarUnverifiedLLaDA-Image employs a Masked Diffusion mechanism that combines a discrete masked language model with a continuous diffusion process.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/843366API
curl https://kongchang.com/api/v1/knowledge/claims/843366MCP
get_claim(id=843366)