Unverified50% confidenceFactExact time
Masked diffusion randomly masks some tokens in the sequence during training, and during generation starts from the masked state and gradually denoises through multiple iterations to restore the complete sequence, with each step able to predict tokens at multiple positions in parallel.
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA single-step generation in discrete diffusion can truly match the training data distribution only when the positions it writes to are mutually independent given the already-fixed tokens.77% similarVerifiedAutoregressive generation refers to the model predicting the next token each time based on all previously generated tokens, progressively unfolding until a complete code block is generated72% similarUnverifiedThe k-partition probing mechanism randomly divides committed token positions into k non-overlapping subsets, re-masking one subset at a time and performing one forward inference pass.72% similarUnverifiedDiffusionGemma采用整片并行去噪的生成机制，先铺开256个位置的噪声，然后逐轮迭代，将置信度最高的位置先固定，其余位置重新注入噪声72% similarUnverifiedDiffusionGemma使用双向注意力，去掉因果掩码，让每个位置能同时关注序列中所有其他位置，机制与BERT编码器相同71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1084316API
curl https://kongchang.com/api/v1/knowledge/claims/1084316MCP
get_claim(id=1084316)