Unverified50% confidenceFactExact time
Binary Discrete Diffusion represents the candidate token set at each position with a binary membership vector and gradually resolves uncertainty through add and remove operations.
1
Sources
50%
Confidence
Long-term
Relevance
10/21/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe separation result generalizes to weighted fixed-count targets, as long as the weight ratios are bounded and the quota proportions are bounded away from 0 and 1, the separation holds even without permutation symmetry.64% similarUnverifiedIn the unfused small-operator mode, Cube computation and Vector computation are often arranged serially, causing one type of compute unit to be idle at any given moment.64% similarUnverifiedDifferentiable assignment relaxes discrete point-assignment decisions into continuous weights (such as softmax-normalized assignment probabilities), allowing gradients to be backpropagated to the assignment parameters.63% similarUnverifiedMatrix multiplication is typically compute-bound in high-batch inference scenarios, while autoregressive decoding with a batch size of 1 is almost invariably bandwidth-bound.62% similarUnverifiedDiffusionGemma 大概率采用掩码扩散方案，前向过程逐步将 token 替换为 [MASK] 标记62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1078871API
curl https://kongchang.com/api/v1/knowledge/claims/1078871MCP
get_claim(id=1078871)