Unverified50% confidenceFactExact time
PARC's structured pruning constraint requires that the pruned attention dimension be divisible by the target core count while keeping the query-to-key/value ratio and the per-head dimension unchanged.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedExisting structured pruning methods usually assign and rank importance weights at the MLP block or attention head level, with relatively coarse granularity, which loses the information about differences in importance among different neurons within an MLP block.72% similarUnverifiedDense attention layers require a complete KV Cache, whereas linear layers rely on hidden states.70% similarUnverified剪枝和量化在保持原架构拓扑的前提下做减法,不改变层的功能语义和连接模式;而Priming允许目标架构在层数、宽度、注意力模式上与原模型不同。69% similarUnverifiedThe principle of tiled processing is to split the feature map into sub-tiles along the spatial dimension for independent computation and then merge them; peak VRAM usage decreases linearly as the number of tiles increases, but overlapping sampling is required to avoid seams and the total computation increases slightly.68% similarVerifiedThe computational complexity of standard self-attention is quadratic with respect to sequence length, and achieving a context of one million tokens requires introducing techniques such as sparse attention, sliding window attention, and KV Cache compression.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130510API
curl https://kongchang.com/api/v1/knowledge/claims/1130510MCP
get_claim(id=1130510)