Unverified50% confidenceFactExact time
DepthGraft reconstructs deep-layer KV from stride-aligned shallow-to-middle-layer source pairs, distributing non-local gradient flow across multiple source layers.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDepthGraft preserves the early-completion property of prefill, i.e., deep-layer KV can be reconstructed from KV already computed in shallow and middle layers, thereby shortening first-token latency.76% similarUnverifiedThe paper "DepthGraft: Structural Regularization through Hierarchical Cross-Layer KV Reconstruction" proposes a hierarchical KV sharing architecture named DepthGraft.73% similarUnverifiedThe mainstream approaches to depth estimation fall into two main categories: geometry-based multi-view stereo (MVS) and deep-learning-based monocular depth estimation.62% similarUnverified深度生成模型通过学习高维数据中的低维流形结构来规避维度灾难62% similarUnverifiedDepth Anything、MiDaS、DPT等基础模型大幅提升了相对深度质量,但输出仍为无尺度的相对值,需要额外锚定手段还原绝对度量62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130756API
curl https://kongchang.com/api/v1/knowledge/claims/1130756MCP
get_claim(id=1130756)