Unverified50% confidenceFactExact time
The null space refers to the "safe directions" in model parameter updates that do not affect the original task.
1
Sources
50%
Confidence
Long-term
Relevance
11/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedNull-space projection methods in knowledge editing only constrain the overall effect at the activation-space level, do not distinguish differences in the importance of parameters themselves to predictions, and their null-space estimation relies on limited samples, so the constraint may fail when generalizing to unseen inputs.71% similarUnverifiedDynamic low-rank approximation is used to expand the null space when the common null space is insufficient.66% similarUnverifiedOmniST's computation has zero dependence on the number of sensor nodes, and the model weight shapes do not change with the number of nodes N.62% similarUnverifiedThe exit code of this evaluation gate module returns nonzero only when a case has zero passes or still fails after an upgrade, to avoid a noisy suite blocking deployments that should not be blocked.61% similarUnverifiedThe fundamental assumption of low-dimensional parameter mapping is that effective parameter updates tend to concentrate in a subspace far smaller than the full parameter space.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1165344API
curl https://kongchang.com/api/v1/knowledge/claims/1165344MCP
get_claim(id=1165344)