Unverified50% confidenceDecision RuleExact time
These theoretical results suggest that the choice of LoRA rank should refer to the bottleneck rank, and that beyond the bottleneck rank threshold, extra rank may not bring observable changes in output.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedThe study proves that for the factorized LoRA objective with weight decay, after an arbitrarily small generic positive semidefinite perturbation, every first-order stationary point is rank-deficient as long as the LoRA rank exceeds the threshold determined by the bottleneck rank.78% similarUnverifiedVRAM requirements do not grow linearly; each time a model scale threshold is crossed, the required VRAM jumps up by a step.73% similarUnverifiedIn ultra-long context inference scenarios, the performance bottleneck is often not compute but the KV cache itself.72% similarUnverified误差累积问题往往源于参照系选择而非单纯模型容量限制72% similarVerifiedSetting the threshold for semantic caching involves a trade-off: a threshold that is too high results in a low hit rate, while one that is too low may incorrectly map semantically different questions and cause response quality issues.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069416API
curl https://kongchang.com/api/v1/knowledge/claims/1069416MCP
get_claim(id=1069416)