Unverified50% confidenceSolutionExact time
SAFE-QE scores each candidate expansion term based on its cross-context reproducibility conditioned on the corpus: if a term repeatedly appears in randomized query contexts yet finds no support in the pseudo-relevant document set, it is likely the product of model-driven hallucinated associations and is down-weighted.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedOn learned sparse retrievers, all expansion methods exhibited performance regression, corroborating concerns that query expansion may harm strong retrievers, while SAFE-QE still maintained a low hallucination rate in this scenario.75% similarUnverifiedIn real-world testing, the QA Agent discovered boundary risks that are difficult for human testers to think of, such as the possibility of experience points being incorrectly awarded during a program's transaction rollback.65% similarUnverifiedQ4 quantization is sufficient for casual conversation and code completion, but for scenarios involving rigorous analysis, long document processing, or the need to reduce hallucinations, switching to Q6 or Q8 versions is recommended.62% similarUnverified通用越狱指单一提示词或操作序列能系统性绕过模型所有安全约束，而针对性越狱只能解锁某一类具体有害行为62% similarUnverifiedOnce a benchmark is made public and repeatedly attacked, targeted overfitting may occur—that is, learning the distribution patterns of the problems without necessarily corresponding to underlying generalized intelligence.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002084API
curl https://kongchang.com/api/v1/knowledge/claims/1002084MCP
get_claim(id=1002084)