Unverified50% confidenceSolutionExact time
生产中典型做法是先用向量检索和BM25快速召回50-100个候选结果，再用reranker精排取top-k送入LLM
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/29/2026
First Seen
Valid until: 11/27/2026
Sources
Related Entities
Related Claims
UnverifiedMeshLLM claims to support more than 40 models, with parameter scales ranging from 5.1B to 235B (MoE).60% similarUnverifiedQwen3.6-27B's benchmark scores place it at the leading position among models of similar scale (20B–30B parameter range).59% similarUnverified在27B混合架构、4B稠密模型、36B MoE三类模型上以8k上下文测试，Prefill阶段性能无变化，Decode阶段打补丁后快约1.4%58% similarVerifiedBM25 (Best Matching 25) is a probabilistic ranking algorithm proposed by Stephen Robertson and others in the 1990s, scoring relevance based on three factors: term frequency, inverse document frequency, and document length normalization.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/817053API
curl https://kongchang.com/api/v1/knowledge/claims/817053MCP
get_claim(id=817053)