Unverified50% confidenceFactExact time
The output layer of classification/decision-type models typically has only a fixed number of categories and does not have the ability to score across an open vocabulary.
1
Sources
50%
Confidence
Long-term
Relevance
10/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTraditional LLM routing typically decides which model to use based solely on the individual request itself (e.g., text length, complexity, or domain classification), making it a stateless, isolated judgment.73% similarUnverified模型每词单向量只能捕捉主导词义，多义词处理能力有限，且只能做相似度而非推理或事实记忆72% similarUnverifiedGeneral benchmarks measure the upper bound of a model's general capabilities, not its actual performance under specific prompt templates and specific output format requirements.71% similarUnverifiedWhen constraining output format through prompts, the model may forget these constraints in long conversations or complex reasoning scenarios.71% similarUnverifiedBERT分类头线性层的输出维度严格等于类别数，一旦num_classes固定，模型输出空间被锁死，无法在不修改网络结构的前提下直接预测新增类别70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/965191API
curl https://kongchang.com/api/v1/knowledge/claims/965191MCP
get_claim(id=965191)