Unverified50% confidenceFactExact time
HakemBench evaluates models across three dimensions—decision quality (macro F1), calibration (normalized Brier score), and selective automation (normalized area under the generalized risk-coverage curve)—and combines them into a composite score via geometric mean.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSHAP值通过博弈论中的沙普利值方法量化每个特征对单一预测结果的贡献度，LIME通过在预测点附近构建局部线性近似模型解释个案决策66% similarUnverifiedModelScope has launched the InternDecision series of models, covering three parameter scales of 0.8B, 2B, and 4B, targeting multimodal and structured decision-making tasks.64% similarVerified混合专家架构MoE通过稀疏激活机制，推理时只激活模型总参数的1/8至1/464% similarUnverifiedSHAP values and LIME are commonly used local interpretability methods capable of quantifying each feature's contribution to a single decision.64% similarUnverified选择性量化思路与LLM领域的GPTQ、AWQ等量化方法一脉相承63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975445API
curl https://kongchang.com/api/v1/knowledge/claims/975445MCP
get_claim(id=975445)