Unverified50% confidenceFactExact time
美国BIS于2023年发布的芯片出口管制规则已将特定算力门槛以上的模型训练纳入管制考量
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
Related Claims
Unverified美国AI监管以模型训练算力阈值（以FLOP为单位）作为触发强制报备义务的量化指标67% similarUnverifiedThe training-time scaling law was proposed by Kaplan et al. in 2020, holding that model capability is primarily determined by parameter count, training data volume, and training compute.67% similarUnverifiedExecutive Order 14110以训练所用浮点运算量（FLOPs）作为触发审查义务的量化阈值，超过10²⁶次FLOPs训练量的模型即需强制报备64% similarUnverifiedGPTQ于2022年采用逐层二阶优化将已训练模型压缩至INT4，是训练后量化的代表61% similarUnverifiedFrugalGPT（Chen et al., 2023）和LLM-Blender等框架试图通过训练专用路由分类器来平衡准确性与开销60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/489145API
curl https://kongchang.com/api/v1/knowledge/claims/489145MCP
get_claim(id=489145)