Unverified50% confidenceFactExact time
工作记忆模型由Baddeley与Hitch于1974年提出，容量约为7±2个信息组块
1
Sources
50%
Confidence
Long-term
Relevance
7/21/2026
First Seen
Sources
Related Claims
UnverifiedThe memory requirements of the Jacobian Lens grow linearly with model scale, roughly 1/8 of the model size.56% similarUnverifiedCowan's embedded-processes model holds that the effective capacity of working memory is closer to 4 chunks.56% similarUnverifiedThe hierarchical memory system contains two components: Memory Bank and Memory Profile56% similarUnverifiedUsing float16 precision as a baseline, each parameter occupies approximately 2 bytes, meaning a 7B model requires approximately 14GB of VRAM/memory, and a 70B model requires approximately 140GB.55% similarVerifiedThe speed bottleneck in large model inference is memory bandwidth rather than insufficient computing power, as generating each token requires loading hundreds of GB of model weights from memory.55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/579860API
curl https://kongchang.com/api/v1/knowledge/claims/579860MCP
get_claim(id=579860)