Unverified50% confidenceFactExact time
The learnable parameters of hyperdimensional classifiers are low-bit angle values, which the paper refers to as phase memory.
1
Sources
50%
Confidence
Long-term
Relevance
10/9/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedThe inference stage of large models is memory-bound (limited by memory bandwidth); generating each token requires reading the entire model weights from VRAM, resulting in extremely low computational intensity.76% similarUnverified上下文图谱为智能体提供了介于临时上下文和模型权重微调之间的中间层记忆机制，可在不改变底层模型的前提下持续积累经验71% similarUnverifiedParametric memory is static, read-only knowledge solidified into model weights during the training phase and does not update through conversation after training is complete.69% similarUnverifiedThe design of the memory management layer is partly inspired by the memory consolidation theory in cognitive science, corresponding to the transformation from episodic memory to semantic memory, with the Ebbinghaus forgetting curve providing the basis for decay function parameter design.69% similarUnverifiedThe degree of model memorization is highly correlated with the number of data repetitions; when the number of parameters greatly exceeds the information entropy of the training data, the model tends to memorize rather than generalize.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/990622API
curl https://kongchang.com/api/v1/knowledge/claims/990622MCP
get_claim(id=990622)