Expired50% confidenceFactTime unknown
LLM 'memory' is merely a temporary stack of context, which differs fundamentally from human memory that continuously updates neural connections through synaptic plasticity
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
Huawei Hermes Agent Manual Explained: Five-Layer Memory Architecture and Multi-Agent Collaboration in Practice
bilibili码农人帥气质佳
Related Claims
UnverifiedLSTM等循环神经网络通过隐状态向量和细胞状态在时间步之间显式传递信息，天然具备序列记忆能力，但无法在序列维度上并行73% similarUnverifiedLSTM 通过引入记忆单元与门控机制（输入门、遗忘门、输出门）来捕捉时间序列中的长期依赖关系73% similarPartially VerifiedEach LLM inference is stateless and constrained by the limited context window; the model has no persistent memory of its own.70% similarUnverifiedThe memory of LLMs depends on the context window, which is the maximum number of tokens the model can process in a single inference.70% similarUnverifiedLLM 在处理复杂任务时容易出现幻觉和逻辑漂移，这与其自回归逐 token 预测机制有关，早期错误会被后续内容放大70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/55328API
curl https://kongchang.com/api/v1/knowledge/claims/55328MCP
get_claim(id=55328)