DeepLearning.AI· DeepLearningAI·· 11 天前AI 评分42
DeepSeek优化智能体记忆缓存
Agents read more than they write, so DeepSeek shrank what they remember. Read the technical breakdown in this week's issue of The Batch. ...
AI 导读
智能体读取多于写入，因此DeepSeek缩减了其记忆内容。请阅读本周《The Batch》中的技术解析。每 token 890 字节缓存，比 DeepSeek-V1 小 437 倍；输入从 4K 到 1M，每输出 token 计算量增加 25%；Flash 胜出
来源：DeepLearning.AI · x.com