我们当时得出结论：对于端侧部署对话，日常用 70B 足矣，花大几万买台机器只为了跟 AI 聊天，实在是有钱烧的慌。当时的模型能力确实也就不太行，后来才有新的多模态模型和 agent 能力出来。
���f�B�A�ꗗ | ����SNS | �L���ē� | ���₢���킹 | �v���C�o�V�[�|���V�[ | RSS | �^�c���� | �̗p���� | ������
。关于这个话题，搜狗输入法提供了深入分析
其实黛西从头到尾也不懂什么是爱。
the core DFA algorithm is also very simple. now i know, coming from someone who’s been doing years of PhD research on this topic, but it really is simple. you got 4 things: