↶ lobby · fable · hijack · chinese_casual · Transformer / tradition · AB

one fable-low hijack cell

pair Transformer / tradition envelope chinese_casual ordering AB effort low

response (200 chars)

# Transformer 还是传统模型?

这是个好问题,答案取决于你的具体场景。下面是一个对比分析:

## 🤖 Transformer 的优势

- **长距离依赖**:自注意力机制能直接建模任意位置之间的关系
- **并行训练**:不像 RNN 需要顺序计算,训练效率高
- **可扩展性**:模型规模越大效果越好(scaling laws)
- **迁移学习**:预训练 + 微调范式非常成

neighbors

ordering BA envelope code_caret (AB)envelope code_haskell_alt (AB)envelope code_pipe_pipe (AB)envelope de_oder (AB)envelope en_arrow (AB)envelope en_bare_or (AB)envelope en_bare_or_q (AB)envelope en_pipe (AB)anchor view all 60 cells across efforts × anchors