[CLS 01 · 学习信号、表示与迁移](/course/notebooks/fb4365d604974dd19df492cb22516e44.html)

# 活动 B：队列身份与参数更新分开看 · 实现代码
[独立阅读](/course/notebook-blocks/f1ec258c1fdf4a93a17b3e1df63f5a70.html)
主线阅读与优先实验

标签供给、文本目标接口、视图/队列/MAE图与表示迁移证据
展开完整实现代码
```
# 独立 CPU 机制；追踪特征身份，不进行图像模型训练。
from collections import deque
queue = deque(['a0','a1','b0','b1','c0','c1','d0','d1'], maxlen=8)
m = 0.9
key_theta = [0.0,2.0]
query_theta = [2.0,4.0]
print('更新前:', list(queue))
for batch in (['e0','e1'], ['f0','f1']):
    key_theta = [m*k+(1-m)*q for k,q in zip(key_theta,query_theta)]
    queue.extend(batch)
    print('加入',batch,'→',list(queue),'key参数',[round(x,4) for x in key_theta])
assert list(queue)==['c0','c1','d0','d1','e0','e1','f0','f1']
assert all(abs(a-b)<1e-12 for a,b in zip(key_theta,[0.38,2.38]))
```