[LAB 01 · 固定表示：标签预算与读出](/course/notebooks/4dd8605ed9904820919ea9e13d9e7d7c.html)

# 本地保存输出（非 Deepnote 云端运行）
[独立阅读](/course/notebook-blocks/9d7eedd691a24080aca7286577ebf886.html)
按问题继续

固定CNN与缓存特征、标签预算、kNN/linear和独立评价

**本地保存输出（非 Deepnote 云端运行）**

![本地保存图](/assets/display/7c292d90e96c12dfbd1803b6.webp)

```
Split sizes: {'train': 256, 'validation': 128, 'heldout': 256}
SOURCE: ba8b738cc67d4ec3cd83d75a5dbe8a70ec548377335ba0abaaa7c23cd14eeba3
Checks passed: file identity; sample separation; preprocessing.

```