# CISC6000 Deep Learning HW2 — Data

来源：https://www.kaggle.com/competitions/cisc6000-deep-learning-hw2/data

采集时间：2026-10-19T07:20:19Z

# Task One
1. cifar.train.npy 48000 pic in numpy array. 3 channel , 32 pixels by 32 pixels

2. cifar.trainlabel 48000 related labels

3. cifar.test.npy 12000 testing pics

4. sample_submission.csv: this is an example that YOU should generate from the model.

# Task Two
1. train_clean_data.zip 30000 pics cleaned image data for training

2. train_noise_data.zip 30000 pics noise image data for training

3. test_clean_data.zip 20000 pics cleaned image data for testing

4. test_noise_data.zip 20000 pics noise image data for testing

5. labels.csv: 20K labels for images in (III) and (IV)

6. M.weights.hdf5 Q2 question pre-train weight.