added README.md
Browse files
README.md
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
datasets:
|
3 |
+
- cifar10
|
4 |
+
metrics:
|
5 |
+
- accuracy
|
6 |
+
library_name: pytorch
|
7 |
+
pipeline_tag: image-classification
|
8 |
+
---
|
9 |
+
|
10 |
+
# custom-cnn-cifar2
|
11 |
+
|
12 |
+
Custom convolutional neural network (CNN) trained on CIFAR-2 (a subset of CIFAR-10 for classifying 'airplane' vs. 'bird').
|
13 |
+
|
14 |
+
This model pertains to Exercise 1 of Chapter 8 of the book "Deep Learning with PyTorch" by Eli Stevens, Luca Antiga, and Thomas Viehmann.
|
15 |
+
|
16 |
+
**Note:** In the exercise, we tried out `(5, 5)` and `(1, 3)` convolution kernel sizes. However, these didn't outperform the baseline network with `(3, 3)` kernel size. Hence, this checkpoint sticks to the `(3, 3)` kernel size.
|
17 |
+
|
18 |
+
Code: https://github.com/sambitmukherjee/dlwpt-exercises/blob/main/chapter_8/exercise_1.ipynb
|
19 |
+
|
20 |
+
Experiment tracking: https://wandb.ai/sadhaklal/custom-cnn-cifar2
|
21 |
+
|
22 |
+
## Metric
|
23 |
+
|
24 |
+
Accuracy on `cifar2_val`: 0.8995
|