File size: 2,314 Bytes
299bc83
 
 
 
91e97d6
 
 
 
 
 
299bc83
 
91e97d6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6b7c50
 
91e97d6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6b7c50
 
 
 
 
 
91e97d6
 
 
 
 
 
 
 
 
 
d6b7c50
91e97d6
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
---
tags:
- model_hub_mixin
- pytorch_model_hub_mixin
datasets:
- zalando-datasets/fashion_mnist
metrics:
- accuracy
library_name: pytorch
pipeline_tag: image-classification
---

# mlp-fashion-mnist

A multi-layer perceptron (MLP) trained on the Fashion-MNIST dataset.

It is a PyTorch adaptation of the TensorFlow model in Chapter 10 of Aurelien Geron's book 'Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow'.

Code: https://github.com/sambitmukherjee/handson-ml3-pytorch/blob/main/chapter10/mlp_fashion_mnist.ipynb

Experiment tracking: https://wandb.ai/sadhaklal/mlp-fashion-mnist

## Usage

```
!pip install -q datasets

from datasets import load_dataset

fashion_mnist = load_dataset("zalando-datasets/fashion_mnist")

features = fashion_mnist['train'].features
id2label = {id: label for id, label in enumerate(features['label'].names)}

import torch
import torchvision.transforms.v2 as v2

tfms = v2.Compose([
    v2.ToImage(),
    v2.ToDtype(torch.float32, scale=True)
])

device = torch.device("cpu")

import torch.nn as nn
from huggingface_hub import PyTorchModelHubMixin

class MLP(nn.Module, PyTorchModelHubMixin):
    def __init__(self):
        super().__init__()
        self.fc1 = nn.Linear(28 * 28, 300)
        self.fc2 = nn.Linear(300, 100)
        self.fc3 = nn.Linear(100, 10)

    def forward(self, x):
        x = x.view(-1, 28 * 28)
        act = torch.relu(self.fc1(x))
        act = torch.relu(self.fc2(act))
        return self.fc3(act)

model = MLP.from_pretrained("sadhaklal/mlp-fashion-mnist")
model.to(device)

example = fashion_mnist['test'][0]

import matplotlib.pyplot as plt

plt.imshow(example['image'], cmap='gray')
print(f"Ground truth: {id2label[example['label']]}")

img = tfms(example['image'])
x_batch = img.unsqueeze(0)

model.eval()
x_batch = x_batch.to(device)
with torch.no_grad():
    logits = model(x_batch)
proba = torch.softmax(logits, dim=-1)

confidence, pred = proba.max(dim=-1)
print(f"Predicted class: {id2label[pred[0].item()]}")
print(f"Predicted confidence: {round(confidence[0].item(), 4)}")
```

## Metric

Accuracy on the test set: 0.8829

---

This model has been pushed to the Hub using the [PyTorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration.