A newer version of the Gradio SDK is available:
6.2.0
title: TSAI S12
emoji: ⚡
colorFrom: green
colorTo: indigo
sdk: gradio
sdk_version: 3.40.1
app_file: app.py
pinned: false
license: mit
The School of AI - ERA(Extensive & Reimagined AI Program) - Assignment 12
This folder consists of Assignment-12 from ERA course offered by - TSAI(The school of AI). Follow https://theschoolof.ai/ for more updates on TSAI
For more details on the assignment, refer to github link: https://github.com/ToletiSri/TSAI_ERA_Assignments/tree/main/S12
As part of the assignment, we have trained a custom resnet model on CIFAR-10 dataset, using pytorch lightning. We have then saved the trained model to file - saved_model.pth. The saved model is uploaded and used in the current Space.
As part of the app, we provide provide the following features to the user:
- ask the user whether he/she wants to see GradCAM images and how many, and from which layer, allow opacity change as well
- allow users to upload new images, as well as provide 10 example images
- ask how many top classes are to be shown (make sure the user cannot enter more than 10)
The custom resnet model used, has the following model architecture
# | Name | Type | Params
0 | loss_criteria | CrossEntropyLoss | 0
1 | accuracy | MulticlassAccuracy | 0
2 | convblockPreparation | Sequential | 1.9 K
3 | convblockL1X1 | Sequential | 74.0 K
4 | convblockL1R1 | Sequential | 295 K
5 | convblockL2X1 | Sequential | 295 K
6 | convblockL3X1 | Sequential | 1.2 M
7 | convblockL3R1 | Sequential | 4.7 M
8 | FinalBlock | Sequential | 0
9 | FC | Sequential | 5.1 K
10 | dropout | Dropout | 0
6.6 M Trainable params 0 Non-trainable params 6.6 M Total params 26.293 Total estimated model params size (MB)
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference