This repository contains a LeNet implementation with ReLUs (instead of Tanh) and two types of pooling layers. The code is a modification of the one published here: https://towardsdatascience.com/implementing-yann-lecuns-lenet-5-in-pytorch-5e05a0911320
All the requirements to get the project running have been mentioned in the Lenet/requirements.txt
file. The Python version required is 3.8.15
.
- Open the Jupyter notebook
Lenet/LeNet_jupiter.ipynb
. - Make sure you have installed all the requirements from
Lenet/requirements.txt
and you have Python version3.8.15
or newer. - Run the cells of the Jupyter notebook one by one.
Range of channels reserved: 100000 - 100019
Channels in use:
- LenetData/Parameters/Default-parameters/P0-classifier.0.bias.txt
- LenetData/Parameters/Default-parameters/P0-classifier.0.weight.txt
- LenetData/Parameters/Default-parameters/P0-classifier.2.bias.txt
- LenetData/Parameters/Default-parameters/P0-classifier.2.weight.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.0.bias.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.0.weight.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.3.bias.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.3.weight.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.6.bias.txt
- LenetData/Parameters/Default-parameters/P0-feature_extractor.6.weight.txt
Range of channels reserved: 90000 - 90049
- 90000 → execution 1 ./LenetData/Inputs/Execution_0/P0-feature_extractor_0.input.txt
- 90001 → execution 2 ./LenetData/Inputs/Execution_1/P0-feature_extractor_0.input.txt
- 90002 → execution 3 ./LenetData/Inputs/Execution_2/P0-feature_extractor_0.input.txt
To insert neural network parameters, a script can be found in Scripts directory insert-parameters-and-inputs.sh
.
Mysql client is required to be installed in your machine and setting up the option to share files with
the server by using SET GLOBAL local_infile = 1;