GitHub - MihailoIsakov/fraction_bitwidth: Experimenting what works when using fixed point values for training data

MihailoIsakov / fraction_bitwidth Public

Experimenting what works when using fixed point values for training data

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 6 Commits
.gitignore		.gitignore
Quantized training bottleneck.ipynb		Quantized training bottleneck.ipynb
Test hardware.ipynb		Test hardware.ipynb
Training on quantized data.ipynb		Training on quantized data.ipynb
Training with fixed point values.ipynb		Training with fixed point values.ipynb
activation_values.py		activation_values.py
activations.py		activations.py
avg_update.pickle		avg_update.pickle
backprop.py		backprop.py
breast_input_30neuron_20bit_16frac.mem		breast_input_30neuron_20bit_16frac.mem
breast_output_30neuron_20bit_16frac.mem		breast_output_30neuron_20bit_16frac.mem
delta_test.py		delta_test.py
deltas.pickle		deltas.pickle
digits_accuracies.pickle		digits_accuracies.pickle
iris_test.pickle		iris_test.pickle
large_acc.pickle		large_acc.pickle
lr_fraction_bar_graph.py		lr_fraction_bar_graph.py
lr_fraction_data.py		lr_fraction_data.py
mnist_accuracies.pickle		mnist_accuracies.pickle
mnist_int.py		mnist_int.py
train_integers.py		train_integers.py
weights_30x30_20bit_16frac.mem		weights_30x30_20bit_16frac.mem