Performing Iterative Pruning - 3.0 English

Vitis AI Optimizer User Guide (UG1333)

Document ID
UG1333
Release Date
2023-01-12
Version
3.0 English

Load the checkpoint saved from the previous fine-tuning stage. Increase the pruning ratio value to achieve higher levels of sparsity. With each pruning step, fine-tune this sparse model. Repeat this pruning and fine-tuning loop until the sparsity reaches the desired value, or until it is observed that the evaluation performance degrades significantly.

model.load_weights("model_sparse_0.2")

input_shape = [28, 28, 1]
input_spec = tf.TensorSpec((1, *input_shape), tf.float32)
runner = IterativePruningRunner(model, input_spec)
sparse_model = runner.prune(ratio=0.5)