eden-emu/opus
15
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Convert PLC weights to blob format

Browse source
This commit is contained in:
Jean-Marc Valin 2023-05-22 03:32:53 -04:00
parent c7b6935bf2
commit 98da335009
4 changed files with 64 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

59
dnn/training_tf2/dump_plc.py
View file

@ -27,6 +27,7 @@
'''
import lpcnet_plc
import io
import sys
import numpy as np
from tensorflow.keras.optimizers import Adam
@ -41,11 +42,17 @@ max_rnn_neurons = 1
max_conv_inputs = 1
def printVector(f, vector, name, dtype='float', dotp=False):
global array_list
if dotp:
vector = vector.reshape((vector.shape[0]//4, 4, vector.shape[1]//8, 8))
vector = vector.transpose((2, 0, 3, 1))
v = np.reshape(vector, (-1));
#print('static const float ', name, '[', len(v), '] = \n', file=f)
if name not in array_list:
array_list.append(name)
f.write('#ifndef USE_WEIGHTS_FILE\n')
f.write('#define WEIGHTS_{}_DEFINED\n'.format(name))
f.write('#define WEIGHTS_{}_TYPE WEIGHT_TYPE_{}\n'.format(name, dtype))
f.write('static const {} {}[{}] = {{\n '.format(dtype, name, len(v)))
for i in range(0, len(v)):
f.write('{}'.format(v[i]))
@ -58,7 +65,8 @@ def printVector(f, vector, name, dtype='float', dotp=False):
else:
f.write(" ")
#print(v, file=f)
f.write('\n};\n\n')
f.write('\n};\n')
f.write('#endif\n\n')
return;
def printSparseVector(f, A, name, have_diag=True):
@ -122,11 +130,11 @@ def dump_sparse_gru(self, f, hf):
reset_after = 1
neurons = weights[0].shape[1]//3
max_rnn_neurons = max(max_rnn_neurons, neurons)
f.write('const SparseGRULayer {} = {{\n {}_bias,\n {}_subias,\n {}_recurrent_weights_diag,\n {}_recurrent_weights,\n {}_recurrent_weights_idx,\n {}, ACTIVATION_{}, {}\n}};\n\n'
.format(name, name, name, name, name, name, weights[0].shape[1]//3, activation, reset_after))
hf.write('#define {}_OUT_SIZE {}\n'.format(name.upper(), weights[0].shape[1]//3))
hf.write('#define {}_STATE_SIZE {}\n'.format(name.upper(), weights[0].shape[1]//3))
hf.write('extern const SparseGRULayer {};\n\n'.format(name));
model_struct.write(' SparseGRULayer {};\n'.format(name));
model_init.write(' if (sparse_gru_init(&model->{}, arrays, "{}_bias", "{}_subias", "{}_recurrent_weights_diag", "{}_recurrent_weights", "{}_recurrent_weights_idx", {}, ACTIVATION_{}, {})) return 1;\n'
.format(name, name, name, name, name, name, weights[0].shape[1]//3, activation, reset_after))
return True
def dump_gru_layer(self, f, hf):
@ -158,11 +166,11 @@ def dump_gru_layer(self, f, hf):
reset_after = 1
neurons = weights[0].shape[1]//3
max_rnn_neurons = max(max_rnn_neurons, neurons)
f.write('const GRULayer {} = {{\n {}_bias,\n {}_subias,\n {}_weights,\n {}_weights_idx,\n {}_recurrent_weights,\n {}, {}, ACTIVATION_{}, {}\n}};\n\n'
.format(name, name, name, name, name, name, weights[0].shape[0], weights[0].shape[1]//3, activation, reset_after))
hf.write('#define {}_OUT_SIZE {}\n'.format(name.upper(), weights[0].shape[1]//3))
hf.write('#define {}_STATE_SIZE {}\n'.format(name.upper(), weights[0].shape[1]//3))
hf.write('extern const GRULayer {};\n\n'.format(name));
model_struct.write(' GRULayer {};\n'.format(name));
model_init.write(' if (gru_init(&model->{}, arrays, "{}_bias", "{}_subias", "{}_weights", "{}_weights_idx", "{}_recurrent_weights", {}, {}, ACTIVATION_{}, {})) return 1;\n'
.format(name, name, name, name, name, name, weights[0].shape[0], weights[0].shape[1]//3, activation, reset_after))
return True
GRU.dump_layer = dump_gru_layer
@ -178,10 +186,10 @@ def dump_gru_layer_dummy(self, f, hf):
def dump_dense_layer_impl(name, weights, bias, activation, f, hf):
printVector(f, weights, name + '_weights')
printVector(f, bias, name + '_bias')
f.write('const DenseLayer {} = {{\n {}_bias,\n {}_weights,\n {}, {}, ACTIVATION_{}\n}};\n\n'
.format(name, name, name, weights.shape[0], weights.shape[1], activation))
hf.write('#define {}_OUT_SIZE {}\n'.format(name.upper(), weights.shape[1]))
hf.write('extern const DenseLayer {};\n\n'.format(name));
model_struct.write(' DenseLayer {};\n'.format(name));
model_init.write(' if (dense_init(&model->{}, arrays, "{}_bias", "{}_weights", {}, {}, ACTIVATION_{})) return 1;\n'
.format(name, name, name, weights.shape[0], weights.shape[1], activation))
def dump_dense_layer(self, f, hf):
name = self.name
@ -202,12 +210,12 @@ def dump_conv1d_layer(self, f, hf):
printVector(f, weights[-1], name + '_bias')
activation = self.activation.__name__.upper()
max_conv_inputs = max(max_conv_inputs, weights[0].shape[1]*weights[0].shape[0])
f.write('const Conv1DLayer {} = {{\n {}_bias,\n {}_weights,\n {}, {}, {}, ACTIVATION_{}\n}};\n\n'
.format(name, name, name, weights[0].shape[1], weights[0].shape[0], weights[0].shape[2], activation))
hf.write('#define {}_OUT_SIZE {}\n'.format(name.upper(), weights[0].shape[2]))
hf.write('#define {}_STATE_SIZE ({}*{})\n'.format(name.upper(), weights[0].shape[1], (weights[0].shape[0]-1)))
hf.write('#define {}_DELAY {}\n'.format(name.upper(), (weights[0].shape[0]-1)//2))
hf.write('extern const Conv1DLayer {};\n\n'.format(name));
model_struct.write(' Conv1DLayer {};\n'.format(name));
model_init.write(' if (conv1d_init(&model->{}, arrays, "{}_bias", "{}_weights", {}, {}, {}, ACTIVATION_{})) return 1;\n'
.format(name, name, name, weights[0].shape[1], weights[0].shape[0], weights[0].shape[2], activation))
return True
Conv1D.dump_layer = dump_conv1d_layer
@ -235,6 +243,12 @@ else:
f = open(cfile, 'w')
hf = open(hfile, 'w')
model_struct = io.StringIO()
model_init = io.StringIO()
model_struct.write('typedef struct {\n')
model_init.write('#ifndef DUMP_BINARY_WEIGHTS\n')
model_init.write('int init_plc_model(PLCModel *model, const WeightArray *arrays) {\n')
array_list = []
f.write('/*This file is automatically generated from a Keras model*/\n')
@ -250,6 +264,19 @@ for i, layer in enumerate(model.layers):
layer_list.append(layer.name)
#dump_sparse_gru(model.get_layer('gru_a'), f, hf)
f.write('#ifndef USE_WEIGHTS_FILE\n')
f.write('const WeightArray lpcnet_plc_arrays[] = {\n')
for name in array_list:
f.write('#ifdef WEIGHTS_{}_DEFINED\n'.format(name))
f.write(' {{"{}", WEIGHTS_{}_TYPE, sizeof({}), {}}},\n'.format(name, name, name, name))
f.write('#endif\n')
f.write(' {NULL, 0, 0, NULL}\n};\n')
f.write('#endif\n')
model_init.write(' return 0;\n}\n')
model_init.write('#endif\n')
f.write(model_init.getvalue())
hf.write('#define PLC_MAX_RNN_NEURONS {}\n\n'.format(max_rnn_neurons))
#hf.write('#define PLC_MAX_CONV_INPUTS {}\n\n'.format(max_conv_inputs))
@ -257,7 +284,11 @@ hf.write('#define PLC_MAX_RNN_NEURONS {}\n\n'.format(max_rnn_neurons))
hf.write('typedef struct {\n')
for i, name in enumerate(layer_list):
hf.write(' float {}_state[{}_STATE_SIZE];\n'.format(name, name.upper()))
hf.write('} PLCNetState;\n')
hf.write('} PLCNetState;\n\n')
model_struct.write('} PLCModel;\n\n')
hf.write(model_struct.getvalue())
hf.write('int init_plc_model(PLCModel *model, const WeightArray *arrays);\n\n')
hf.write('\n\n#endif\n')