malar
/
speech-scoring
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

resuming segment training

master
Malar Kannan 2017-12-10 21:58:55 +05:30
parent 03edd935ea
commit 52bbb69c65
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
segment_model.py
View File

@ -54,8 +54,6 @@ def segment_model(input_dim):
return Model(inp, oup) return Model(inp, oup)
def simple_segment_model(input_dim): def simple_segment_model(input_dim):
# input_dim = (100,100)
# input_dim = (506,743)
inp = Input(shape=input_dim) inp = Input(shape=input_dim)
b_gr1 = Bidirectional(GRU(256, return_sequences=True),merge_mode='sum')(inp) b_gr1 = Bidirectional(GRU(256, return_sequences=True),merge_mode='sum')(inp)
# b_gr1 # b_gr1
@ -75,7 +73,7 @@ def load_model_arch(mod_file):
model_f.close() model_f.close()
return mod return mod
def train_segment(collection_name = 'test'): def train_segment(collection_name = 'test',resume_weights='',initial_epoch=0):
# collection_name = 'story_test' # collection_name = 'story_test'
batch_size = 64 batch_size = 64
# batch_size = 4 # batch_size = 4
@ -101,7 +99,7 @@ def train_segment(collection_name = 'test'):
embeddings_freq=0, embeddings_freq=0,
embeddings_layer_names=None, embeddings_layer_names=None,
embeddings_metadata=None) embeddings_metadata=None)
cp_file_fmt = model_dir+'/siamese_speech_model-{epoch:02d}-epoch-{val_loss:0.2f}\ cp_file_fmt = model_dir+'/speech_segment_model-{epoch:02d}-epoch-{val_loss:0.2f}\
-acc.h5' -acc.h5'
cp_cb = ModelCheckpoint( cp_cb = ModelCheckpoint(
@ -115,14 +113,16 @@ def train_segment(collection_name = 'test'):
# train # train
rms = RMSprop() rms = RMSprop()
model.compile(loss=categorical_crossentropy, optimizer=rms, metrics=[accuracy]) model.compile(loss=categorical_crossentropy, optimizer=rms, metrics=[accuracy])
write_model_arch(model,model_dir+'/siamese_speech_model_arch.yaml') write_model_arch(model,model_dir+'/speech_segment_model_arch.yaml')
epoch_n_steps = step_count(n_records,batch_size) epoch_n_steps = step_count(n_records,batch_size)
if resume_weights != '':
model.load_weights(resume_weights)
model.fit_generator(tr_gen model.fit_generator(tr_gen
, epochs=1000 , epochs=10000
, steps_per_epoch=epoch_n_steps , steps_per_epoch=epoch_n_steps
, validation_data=(te_x, te_y) , validation_data=(te_x, te_y)
, max_queue_size=32 , max_queue_size=32
, callbacks=[tb_cb, cp_cb]) , callbacks=[tb_cb, cp_cb],initial_epoch=initial_epoch)
model.save(model_dir+'/speech_segment_model-final.h5') model.save(model_dir+'/speech_segment_model-final.h5')
# y_pred = model.predict([te_pairs[:, 0], te_pairs[:, 1]]) # y_pred = model.predict([te_pairs[:, 0], te_pairs[:, 1]])
@ -133,4 +133,4 @@ def train_segment(collection_name = 'test'):
if __name__ == '__main__': if __name__ == '__main__':
# pass # pass
train_segment('story_phrases') train_segment('story_phrases','./models/segment/story_phrases.1000/speech_segment_model-final.h5',1001)