Hello,
I am using ......
optimizer = RMSprop(lr=1e-4, decay=1e-6)
N_FOLDS = 5
EPOCHS = 20
INIT_LR = 1e-4
Train_Batch_Size = 32
Valid_Batch_Size = 32
decay_rate = 0.95
decay_step = 1
skf = StratifiedKFold(n_splits = N_FOLDS , random_state = 7, shuffle = True) tensorboard_callback = TensorBoard(log_dir=log_dir, histogram_freq=1)
callbacks = [ModelCheckpoint(filepath='best_lstm_model.h5', monitor='val_loss',mode='min',verbose=1,save_best_only=True,save_weights_only=True), LearningRateScheduler(lambda epoch : INIT_LR * pow(decay_rate, floor(epoch / decay_step))), tensorboard_callback]
I don't know what's wrong here? Can anyone help me? Thank you.
One figure shows the *loss* (cross-entropy?), the other the chosen metric (accuracy in your case). Note that accuracy is hardly ever a good metric to consider.
Only the training curve, it seems. Check if the training data is randomized/shuffled well enough.
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training