benchmark: I used it for training my proposed model about fashion-mnist

[AFAgarap]

I did another training with the following hyper-parameters:

BATCH_SIZE = 256
CELL_SIZE = 256
DROPOUT_P_KEEP = 0.85
EPOCHS = 10
LEARNING_RATE = 1e-3
SVM_C = 1

Trained using tf.train.AdamOptimizer(), and used tf.nn.dynamic_rnn().

Here's the result:

Epoch : 0 completed out of 10, loss : 176.0813446044922, accuracy : 0.875
Epoch : 1 completed out of 10, loss : 198.86962890625, accuracy : 0.84375
Epoch : 2 completed out of 10, loss : 160.88693237304688, accuracy : 0.87109375
Epoch : 3 completed out of 10, loss : 206.49061584472656, accuracy : 0.84765625
Epoch : 4 completed out of 10, loss : 169.35137939453125, accuracy : 0.84375
Epoch : 5 completed out of 10, loss : 158.23252868652344, accuracy : 0.8984375
Epoch : 6 completed out of 10, loss : 210.0270233154297, accuracy : 0.85546875
Epoch : 7 completed out of 10, loss : 173.9663848876953, accuracy : 0.875
Epoch : 8 completed out of 10, loss : 190.8488006591797, accuracy : 0.8671875
Epoch : 9 completed out of 10, loss : 161.12210083007812, accuracy : 0.89453125
Accuracy : 0.8551000356674194

The implementation I used for this test may be found here.