W1 Q6 accuracy not converging

UNQ_C6 not giving same accuracy as expected output, I am getting this again and again, all unitest passed before and after this

Step 1: Total number of trainable weights: 2327042 Step 1: Ran 1 train steps in 1.46 secs Step 1: train CrossEntropyLoss | 5.54816008 Step 1: eval CrossEntropyLoss | 5.58890200 Step 1: eval Accuracy | 4.00000000

I​ tried changing mean_layer = tl.Mean(axis=1, keepdims = True) fromQ5 but if I use keepdims = False or just mean_layer = tl.Mean(axis=1) I get a different error.

P​lease help!!

Finally solved it !!

For me it was the for loop in the data generator, I was setting wrong the start of the negative examples