Why is my model fitting only on the batch data of 36 observations, Batch size is 100 and the Total Observations are 3579. which means only a single batch gets Frontpropogated. Why?

Hi, I believe that the log indicates 36 because 36 is the number of data contained in each of the 100 lots (3579/100 = 36). At each epoch, however, a new batch is considered and therefore a new set of data (always 36). The same thing happens in the video "Learning" of the Section "Deeper example" at 02:38. From your log I see that for you too the accuracy is about 80% and not 90% as presented in the video. Did you find any explanation for this?