Hands-On Deep Learning with Go

Test and validation

We won't cover the code for testing in extensive detail as we've already covered how to get our images out of our output, but note that y is now also 784 columns wide:

arrayOutput := m.predVal.Data().([]float64)yOutput := tensor.New(tensor.WithShape(bs, 784), tensor.WithBacking(arrayOutput))for j := 0; j < yOutput.Shape()[0]; j++ {    rowT, _ := yOutput.Slice(sli{j, j + 1})    row := rowT.Data().([]float64)    img := visualizeRow(row)    f, _ := os.OpenFile(fmt.Sprintf("images/%d - %d output.jpg", b, j), os.O_CREATE|os.O_WRONLY|os.O_TRUNC, 0644)    jpeg.Encode(f, img, &jpeg.Options{jpeg.DefaultQuality})    f.Close()}

Now, here's the fun part; getting results out of our autoencoder:

You'll notice that the results are noticeably less well ...