VAE | Tensorflow implementation of Auto-Encoding Variational Bayes | Machine Learning library

I am trying to develop a VAE using this dataset, I have created and encoder and decoder by myself using keras tutorial, I only used Dense layers but now I wanted to add Conv1D layers too, however, after adding 1 conv layer to the encoder I get: Input 0 of layer "conv1d_3" is incompatible with the layer: expected min_ndim=3, found ndim=2. Full shape received: (None, 3)

I have found many questions like this but haven't found the exact answer, I want to add more Conv1D layers to the encoder and decoder, what do I need to change in both of them to add Con1D layers?

The code:

I am developing a VAE using this dataset. I have used keras tutorial code and developed my own VAE. However, when I run fit() function I get: Invalid reduction dimension 1 for input with 1 dimensions. for '{{node Sum}} = Sum[T=DT_FLOAT, Tidx=DT_INT32, keep_dims=false](Mean, Sum/reduction_indices)' with input shapes: [?], [2] and with computed input tensors: input[1] = <1 2>. What do I have to change?

The code:

I am developing a VAE using this: dataset

I have used keras tutorial code and I have developed my own encoder and decoder, the problem is that when I run vae.fit() I get 'int' object is not subscriptable. What am I doing wrong?

I am training a VQVAE with this dataset (64x64x3). I have downloaded it locally and loaded it with keras in Jupyter notebook. The problem is that when I ran fit() to train the model I get this error: ValueError: Layer "vq_vae" expects 1 input(s), but it received 2 input tensors. Inputs received: [, ] . I have taken most of the code from here and adapted it myself. But for some reason I can't make it work for other datasets. You can ignore most of the code here and check it in the page, help is much appreciated.

The code I have so far:

I'm working on a procfs kernel extension for macOS and trying to implement a feature that emulates Linux’s /proc/cpuinfo similar to what FreeBSD does with its linprocfs. Since I'm trying to learn, and since not every bit of FreeBSD code can simply be copied over to XNU and be expected to work right out of the jar, I'm writing this feature from scratch, with FreeBSD and NetBSD's linux-based procfs features as a reference. Anyways...

Under Linux, $cat /proc/cpuinfo showes me something like this:

I'm using the CIFAR-10 pre-trained VAE from lightning-bolts. It should be able to regenerate images with the quality shown on this picture taken from the docs (LHS are the real images, RHS are the generated)

However, when I write a simple script that loads the model, the weights, and tests it over the training set, I get a much worse reconstruction (top row are real images, bottom row are the generated ones):

Here is a link to a self-contained colab notebook that reproduces the steps I've followed to produce the pictures.

Am I doing something wrong on my inference process? Could it be that the weights are not as "good" as the docs claim?

Thanks!

I want to have two 50% divs, but the content of the first div got a min-size.

The variational autoencoder loss function is this: Loss = Loss_reconstruction + Beta * Loss_kld. I am trying to efficiently implement Kullback-Liebler Divergence Cyclic Annealing--that is changing the weight of beta dynamically during training. I subclass the tf.keras.callbacks.Callback class as a start, but I don't know how I can update a tf.keras.Model variable from a custom keras callback. Furthermore, I would like to track how the betas change at the end of each training step (on_train_batch_end), and right now I have a list in the callback class, but I know python lists don't play well with TensorFlow. When I fit the model, I get a warning that my on_train_batch_end function is slower than the processing of the batch itself. I think I should use a tf.TensorArray instead of python lists, but then the tf.TensorArray method write cannot use a tf.Variable for the index (i.e., as the number of steps changes, the index in the tf.TensorArray to which a new beta for that step should be written changes)... is there a better way to store value changes? It looks like this github shows a solution that doesn't involve a custom tf.keras.Model and that uses a different kind of KL annealing. Below is a callback function and dummy VAE.