David Chan

I found that the error was resolved for me by adding a check to tflearn/helpers/trainer.py at line 134: ```{python} if to_restore: self.restorer = tf.train.Saver( var_list=to_restore, max_to_keep=max_checkpoints, keep_checkpoint_every_n_hours=keep_checkpoint_every_n_hours) ``` However this...

Thanks for your interest on developing for TSNE-CUDA! I don't think there's a huge amount of complexity, at least mathematically in extending it to 3D visualization. There is, however, a...

I'm actually just now starting to get back into active development for this code - and it's on my ToDo list (as one of the most requested features).

Approximately what sized datasets are you thinking of for 3D visualization? I've been going through our code and the potential implementation, and the scaling for 3D isn't very good when...

The average gradient norm is basically the norm of the gradient of the KL-divergence with respect to the particle positions. Thus, it can be a proxy for how stable the...

Currently, no - but I'll consider working it into the next version, and we're always welcome to PRs if anyone wants to contribute! Here would be a good place in...

This is not currently a planned feature, however it certainly seems possible. You might consider looking at the file: https://github.com/CannyLab/tsne-cuda/blob/master/src/ext/pymodule_ext.cu, which is exposed in the library as a binding method...

Just an FYI: the real reason that this argument is ignored is because I'm not entirely certain that we can enforce strict random seeding. There may be external sources of...

The reason is probably the kernel-level optimizations here: https://github.com/CannyLab/tsne-cuda/blob/master/src/include/options.h, or an upstream sub-optimization in FAISS. Every GPU architecture is different, so it needs to be tuned to get the best...

In terms of the PTXAS warnings - they're upstream issues that come from thrust, a low-level computing library that we use for simple GPU operations.