Alexander Novikov

It's an adder which adds entire episodes into replay. For this to make sense your agent has to be recurrent. An alternative is e.g. adding transitions (SARSA tuples) into replay:...

Not exactly. When you work with SARSA tuples it indeed works like this. When working with sequences, they are stored as a tuple `(observations, actions, rewards, discounts)`, where each tensor...

Please change this line: https://github.com/oseledets/ttpy/blob/master/tt/core/vector.py#L189 from `answ_cores[-1] = _np.dot(answ_cores[-1], running_fact)` to ``` if len(answ_cores) > 0: answ_cores[-1] = _np.dot(answ_cores[-1], running_fact) else: # Result is a number. answ_cores = [reshape(running_fact, (1,...

You can also try TensorFlow version of Tensor Train toolbox. It is not yet tested on Windows but should work without tricks. github.com/Bihaqo/t3f

It seems that the problem is that Nose runs all the test files in the same process, polluting Tensorflow state (in some of the test files eager execution is turned...

Sorry for an embarrassingly long reply. As this library focuses on a particular tensor decomposition (Tensor Train decomposition) and as I don't know a good way of using TT-decomposition for...

Hi A few things 1) I had reasons to call `max_tt_rank` and `tt_rank` differently, but now that you questioned it, I realised that those reasons were never convincing enough and...

So the idea is that if your input dims are `[a1, a2, a3]` and your output dims are `[b1, b2, b3]`, then your TT-ranks should be smaller than `np.minimum([1, a1*b1,...

Can you please provide a small reproducible example? Because I'm not exactly sure I got you correctly. By "contacting all indexes but a few" you mean something like Y[i_3, i_5]...

1. You still can use a batch of objects. pairwise_flat_inner(x, w) is a matrix of cross products of size x.batch_size x w.batch_size 2. Good point, I see. And where do...