-
meigel 11:46 AM
pls check
http://proceedings.mlr.press/v84/izmailov18a.html
PMLRPMLR
Scalable Gaussian Processes with Billions of Inducing Inputs via Tensor Train Decomposition
We propose a …
-
Found a bunch of repeated words, many of which appear to be erroneous (e.g. "An event is an an element"). I haven't checked all of them though.
Repeated token 'an' at:
discuss in Section 2.1.1.4…
-
Goodfellow I, et. al. [Generative adversarial nets](https://proceedings.neurips.cc/paper/2014/file/5ca3e9b122f61f8f06494c97b1afccf3-Paper.pdf).
-
It looks like splitting is not performed in the `_BaseCouplingBlock` properly when its `channels` attribute is equal to 1, leading to errors when input is passed to the subnet later on. The offending …
-
As you mentioned, this implementation meets with exploding gradients. Is there any reason for this? Is it due to the reversible connection or implementation problem?
-
When I run the code I get this error :
Traceback (most recent call last):
File "", line 1, in
File "/home/.pycharm_helpers/pydev/_pydev_bundle/pydev_umd.py", line 197, in runfile
pydev_…
-
I have a question about the calmping in `RNVPCouplingBlock` and `GLOWCouplingBlock`.
Code from `RNVPCouplingBlock`:
```python
def e(self, s):
# return torch.exp(torch.clamp(s, -self.cl…
-
Excellent library! I'm wondering if there are any tips/guidance from the community on how to choose the network architecture and depth. Currently I'm trying to do unconditional generation of ultrasoun…
-
It is a great job!
The inverse-operation is quite novel for me,I notice that in the paper, the formula is
> hl+1 = h | **exp**( (hl 2)) + φ(hl 2);
hl+12 = hl 2 exp(ρ(hl 1+1)) + η(hl 1+1)
Howev…
-
To make an actual contribution I have to design a new component of a Normalising Flow architecture that makes it a 'causal flow'. To do that I have identify an issue with the simple extension of the C…