Comments (2)
I came across your issue after the same error, I presume: ValueError: Expected parameter probs (Tensor of shape (1, 128, 1, 28, 28)) of distribution LogitRelaxedBernoulli(probs: torch.Size([1, 128, 1, 28, 28])) to satisfy the constraint Interval(lower_bound=0.0, upper_bound=1.0), but found invalid values
i'm not sure but my hunch is this is due to a change in pytorch. i'm using pytorch v.1.13 and had that error. this library assumes an earlier version of both geoopt and pytorch, i think.
if you resolve the error, please share! hope my reply helps.
from pvae.
I think the reason for this is that the constructor for torch.distributions.RelaxedBernoulli
changed at some point. The decoder here is outputting logits, and to initialize a RelaxedBernoulli with logits, one must use the logits=
keyword argument:
So, I modified the following in pvae/models/vae.py
and training works:
class VAE(nn.Module):
...
def forward(self, x, K=1):
...
- px_z = self.px_z(*self.dec(zs))
+ temperature, decoder_output = self.dec(zs)
+ px_z = self.px_z(temperature, logits=decoder_output)
note, perhaps obviously, this assumes your px_z
(likelihood distribution) is parameterized by temperature
and logits
.
from pvae.
Related Issues (7)
- Question about the implementation of wrapped normal HOT 2
- normdist2plane potential implementation error?
- subprocess.CalledProcessError: Command '['git', 'rev-parse', '--verify', 'HEAD']' returned non-zero exit status 128 HOT 2
- Negative KL loss HOT 3
- replicate the Graph embeddings experiment
- Error when trying to train mnist poincare experiment
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from pvae.