Giter Club home page Giter Club logo

tyxe's Introduction

TyXe: Pyro-based BNNs for Pytorch users

Disclaimer: Please note that TyXe moved to https://github.com/TyXe-BDL/TyXe and this codebase here is probably not representing the library in its current shape.

tyxe's People

Contributors

hpplyt avatar karalets avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

Forkers

patrickens

tyxe's Issues

Predict interface

The predict method currently returns either the sampled predictions from the network (without data noise) or an aggregation of those. We should have a clear interface for predicting either with or without data noise, so it would probably be worth adding a method that returns fully sampled predictions.

We could potentially use the forward method for one of those cases, which would give us a more pytorch-like interface, since then users could just call bnn(x, num_samples) to make predictions.

variational BNN regression example does not work anymore with pyro 1.7.0

Thank you very much for the library!

FYI: the regression example with VariationalBNN does not seem to work anymore.

pyro.clear_param_store()
optim = pyro.optim.Adam({"lr": 1e-3})
elbos = []
def callback(bnn, i, e):
    elbos.append(e)
    
with tyxe.poutine.local_reparameterization():
    bnn.fit(loader, optim, 10000, callback)
...
C:\Miniconda3\envs\pta\lib\site-packages\torch\distributions\transforms.py in _inv_call(self, y)
    155         """
    156         if self._cache_size == 0:
--> 157             return self._inverse(y)
    158         x_old, y_old = self._cached_x_y
    159         if y is y_old:

C:\Miniconda3\envs\pta\lib\site-packages\torch\distributions\transforms.py in _inverse(self, y)
    430 
    431     def _inverse(self, y):
--> 432         if y.dim() < self.codomain.event_dim:
    433             raise ValueError("Too few dimensions on input")
    434         return self.base_transform.inv(y)

AttributeError: 'PyroParam' object has no attribute 'dim'

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

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.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.