Comments (7)
Great contribution @kevinykuo !
I was just reviewing your project before adding the link to the README and noticed that you are using an old version of the code.
Yesterday we released a new version, with a cleaner Python API and broader support for discrete variables, so I think it would be better if you replaced the current installation and the code copy with a simple ctgan==0.2.0
install from PyPI (use plain pip instead of conda). If you do this you should be able to also remove all the other dependencies (numpy, scipy...), as they will be pulled alongside ctgan
by pip
.
Also notice that we moved epochs
from __init__
to fit
, as you suggested, but we kept batch_size
in the __init__
, as this is something that gets stored in the instance and used later on when sampling. So maybe it would be a good idea if you mirrored the same API structure, so both implementations are consistent with each other?
Also, keep an eye on the releases, as other things will be added soon as well (like the save/load functionality)
from ctgan.
Yep I have been watching the repo, and will refactor accordingly. There were some hacks I had to make to get it to work with R. Ideally we can just grab ctgan from pip to lessen maintenance burden :)
I'll keep an eye out on new features, and ensure consistency as much as possible to reduce friction for teams working with different APIs. Note that the R package currently supports save/load already, and we need to have a custom implementation there since we also serialize R-specific metadata. This is partly because we also made the design decision to infer discrete columns from column types, which is more Rtistic (by convention, categorical variables are factors or characters.)
from ctgan.
@csala I was able to cleanly remove the inline hacks and depend directly on the new ctgan release! (see PR kasaai/ctgan#7) :)
The only issue left is a cosmetic one, which I've filed here #14
from ctgan.
Actually, a slightly less cosmetic issue here: #16 :P
from ctgan.
@csala do you mind cutting a patch/minor release incorporating the last couple PRs so we can streamline the dependencies for the R package and get it on CRAN? Planning to tell people about SDV/CTGAN at the RStudio developer conference next week :)
from ctgan.
Sure, we'll go for it today @kevinykuo
@csala I was able to cleanly remove the inline hacks and depend directly on the new ctgan release! (see PR kasaai/ctgan#7) :)
The only issue left is a cosmetic one, which I've filed here #14
Great news btw! Thanks for taking care of it!
from ctgan.
Release v0.2.1 is out @kevinykuo !
The link to R package has also been added to the readme, so closing this.
from ctgan.
Related Issues (20)
- TypeError while ctgan.fit() HOT 6
- Improve DataSampler efficiency
- ValueError: mismatch of shapes when sampling data for compas dataset HOT 2
- Return loss values as float values not PyTorch objects
- Transition from using setup.py to pyproject.toml to specify project metadata
- Remove bumpversion and use bump-my-version
- Switch to using ruff for Python linting and code formatting
- Add dependency checker
- Remove scikit-learn dependency
- CTGAN using deprecated 'sklearn' HOT 2
- Replace integration test that uses the iris demo data
- Add bandit workflow
- Feature Request: More verbose logging HOT 3
- Fix minimum version workflow when pointing to github branch
- Deployment requirements based on libtorch or ONNX HOT 4
- How to load this model directly to generate data after saving it HOT 4
- Cleanup automated PR workflows
- Remove FutureWarning: Setting an item of incompatible dtype is deprecated
- Only run unit and integration tests on oldest and latest python versions for macos
- [HELP] CTGAN has Reproducibility? HOT 8
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 ctgan.