pyro
jupytext
pyro | jupytext | |
---|---|---|
9 | 20 | |
8,364 | 6,425 | |
0.5% | - | |
8.4 | 8.8 | |
10 days ago | 1 day ago | |
Python | Python | |
Apache License 2.0 | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
pyro
-
Show HN: Designing Bridges with PyTorch
Mostly I use pytorch for statistical modeling https://pyro.ai . Under the hood that package uses a lot of Monte Carlo integration and variational methods (i.e. integration by optimization). It does support neural nets, but probably >80% of pyro users stick to simpler hierarchical Bayesian models.
- Pyro: The Universal, Probablistic Programming Language
- The Jupyter+Git problem is now solved
- Pyro: Deep universal probabilistic programming with Python and PyTorch
-
Computational Bayesian Inference Techniques
Amortized Variational Inference (Like done in pyro.ai with neural networks)
-
[P] torchegranate: a PyTorch rewrite of the pomegranate library for probabilistic modeling
Can you compare this to Pyro, which is also built on top of PyTorch?
- [Q] Updated book or review paper on MCMC methods
- Is anyone here working in uncertainty estimation in neural networks?
-
[D] Do you train and deploy models using just one framework or multiple frameworks at work?
Using pyod, statmodels, scikit-learn, Tensorflow and pyro.ai (that is using PyTorch as backend). I always use the same framework for training and for production.
jupytext
- The Jupyter+Git problem is now solved
-
Do you git commit jupyter notebooks?
Jupytext (https://github.com/mwouts/jupytext) has been designed exactly for this
-
The hatred towards jupyter notebooks
jupytext is your friend.
-
Edit notebooks in Google cloud
So if you run your own jupyter server, -jupy+text can be a great workflow : it takes your notebook synchronized with other formats (python file, makdown, ...), so you can edit your py/md file with neovim, and refresh the browser to execute the notebook.
-
Rant: Jupyter notebooks are trash.
Automatically convert ipynb files to py when saving them on JupyterLab
-
Two questions regarding working with jupyter notebooks (git, vim)
I don't use Jupyter so I don't know for sure, but on a quick glance you might want to look at https://github.com/mwouts/jupytext to see if that could help at all.
-
JupyterLite is a JupyterLab distribution that runs in the browser
The format is only partially invented, it follows Jupytext [0], but adds support for cell metadata. There is no obvious way to get that in fenced codeblocks, especially with the ability to spread it over multiple lines so it plays well with version control.
One more consideration is that it's not "Markdown with code blocks interspersed", one might as well use plaintext or AsciiDoc.
Of course there are tradeoffs.. I wish I had more time to work on it.
[0]: https://github.com/gzuidhof/starboard-notebook/blob/master/d...
[1]: https://github.com/mwouts/jupytext
-
Many write research papers in R Markdown - What is the alternative setup in Python?
Using jupytext (allows you to open .md files as notebooks) + jupyter gives you pretty much the same experience. The main issue is that the cell's output will be discarded. To fix it, you can use ploomber to generate an output HTML, so the workflow goes like this:
-
Jupyter Notebooks.
First, the format. The ipynb format does not play nicely with git since it stores the cell's source code and output in the same file. But Jupyter has built-in mechanisms to allow other formats to look like notebooks. For example, here's a library that allows you to store notebooks on a postgres database (I know this isn't practical, but it's a great example). To give more practical advice, jupytext allows you to open .py files as notebooks. So you can develop interactively but in the backend, you're storing .py files.
What are some alternatives?
PyMC - Bayesian Modeling and Probabilistic Programming in Python
jupyter - An interface to communicate with Jupyter kernels.
scikit-learn - scikit-learn: machine learning in Python
rmarkdown - Dynamic Documents for R
Prophet - Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.
sagemaker-run-notebook - Tools to run Jupyter notebooks as jobs in Amazon SageMaker - ad hoc, on a schedule, or in response to events
trueskill - An implementation of the TrueSkill rating system for Python
nbdev - Create delightful software with Jupyter Notebooks
probability - Probabilistic reasoning and statistical analysis in TensorFlow
papermill - 📚 Parameterize, execute, and analyze notebooks
Keras - Deep Learning for humans
nbdime - Tools for diffing and merging of Jupyter notebooks.