https://github.com/google-research/tensorflow-coder
Science Score: 23.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
○codemeta.json file
-
○.zenodo.json file
-
✓DOI references
Found 5 DOI reference(s) in README -
✓Academic publication links
Links to: acm.org -
○Committers with academic emails
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (13.0%) to scientific vocabulary
Repository
Basic Info
- Host: GitHub
- Owner: google-research
- License: apache-2.0
- Language: Python
- Default Branch: master
- Size: 30.3 MB
Statistics
- Stars: 278
- Watchers: 14
- Forks: 34
- Open Issues: 1
- Releases: 0
Metadata Files
README.md
TensorFlow Coder (TF-Coder)
TF-Coder is a program synthesis tool that helps you write TensorFlow code. First, the tool asks for an input-output example of the desired tensor transformation. Then, it runs a combinatorial search to find TensorFlow expressions that perform that transformation. TF-Coder’s output is real TensorFlow code that you can include in your projects.
Quick Links
The TF-Coder tool is ready-to-use at this link. Everything is already packaged together in a Colab notebook, so no installation or download is needed.
For more information about TF-Coder, see the following documents: * TF-Coder Tutorial: walks you through using TF-Coder to solve tensor manipulation tasks, and provides tips on getting the most out of TF-Coder. * User Journeys: illustrates several realistic scenarios where TF-Coder can help accelerate your TensorFlow development in different ways.
Contents
What is TF-Coder?
When manipulating tensors, one must keep track of multiple dimensions, tensor shape and DType compatibility, and of course mathematical correctness. Additionally, there are hundreds of TensorFlow operations, and finding the right ones to use can be a challenge.
TensorFlow Coder, or TF-Coder, can help you write tricky tensor manipulations in TensorFlow. Instead of coding your tensor manipulation directly, you can just demonstrate it through an illustrative input-output example, and TF-Coder can produce the corresponding code automatically. TF-Coder performs an efficient combinatorial search over compositions of TensorFlow operations, until it finds a TensorFlow expression that matches the given input-output example.
TF-Coder allows you to: * Program in TensorFlow by example * Find the right function to use * Automatically combine functions in clever ways * Spend less time debugging
TF-Coder is primarily a development tool for TensorFlow users. If you just want to use TF-Coder as a tool, you don’t need to install anything, as the tool is ready-to-use in this Colab notebook.
Caveats
There are limitations to TF-Coder. It can currently find solutions involving 3-4 operations within a minute of searching, but solutions involving 6 or more operations are too complex to find in a reasonable amount of time. Furthermore, TF-Coder currently does not support complex or string tensors, or RaggedTensors. The full list of supported operations can be found in the Colab notebook.
In addition, TF-Coder only guarantees that its solutions work for the given input-output example. The tool searches for a simple TensorFlow expression that matches the provided input-output example, but sometimes this solution is too simple and doesn’t generalize in the intended way. It can be helpful to make the example as unambiguous as possible, which can often be achieved by adding more numbers to the input and output tensors. Please review TF-Coder’s solutions to ensure that they correctly implement the intended behavior.
In the Colab tool, we would like to log the problems given to TF-Coder and the resulting solutions, so that we can improve the tool and build a dataset that will accelerate program synthesis research in general, but this data collection is completely optional.
Tutorial and Further Reading
For more information about TF-Coder, see the following documents: * TF-Coder Tutorial: walks you through using TF-Coder to solve tensor manipulation tasks, and provides tips on getting the most out of TF-Coder. * User Journeys: illustrates several realistic scenarios where TF-Coder can help accelerate your TensorFlow development in different ways. * Our research paper: describes the technology behind TF-Coder.
Optional: Using TF-Coder Outside Colab
Because TF-Coder is primarily a development tool and not a library that you use in your code, we hope that the provided Colab notebook is sufficient for your use cases.
However, if you would rather not use the Colab notebook, you can still install
TF-Coder as a Python package yourself:
pip install --user tensorflow-coder
To run the TF-Coder search as a library, follow the code example in
tf_coder_main.py.
To run TF-Coder on our benchmarks, run:
python3 tf_coder/value_search/value_search_main.py
To run tests, clone the repository and run pytest.
Citation
If you find TF-Coder helpful for a research project, you may cite our research
paper as follows:
@article{TFCoder,
title={{TF-Coder}: Program Synthesis for Tensor Manipulations},
author={Kensen Shi and David Bieber and Rishabh Singh},
journal={ACM Transactions on Programming Languages and Systems (TOPLAS)},
volume={44},
number={2},
pages={1--36},
year={2022},
publisher={Association for Computing Machinery},
url={https://doi.org/10.1145/3517034},
doi={10.1145/3517034},
}
Disclaimer
This is a research project, not an official Google product.
To report a bug or make a feature request, please raise a GitHub issue.
Owner
- Name: Google Research
- Login: google-research
- Kind: organization
- Location: Earth
- Website: https://research.google
- Repositories: 226
- Profile: https://github.com/google-research
GitHub Events
Total
- Issues event: 1
- Watch event: 3
- Issue comment event: 3
- Fork event: 1
Last Year
- Issues event: 1
- Watch event: 3
- Issue comment event: 3
- Fork event: 1
Committers
Last synced: 7 months ago
Top Committers
| Name | Commits | |
|---|---|---|
| Kensen Shi | 6****s@u****m | 21 |
| Christopher Akiki | c****i@p****m | 1 |
Issues and Pull Requests
Last synced: 8 months ago
All Time
- Total issues: 5
- Total pull requests: 3
- Average time to close issues: 6 months
- Average time to close pull requests: about 1 month
- Total issue authors: 5
- Total pull request authors: 3
- Average comments per issue: 1.2
- Average comments per pull request: 1.33
- Merged pull requests: 1
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 1
- Pull requests: 0
- Average time to close issues: 6 days
- Average time to close pull requests: N/A
- Issue authors: 1
- Pull request authors: 0
- Average comments per issue: 3.0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Top Authors
Issue Authors
- programsynthesis-cn (1)
- g-dspencer (1)
- liujqian (1)
- jnd77 (1)
- DLPerf (1)
Pull Request Authors
- cakiki (1)
- dbieber (1)
- DLPerf (1)
Top Labels
Issue Labels
Pull Request Labels
Packages
- Total packages: 2
-
Total downloads:
- pypi 85 last-month
-
Total dependent packages: 0
(may contain duplicates) -
Total dependent repositories: 2
(may contain duplicates) - Total versions: 9
- Total maintainers: 1
pypi.org: tensorflow-coder
TensorFlow Coder (TF-Coder): A Program Synthesis Tool for TensorFlow
- Homepage: https://github.com/google-research/tensorflow-coder
- Documentation: https://tensorflow-coder.readthedocs.io/
- License: Apache Software License
-
Latest release: 0.0.5
published over 4 years ago
Rankings
Maintainers (1)
pypi.org: tensorflow-coder-colab-logging
Logging utilities for TensorFlow Coder's Colab interface
- Homepage: https://github.com/google-research/tensorflow-coder
- Documentation: https://tensorflow-coder-colab-logging.readthedocs.io/
- License: Apache Software License
-
Latest release: 0.0.4
published over 4 years ago