ALEA Data Generator
This is a basic synthetic data generation/perturbation library designed to support the creation or augmentation designed by the ALEA Institute to support the creation and augmentation of data without relying on "tainted" LLMs.
Data generation techniques in this library:
- do not require the use of any LLM or external data source
- can be used with KL3M, our Fairly Trained LLM
Supported Patterns
The following data generation patterns are supported:
- Simple string templates with sampled values (e.g.,
This Agreement, by and between <|company:a|> and <|company:b|>, is made as of <|date|>.)- Faker integration for common data types (e.g., names, addresses, dates, etc.)
- Large templates with sampled values (e.g.,
jinja2templates in files) - Common document types (e.g., emails, contracts, memos, etc. using templates)
- Data perturbation (e.g., realistic errors introduced by humans, OCR, or other automated systems)
- Skipping, doubling, or transposing/swapping characters
- Skipping, doubling, or transposing/swapping tokens
- QWERTY and mobile keyboard mistakes (off-by-one key, shift errors, etc.)
- Homophones (e.g.,
theirvs.there) - Synonyms (e.g.,
bigvs.large) - Negation/antonyms (e.g.,
bigvs.small) - Capitalization errors (e.g.,
bigvs.Big) - Punctuation errors (e.g.,
bigvs.big.) - OCR-like errors (e.g., misreading characters, smudges, etc.) -
- Representation conversion (e.g.,
429tofour hundred twenty-nineorfour twenty-nine) - Format conversion (e.g., Markdown <-> HTML variants)
Future Roadmap
- Document image generation for document/OCR models
License
The ALEA Data Generator library is released under the MIT License. See the LICENSE file for details.
Some of the data generation techniques used in this library may also retrieve data from external sources,
which have their own licensing terms. These terms are documented in the alea-data-sources here:
See, e.g., the CMU Pronouncing Dictionary (cmudict), which is used in tasks like homophonic errors:
Support
If you encounter any issues or have questions about using the ALEA Data Generator library, please open an issue on GitHub.
Learn More
To learn more about ALEA and its software and research projects like KL3M and leeky, visit the ALEA website.
Metadata
Release files for alea-data-generator 0.1.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| alea_data_generator-0.1.2.tar.gz | 25.1 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| alea_data_generator-0.1.2-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 66.8 kB
Release files / alea_data_generator-0.1.2.tar.gz
| Download URL | alea_data_generator-0.1.2.tar.gz |
|---|---|
| Size | 25.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
776f085a5c7cb31a48321edcaee407bacd9c965e25f129a958eaf7fd793a7b61
|
|
BLAKE2b-256 checksum How to use checksums |
ac6fdaf373ca824667244fbaa6dcb752007907210ae134c33e740e86943462e9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.8.4 CPython/3.12.3 Linux/6.8.0-50-generic
|
Release files / alea_data_generator-0.1.2-py3-none-any.whl
| Download URL | alea_data_generator-0.1.2-py3-none-any.whl |
|---|---|
| Size | 41.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
17db9ee84c8e4ce6d6dd97ac2cde01f75888dd00805f7265414b00ac9a47a0af
|
|
BLAKE2b-256 checksum How to use checksums |
6a67985ea3d47208a058b7996c3a458428b12e68c6f65ee2b9b39741e83bf8ba
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/1.8.4 CPython/3.12.3 Linux/6.8.0-50-generic
|