Releases: accelerated-text/accelerated-text
v0.9.2
We are bringing Python integration with this release! You can now load your data, choose the document plans, run text generation, and get results - all from your Python code. For the details on other bug fixes and improvements, see linked issues.
v0.9.0
This version comes with better organization of the document plan files and introduction of the text generation project. Now you can set up a separate location for all your customization and document plans based on a project template. We have also made the whole package much leaner with refactored UI.
Changelog
- Previous versions of Accelerated Text had some document plans present by default. Not anymore. Now it starts with an empty environment. You will have to create a new project following the template. We will be following up with sample text generation projects.
- Text Segment blocks now can be used to introduce paragraphs into the generated text.
- We have introduced a better organization of exported document plans. Now they are saved into separate folders based on their type.
- Refactored front-end code to achieve a significant reduction in UI compilation times and resource use.
- Documentation improvements: details on how to run on Windows and MacOS.
- MS Excel support for data file upload.
Autumn release
The highlights of this release are data enrichment and reader model functionality. With the first, raw incoming data can be modified before it appears in the generated text. With the latter, text generation can be adopted to different target audiences.
Given the two new features and a data point coming from the database Q1 profit = 130050, a more promotional text for the public Q1 profits were over $130k can be generated. While Q1 profits were $130050 version would be generated for the financial analyst.
Changelog
- Reader models personalises text based on predefined user groups.
- Data enrichment defines a set of rules that transform data before the generation.
- Fixed exporting generated text from a large input data file.
- Performance improvements for batch generation.
Summer release
The highlight of this release is the improved support for the list blocks. Now we can enumerate things in text better. Complete list of improvements:
- Ability to edit dictionary item forms in UI
- Faster startup time with pre-built docker images
- Improved list block text generation
- Multiple definition blocks for the same variable now produce variations
- Support for template-like AMRs
- Various bug fixes
Spring release - a first public version
With this release Accelerated Text has:
- Full documentation
- All NLG components in place
- Basic Abstract Meaning Representation construction blocks
- Basic Grammatical construction blocks