Releases: funstory-ai/aifw
OneAIFW release 0.5.1 -- Add python binding and web server
- Add python binding aifw-py for aifw core library and NER pipeline
- Add python web server based on aifw-py
- Add config API interface for config aifw pipeline in runtime
- Fix issue of incorrect alignment when free memory
build_docker=true
OneAIFW release v0.5
It adds independent control over anonymized entities; each entity type can independently control whether to anonymize.
It also includes sophisticated Chinese address recognition capabilities.
OneAIFW release 0.4 - Self manage models resource
The aifw-js library now manages its model file resources by default.
The init interface function has been modified; it no longer requires parameters by default. If you need to download resources from a specified address, you must pass in the corresponding parameters.
Currently, it supports downloading from different remote sources. Different download sources are used for users in China and overseas.
Release v0.3.1 - improve to recognize zh address
Improve recognization of zh address for complex address
Release v0.3 - py-origin and aifw-js is ready
Add support authorization in OneAIFW http server.
The aifw-js package is ready to integrate to imt-translator extension of web browser.
Optional to create aifw docker image.
build_docker=true
The OneAIFW release version v0.2.1
Chang format of response of http API interface to {output, error}
The OneAIFW release version v0.2
Add aifw core library that written by Zig language.
Add aifw-js library that is a JS wrapper library for aifw core library.
Add four http API interface: mask_text, restore_text, mask_text_batch, restore_text_batch
The OneAIFW release version v0.1
The OneAIFW is can success to anonymize the privacy information in text.