arekit-ss-0.25.0

What's new

See the following thread: #79

Full Changelog: v0.24.0...v0.25.0

arekit-ss-0.24.0

What's new in 0.24.0

Main feature is support of the custom documents

Backend Updates and Extended Schemata

• Switch to the AREkit==0.24.0 (be14ffc)

Quality of the sampled data

According to the #52 experiments:

• #70
• #71

Dynamic prompting support

When analysing NEREL-bio outputs, it was found that some labels and relations are better to describe more precisely with prompts

• #67
• #66

Simplify new sources registration

There were few feedbacks on bad API with labels and that whole project might be crashed because of the incompletely registered source

• #62
• #58
• #59

Other / Minor updates

• #68
• #69
• #42
• doc_ids argument support which allows to select the specific doc_ids for processing (90488c2)
• [output_dir] customization removed (234ae21)
• #72
• dest_lang is optional parameter (by default is src_lang) (234ae21)
• #74
• 🔧 #75
• #76

arekit-ss-0.23.1

Main goal of this release: #34
At this stage, we have a text parser presets that could be splitted into smaller parts and utilized in future.

Patch

• ❌ #57
• 🔧 #60
• 🔧 #61

Main Changes

• 🔧 remove update_arekit.sh script
• 📓 provide logo.png for the project
• 📓 rename reference section with Powered by AREkit.
• prompting techniques could be mentioned not in reference, but for a prompt-related comment
• #37
• 🔧 #43
• 📓 : Explain how to read the project name ("arekit double s")
• 📓 Mention image of the results in the README.md (Google colab 🪄 results)
• Move examples from the notebook into the test folder
• Clarify in filename that nn is only for sentiment https://github.com/nicolay-r/arekit-ss/blob/7785455be33c6be77cdf2ec967c8eb35c6bfa10f/arekit_ss/framework/arekit/rows_nn.py#L14
• quick fix ident https://github.com/nicolay-r/arekit-ss/blob/7785455be33c6be77cdf2ec967c8eb35c6bfa10f/arekit_ss/text_parser/text_nn_frames.py#L35