- Add original infnet setup (Tu, Gimpel) in the reverse order #32
- Add unnormalized score loss for training tasknn #30
- Single sequence tagging base for all cases of srl and ner #63 (dhruvdcoder)
- Changes to make sequence-tagging work again #61 (dhruvdcoder)
- Adds SPEN Loss and configs #58 (purujitgoyal)
- Adapters and configs with blurb genre dataset #55 (dhruvdcoder)
- Adds Best Run Configs #50 (purujitgoyal)
- Made changes for applying ranking loss directly to the TaskNN #48 (leejayyoon)
- Score-nn Evaluation #42 (purujitgoyal)
- Text MLC with any text encoder including BERT #37 (dhruvdcoder)
- Default "nce discrete" jsonnet has "-" signs now. #51 (leejayyoon)
- Making GBI work again. #47 (dhruvdcoder)
- Add new dependency version info in README #27 (leejayyoon)
- Tracking notebooks using jupytext #54 (dhruvdcoder)
- Automatic generation of changelog #23
- NYT and RCV text datasets, readers and configs #59 (dhruvdcoder)
- No cross entropy configs for DVN+tasknn #53 (dhruvdcoder)
- Feat/from pretrained dvn #52 (dhruvdcoder)
- Added sweeps & jsonnet for eurlex (and other small sweep config changes) #45 (leejayyoon)
- Organized sweep configs for general data & specific datasets. #40 (leejayyoon)
- Scripts to count instances and print env vars in jsonnet #60 (dhruvdcoder)
Merged pull requests:
- Added expr and spo, go and fun datasets. #44 (dhruvdcoder)
v0.1.0 (2021-07-08)
* This Changelog was automatically generated by github_changelog_generator