Changeset
What's new:
#56
Most latest updates:
- Rename project to
book-persona-retriever
- #55
- #58
- fill information into the references section
-
add logo image from the paper - 📙 Jupyter Notebook: cite studies
- Sync updated images with the one from the final paper.
- Emphasize speaker-names in dataset instead of the 1,2,3,4,5
- Simplify description by moving
se
-related content into the sepratedREADME.md
file - Clarify the nature of dataset:
* ProjectGutenberg licence limitations,
* how to compose split-based.
* version of the existed split by refering to the old branches. (4.5) - reforge another dataset for setting up ParlAI tasks.
- This code showcases the ParlAI problem registration for fixed split on to train and dev, composed from the 5 folds. We did not discover an easier way for embedding folds in ParlAI rather than providing predefined splits for individual experiments.
- Provide sections for limitations
⚠️
* NER
* corefernce resolution
*Clarify theproject-gutenberg
project details
Full Changelog: https://github.com/nicolay-r/book-persona-retriever/commits/v1.0