|
…
|
||
|---|---|---|
| .. | ||
| experiments | ||
| README.md | ||
| __init__.py | ||
| teams.py | ||
| teams_experiments.py | ||
| teams_pretrainer.py | ||
| teams_pretrainer_test.py | ||
| teams_task.py | ||
| teams_task_test.py | ||
| train.py | ||
README.md
TEAMS (Training ELECTRA Augmented with Multi-word Selection)
Note: This project is working in progress and please stay tuned.
TEAMS is a text encoder pre-training method that simultaneously learns a generator and a discriminator using multi-task learning. We propose a new pre-training task, multi-word selection, and combine it with previous pre-training tasks for efficient encoder pre-training. We also develop two techniques, attention-based task-specific heads and partial layer sharing, to further improve pre-training effectiveness.
Our academic paper [1] which describes TEAMS in detail can be found here: https://arxiv.org/abs/2106.00139.
References
[1] Jiaming Shen, Jialu Liu, Tianqi Liu, Cong Yu and Jiawei Han, "Training ELECTRA Augmented with Multi-word Selection", Findings of the Association for Computational Linguistics: ACL 2021.