You can not select more than 25 topics
Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
syiswell
b852d11257
|
1 year ago | |
|---|---|---|
| .. | ||
| data | 1 year ago | |
| saved_models | 1 year ago | |
| utils | 1 year ago | |
| README.md | 1 year ago | |
| prepare_data.py | 1 year ago | |
| run.py | 1 year ago | |
| to_bioes.py | 1 year ago | |
README.md
Prerequisites
python 3.7.12
pytorch 1.8.1
cuda 10.2
transformers 4.12.5
json 2.0.9
numpy 1.21.2
pandas 1.3.4
dgl-cu102 0.6.1
nltk 3.6.5
Descriptions
data - contains dataset.
bert-base-uncased: put the download Pytorch bert model here (config.json, pytorch_model.bin, vocab.txt) (https://huggingface.co/bert-base-uncased/tree/main).rr-submission: contains RR dataset, download from (https://github.com/LiyingCheng95/ArgumentPairExtraction/tree/master/data/rr-submission)
saved_models - contains saved models, training logs and results.
utils - utils code.
config.py: parameter setting.fun.py: functions.metrics.py: evaluation metrics.models.py: proposed model.
prepare_data.py - prepare the input data.
run.py - train and evaluate the proposed transition-based model.
to_bioes.py - change to bioes label.
Usage
python prepare_data.py
python to_bioes.py
python run.py