Datasets:

Modalities:
Tabular
Text
Formats:
csv
Languages:
English
ArXiv:
Libraries:
Datasets
pandas
License:
CREHate / finetune /config_files /BERTweet-CREHate-ALL-train.json
Nayeon Lee
Upload 35 files
f1b5d0c verified
raw
history blame contribute delete
908 Bytes
{
"model" : "vinai/bertweet-base",
"max_length":128,
"checkpoint_filename" : "./checkpoints/[US]BERTweet-base_CREHate_checkpoint.pth.tar",
"best_filename" : "./checkpoints/[US]BERTweet-base_CREHate_best.pth.tar",
"additional_tokens" : ["[US]","[AU]","[GB]","[ZA]","[SG]","@USER","URL"],
"remove_special_tokens" :true,
"train_data" : "./data_splits/CREHate_train.csv",
"val_data" : "./data_splits/CREHate_valid.csv",
"test_data" : "./data_splits/CREHate_test_res.csv",
"test_res" : "./data_splits/CREHate_test_res.csv",
"sent_col" : "Text" ,
"label_col" : "United_States_Hate" ,
"num_labels" : 2,
"test_col" : "Text",
"test_res_col" : "BERTweet-CREHate-US",
"batch_size" : 32 ,
"num_workers" : 4,
"distributed":true,
"train" : true,
"evaluate" : true,
"test" : true,
"resume" : false,
"resume_model" : ""
}