bert-large-sst2 / run_arguments.json
amodaresi's picture
initial model upload
4dd37fb
raw
history blame
711 Bytes
{
"model_name_or_path": "bert-large-uncased",
"task_name": "sst2",
"output_dir": "./output/bert-large/sst2/0001/42/",
"learning_rate": 3e-05,
"num_train_epochs": 5,
"per_device_eval_batch_size": 32,
"per_device_train_batch_size": 32,
"weight_decay": 0.1,
"seed": 42,
"warmup_steps": 0,
"do_train": true,
"do_eval": true,
"do_predict": false,
"pad_to_max_length": false,
"max_seq_length": 128,
"report_to": [],
"save_strategy": "no",
"evaluation_strategy": "epoch",
"logging_steps": 2500,
"use_fast_tokenizer": true,
"group_by_length": true,
"save_training_dynamics": false,
"save_training_dynamics_after_epoch": false
}