Files
MSTSb_paraphrase-xlm-r-mult…/train_arguments.json

1 line
190 B
JSON
Raw Permalink Normal View History

{"num_epochs": 2, "train_batch_size": 64, "evaluation_steps": 1000, "scheduler": "WarmupLinear", "warmup_steps": 288, "optimizer_params": {"lr": 4e-05}, "use_amp": true, "weight_decay": 0.1}