| {"log": "trained", "date": "2023-05-22T09:40:10", "elapsed": "00:10:24", "model": "kkuramitsu/mt5-mini9L", "max_length": 128, "target_max_length": 128, "batch_size": 64, "gradient_accumulation_steps": 1, "train_steps": 1106, "accelerator": "gpu", "devices": "auto", "precision": "16", "strategy": "deepspeed", "gradient_clip_val": 1.0, "compile": true, "solver": "adamw", "lr": 0.0003, "warmup_steps": 1, "training_steps": 100000, "adam_epsilon": 1e-08, "weight_decay": 0.0, "epoch": 11, "step": 605, "saved": "model"} | |