SUMMARY MODEL:

Model Params Size: 222882048
Model Params Size Formatted: 222.88 M
Model Disk Size: 891648255
Model Disk Size Formatted: 891.65 MB


TRAINING AND VALIDATION RESULTS:

Training batch size: 64
Validation batch size: 128
Total expected epochs: 40
Total expected trainig steps: 94040
Total expected trainig steps 2: 94040
Total trained epochs: 40.0
Total trained steps: 94040
Elapsed time: 54841.370337963104 seconds
Elapsed time (formatted): 15:14:01
Total flos: 3.6650496018087936e+18
Total flos (formatted): 3.665050e+18
Best epoch val_loss: 0.20909027755260468
Best model checkpoint: /root/pretrain_utg4java_02/checkpoint-91689



SUMMARY DATASETS:

Loaded Dataset:
DatasetDict({
    train: Dataset({
        features: ['text'],
        num_rows: 150523
    })
    valid: Dataset({
        features: ['text'],
        num_rows: 18816
    })
    test: Dataset({
        features: ['text'],
        num_rows: 18815
    })
})

Tokenized Dataset:
DatasetDict({
    train: Dataset({
        features: ['input_ids'],
        num_rows: 150523
    })
    valid: Dataset({
        features: ['input_ids'],
        num_rows: 18816
    })
    test: Dataset({
        features: ['input_ids'],
        num_rows: 18815
    })
})