- Replace context.iteration with consumed_samples (global sample count) - Add optimizer_step property derived from consumed_samples - Checkpoint meta.json stores consumed_samples, drops iteration - CLI --start_batch renamed to --start_samples (per-rank samples) - Checkpoint dir naming: epoch_X_step_Y instead of epoch_X_iter_Y - Metric log entries use step and consumed_samples fields - Backward compat removed (old iteration checkpoints unsupported) |
||
|---|---|---|
| .. | ||
| __init__.py | ||
| base.py | ||
| model_config.py | ||
| preprocess_config.py | ||
| train_config.py | ||