|
58 | 58 | (Lamb, {'lr': 1e-1, 'weight_decay': 1e-3, 'adam': True, 'eps': 1e-8}, 500), |
59 | 59 | (Lamb, {'lr': 1e-1, 'weight_decay': 1e-3, 'pre_norm': True, 'eps': 1e-8}, 500), |
60 | 60 | (LARS, {'lr': 1e-1, 'weight_decay': 1e-3}, 500), |
61 | | - (RaLamb, {'lr': 2e-1, 'weight_decay': 1e-3}, 500), |
62 | | - (RaLamb, {'lr': 5e-1, 'weight_decay': 1e-3, 'pre_norm': True}, 500), |
| 61 | + (RaLamb, {'lr': 1e-2, 'weight_decay': 1e-4}, 200), |
| 62 | + (RaLamb, {'lr': 1e-2, 'weight_decay': 1e-4, 'pre_norm': True}, 200), |
63 | 63 | # (RaLamb, {'lr': 1e-1, 'weight_decay': 1e-3, 'degenerated_to_sgd': True}, 200), |
64 | 64 | (MADGRAD, {'lr': 1e-2, 'weight_decay': 1e-3}, 500), |
65 | 65 | (MADGRAD, {'lr': 1e-2, 'weight_decay': 1e-3, 'eps': 0.0}, 500), |
|
0 commit comments