|
30 | 30 | AdaBound, |
31 | 31 | AdaDelta, |
32 | 32 | AdaFactor, |
| 33 | + AdaGC, |
33 | 34 | AdaHessian, |
34 | 35 | Adai, |
35 | 36 | Adalite, |
|
86 | 87 | SGDSaI, |
87 | 88 | Shampoo, |
88 | 89 | SignSGD, |
| 90 | + SimplifiedAdEMAMix, |
89 | 91 | SophiaH, |
90 | 92 | StableAdamW, |
91 | 93 | StableSPAM, |
|
513 | 515 | (AdamG, {'lr': 1e0, 'p': 0.5}, 60), |
514 | 516 | (AdEMAMix, {'lr': 1e0}, 3), |
515 | 517 | (AdEMAMix, {'lr': 1e0, 't_alpha_beta3': 5}, 3), |
| 518 | + (SimplifiedAdEMAMix, {'lr': 1e0, 'beta1_warmup': 2, 'weight_decay': 1e-3}, 3), |
516 | 519 | ( |
517 | 520 | SOAP, |
518 | 521 | {'lr': 1e0, 'shampoo_beta': 0.95, 'precondition_frequency': 1, 'merge_dims': False, 'precondition_1d': True}, |
|
567 | 570 | (EXAdam, {'lr': 1e-1, 'weight_decay': 1e-3}, 5), |
568 | 571 | (SCION, {'lr': 5e-1, 'constraint': False, 'weight_decay': 1e-3}, 5), |
569 | 572 | (SCION, {'lr': 1e-1, 'constraint': True, 'lmo_type': 'col_norm'}, 10), |
| 573 | + (AdaGC, {'lr': 1e-1, 'warmup_steps': 2, 'weight_decay': 1e-3}, 5), |
570 | 574 | (Ranger25, {'lr': 1e-1}, 3), |
571 | 575 | (Ranger25, {'lr': 1e-1, 't_alpha_beta3': 5}, 3), |
572 | 576 | (Ranger25, {'lr': 5e-2, 'stable_adamw': False, 'orthograd': False, 'eps': None, 'lookahead_merge_time': 2}, 3), |
|
0 commit comments