| import torch | |
| def get_optimizer( | |
| configs, model: torch.nn.Module | |
| ) -> torch.optim.Optimizer: | |
| param_groups = [{"params": model.parameters(), "lr": configs.lr}] | |
| if configs.type == "adam": | |
| optimizer = torch.optim.Adam( | |
| param_groups, | |
| lr=configs.lr, | |
| weight_decay=configs.weight_decay, | |
| betas=(configs.beta1, configs.beta2), | |
| ) | |
| else: | |
| raise ValueError(f"Invalid optimizer: [{configs.type}]") | |
| return optimizer |