import torch from torch import nn class LitEma(nn.Module): def __init__(self, model, decay=0.9999, init_num_updates = 0, use_num_upates=True): super().__init__() if decay < 0.0 or decay > 1.0: raise ValueError('Decay must be between 0 and 1') self.m_name2s_name = {} self.register_buffer('decay', torch.tensor(decay, dtype=torch.float32)) self.register_buffer('num_updates', torch.tensor(init_num_updates, dtype=torch.int) if use_num_upates else torch.tensor(-1, dtype=torch.int)) # 0 for name, p in model.named_parameters(): if p.requires_grad: # remove as '.'-character is not allowed in buffers s_name = name.replace('.', '') self.m_name2s_name.update({name: s_name}) self.register_buffer(s_name, p.clone().detach().data) self.collected_params = [] def reset_num_updates(self): del self.num_updates self.register_buffer('num_updates', torch.tensor(0, dtype=torch.int)) def forward(self, model): decay = self.decay if self.num_updates >= 0: self.num_updates += 1 decay = min(self.decay, (1 + self.num_updates) / (10 + self.num_updates)) one_minus_decay = 1.0 - decay with torch.no_grad(): m_param = dict(model.named_parameters()) shadow_params = dict(self.named_buffers()) for key in m_param: if m_param[key].requires_grad: sname = self.m_name2s_name[key] shadow_params[sname] = shadow_params[sname].type_as(m_param[key]) shadow_params[sname].sub_(one_minus_decay * (shadow_params[sname] - m_param[key])) else: assert not key in self.m_name2s_name def copy_to(self, model): m_param = dict(model.named_parameters()) shadow_params = dict(self.named_buffers()) for key in m_param: if m_param[key].requires_grad: m_param[key].data.copy_(shadow_params[self.m_name2s_name[key]].data) else: assert not key in self.m_name2s_name def store(self, parameters): """ Save the current parameters for restoring later. Args: parameters: Iterable of `torch.nn.Parameter`; the parameters to be temporarily stored. """ self.collected_params = [param.clone() for param in parameters] def restore(self, parameters): """ Restore the parameters stored with the `store` method. Useful to validate the model with EMA parameters without affecting the original optimization process. Store the parameters before the `copy_to` method. After validation (or model saving), use this to restore the former parameters. Args: parameters: Iterable of `torch.nn.Parameter`; the parameters to be updated with the stored parameters. """ for c_param, param in zip(self.collected_params, parameters): param.data.copy_(c_param.data)