| 3 | |
| 4 | |
| 5 | class LitEma(nn.Module): |
| 6 | def __init__(self, model, decay=0.9999, use_num_upates=True): |
| 7 | super().__init__() |
| 8 | if decay < 0.0 or decay > 1.0: |
| 9 | raise ValueError("Decay must be between 0 and 1") |
| 10 | |
| 11 | self.m_name2s_name = {} |
| 12 | self.register_buffer("decay", torch.tensor(decay, dtype=torch.float32)) |
| 13 | self.register_buffer( |
| 14 | "num_updates", |
| 15 | torch.tensor(0, dtype=torch.int) if use_num_upates else torch.tensor(-1, dtype=torch.int), |
| 16 | ) |
| 17 | |
| 18 | for name, p in model.named_parameters(): |
| 19 | if p.requires_grad: |
| 20 | # remove as '.'-character is not allowed in buffers |
| 21 | s_name = name.replace(".", "") |
| 22 | self.m_name2s_name.update({name: s_name}) |
| 23 | self.register_buffer(s_name, p.clone().detach().data) |
| 24 | |
| 25 | self.collected_params = [] |
| 26 | |
| 27 | def reset_num_updates(self): |
| 28 | del self.num_updates |
| 29 | self.register_buffer("num_updates", torch.tensor(0, dtype=torch.int)) |
| 30 | |
| 31 | def forward(self, model): |
| 32 | decay = self.decay |
| 33 | |
| 34 | if self.num_updates >= 0: |
| 35 | self.num_updates += 1 |
| 36 | decay = min(self.decay, (1 + self.num_updates) / (10 + self.num_updates)) |
| 37 | |
| 38 | one_minus_decay = 1.0 - decay |
| 39 | |
| 40 | with torch.no_grad(): |
| 41 | m_param = dict(model.named_parameters()) |
| 42 | shadow_params = dict(self.named_buffers()) |
| 43 | |
| 44 | for key in m_param: |
| 45 | if m_param[key].requires_grad: |
| 46 | sname = self.m_name2s_name[key] |
| 47 | shadow_params[sname] = shadow_params[sname].type_as(m_param[key]) |
| 48 | shadow_params[sname].sub_(one_minus_decay * (shadow_params[sname] - m_param[key])) |
| 49 | else: |
| 50 | assert not key in self.m_name2s_name |
| 51 | |
| 52 | def copy_to(self, model): |
| 53 | m_param = dict(model.named_parameters()) |
| 54 | shadow_params = dict(self.named_buffers()) |
| 55 | for key in m_param: |
| 56 | if m_param[key].requires_grad: |
| 57 | m_param[key].data.copy_(shadow_params[self.m_name2s_name[key]].data) |
| 58 | else: |
| 59 | assert not key in self.m_name2s_name |
| 60 | |
| 61 | def store(self, parameters): |
| 62 | """ |