Loading main_Critic.py 0 → 100644 +62 −0 Original line number Diff line number Diff line # import # hyperparameters epochs=20 batch_size=20 learning_rate=0.001 train_dataloader = torch.utils.data.DataLoader(dataset.train, batch_size=batch_size, shuffle=True, collate_fn = lambda x: x) # load state dict from trainiing model_actor_only = ActorOnlySummarisationModel() model_actor_only.load_state_dict('model_actor_only_wts.pth') model_actor_only.eval() m = Critic(model_actor_only) since = time.time() val_loss_history = [] best_model_wts = copy.deepcopy(m.state_dict()) for e in range(epochs): print('Epoch {}/{}'.format(e, epochs - 1)) print('-' * 10) # train phase epoch_loss = m._epoch(train_dataloader) # collate_fn print('Train Loss: {:.4f}'.format(epoch_loss)) # validation phase val_epoch_loss = m._epoch(dataset.validation) val_loss_history.append(val_epoch_loss) print('Validation Loss: {:.4f}'.format(val_epoch_loss)) # epoch completed, deep copy the best model sofar if val_epoch_loss < best_loss: best_loss = val_epoch_loss best_model_wts = copy.deepcopy(m.state_dict()) # after training completed time_elapsed = time.time() - since print('Training complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60)) print('Best val loss: {:4f}'.format(best_loss)) # write val_rouge_history in file # load best model weights m.load_state_dict(best_model_wts) # testing since = time.time() test_loss = m.test(dataset.test) print('Test Loss: {:.4f}'.format(test_loss)) # after testing completed time_elapsed = time.time() - since print('Testing complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60)) models.py +112 −30 Original line number Diff line number Diff line Loading @@ -221,18 +221,64 @@ class Critic(nn.Module): nn.functional.relu(self.layer_2( utils.gaussian(self.layer_1(double_document))))))) def _train(self, dataset, epochs=200, batch_size=20, learning_rate=0.001, shuffle=True, pos_samples=0.5): # move to main # def _train(self, dataset, epochs=200, batch_size=20, learning_rate=0.001, shuffle=True, pos_samples=0.5): # move to main optimizer = torch.optim.Adam(self.parameters(), lr=learning_rate) loss_fn = nn.MSELoss() # optimizer = torch.optim.Adam(self.parameters(), lr=learning_rate) # loss_fn = nn.MSELoss() for _ in range(epochs): # for _ in range(epochs): training_dataloader = torch.utils.data.DataLoader(dataset, batch_size=batch_size, shuffle=shuffle) # training_dataloader = torch.utils.data.DataLoader(dataset, batch_size=batch_size, shuffle=shuffle) for batch in training_dataloader: # for batch in training_dataloader: optimizer.zero_grad() # optimizer.zero_grad() # for datapoint in batch: # r = np.random.random() # if r > pos_samples: # k = np.random.choice(len(datapoint.p_searchspace)) # sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs # score = self.__call__(sample, datapoint.gold_sent_vecs) # loss = loss_fn(score, datapoint.top_rouge[k]) # else: # if len(datapoint.sent_vecs) >= 3: # narray = np.random.choice(len(datapoint.sent_vecs), 3, replace = False) # narray.sort() # sample = datapoint.sent_vecs[narray] # else: # continue # handle len(sent_vecs) < 3 # score = self.__call__(sample, datapoint.gold_sent_vecs) # loss = loss_fn(score, utils.rouge(raw_document[narray]), raw_summary)) # # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? # loss.backward() # optimizer.step() # # eval # # test with rouge def training_epoch(self, dataloader, learning_rate=None): if learning_rate != None: for g in self.optimizer.param_groups: g['lr'] = learning_rate self.train() pos_samples= 0.5 epoch_loss = 0.0 for batch in train_dataloader: self.optimizer.zero_grad() for datapoint in batch: Loading @@ -243,7 +289,7 @@ class Critic(nn.Module): sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs score = self.__call__(sample, datapoint.gold_sent_vecs) loss = loss_fn(score, datapoint.top_rouge[k]) loss = self.loss_fn(score, datapoint.top_rouge[k]) else: if len(datapoint.sent_vecs) >= 3: Loading @@ -254,15 +300,51 @@ class Critic(nn.Module): continue # handle len(sent_vecs) < 3 score = self.__call__(sample, datapoint.gold_sent_vecs) loss = loss_fn(score, utils.rouge(raw_document[narray]), raw_summary)) loss = self.loss_fn(score, rouge(datapoint.raw_document[narray]), datapoint.raw_summary) # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? epoch_loss += loss.item() loss.backward() optimizer.step() self.optimizer.step() return epoch_loss / len(dataloader.dataset) def test(self, dataloader): self.eval() pos_samples= 0.5 epoch_loss = 0.0 with torch.no_grad(): for batch in dataloader: # eval for datapoint in batch: # test with rouge r = np.random.random() if r > pos_samples: k = np.random.choice(len(datapoint.p_searchspace)) sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs score = self.__call__(sample, datapoint.gold_sent_vecs) loss = self.loss_fn(score, datapoint.top_rouge[k]) else: if len(datapoint.sent_vecs) >= 3: narray = np.random.choice(len(datapoint.sent_vecs), 3, replace = False) narray.sort() sample = datapoint.sent_vecs[narray] else: continue # handle len(sent_vecs) < 3 score = self.__call__(sample, datapoint.gold_sent_vecs) loss = self.loss_fn(score, rouge(datapoint.raw_document[narray]), datapoint.raw_summary) # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? epoch_loss += loss.item() return epoch_loss / len(dataloader.dataset) Loading
main_Critic.py 0 → 100644 +62 −0 Original line number Diff line number Diff line # import # hyperparameters epochs=20 batch_size=20 learning_rate=0.001 train_dataloader = torch.utils.data.DataLoader(dataset.train, batch_size=batch_size, shuffle=True, collate_fn = lambda x: x) # load state dict from trainiing model_actor_only = ActorOnlySummarisationModel() model_actor_only.load_state_dict('model_actor_only_wts.pth') model_actor_only.eval() m = Critic(model_actor_only) since = time.time() val_loss_history = [] best_model_wts = copy.deepcopy(m.state_dict()) for e in range(epochs): print('Epoch {}/{}'.format(e, epochs - 1)) print('-' * 10) # train phase epoch_loss = m._epoch(train_dataloader) # collate_fn print('Train Loss: {:.4f}'.format(epoch_loss)) # validation phase val_epoch_loss = m._epoch(dataset.validation) val_loss_history.append(val_epoch_loss) print('Validation Loss: {:.4f}'.format(val_epoch_loss)) # epoch completed, deep copy the best model sofar if val_epoch_loss < best_loss: best_loss = val_epoch_loss best_model_wts = copy.deepcopy(m.state_dict()) # after training completed time_elapsed = time.time() - since print('Training complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60)) print('Best val loss: {:4f}'.format(best_loss)) # write val_rouge_history in file # load best model weights m.load_state_dict(best_model_wts) # testing since = time.time() test_loss = m.test(dataset.test) print('Test Loss: {:.4f}'.format(test_loss)) # after testing completed time_elapsed = time.time() - since print('Testing complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
models.py +112 −30 Original line number Diff line number Diff line Loading @@ -221,18 +221,64 @@ class Critic(nn.Module): nn.functional.relu(self.layer_2( utils.gaussian(self.layer_1(double_document))))))) def _train(self, dataset, epochs=200, batch_size=20, learning_rate=0.001, shuffle=True, pos_samples=0.5): # move to main # def _train(self, dataset, epochs=200, batch_size=20, learning_rate=0.001, shuffle=True, pos_samples=0.5): # move to main optimizer = torch.optim.Adam(self.parameters(), lr=learning_rate) loss_fn = nn.MSELoss() # optimizer = torch.optim.Adam(self.parameters(), lr=learning_rate) # loss_fn = nn.MSELoss() for _ in range(epochs): # for _ in range(epochs): training_dataloader = torch.utils.data.DataLoader(dataset, batch_size=batch_size, shuffle=shuffle) # training_dataloader = torch.utils.data.DataLoader(dataset, batch_size=batch_size, shuffle=shuffle) for batch in training_dataloader: # for batch in training_dataloader: optimizer.zero_grad() # optimizer.zero_grad() # for datapoint in batch: # r = np.random.random() # if r > pos_samples: # k = np.random.choice(len(datapoint.p_searchspace)) # sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs # score = self.__call__(sample, datapoint.gold_sent_vecs) # loss = loss_fn(score, datapoint.top_rouge[k]) # else: # if len(datapoint.sent_vecs) >= 3: # narray = np.random.choice(len(datapoint.sent_vecs), 3, replace = False) # narray.sort() # sample = datapoint.sent_vecs[narray] # else: # continue # handle len(sent_vecs) < 3 # score = self.__call__(sample, datapoint.gold_sent_vecs) # loss = loss_fn(score, utils.rouge(raw_document[narray]), raw_summary)) # # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? # loss.backward() # optimizer.step() # # eval # # test with rouge def training_epoch(self, dataloader, learning_rate=None): if learning_rate != None: for g in self.optimizer.param_groups: g['lr'] = learning_rate self.train() pos_samples= 0.5 epoch_loss = 0.0 for batch in train_dataloader: self.optimizer.zero_grad() for datapoint in batch: Loading @@ -243,7 +289,7 @@ class Critic(nn.Module): sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs score = self.__call__(sample, datapoint.gold_sent_vecs) loss = loss_fn(score, datapoint.top_rouge[k]) loss = self.loss_fn(score, datapoint.top_rouge[k]) else: if len(datapoint.sent_vecs) >= 3: Loading @@ -254,15 +300,51 @@ class Critic(nn.Module): continue # handle len(sent_vecs) < 3 score = self.__call__(sample, datapoint.gold_sent_vecs) loss = loss_fn(score, utils.rouge(raw_document[narray]), raw_summary)) loss = self.loss_fn(score, rouge(datapoint.raw_document[narray]), datapoint.raw_summary) # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? epoch_loss += loss.item() loss.backward() optimizer.step() self.optimizer.step() return epoch_loss / len(dataloader.dataset) def test(self, dataloader): self.eval() pos_samples= 0.5 epoch_loss = 0.0 with torch.no_grad(): for batch in dataloader: # eval for datapoint in batch: # test with rouge r = np.random.random() if r > pos_samples: k = np.random.choice(len(datapoint.p_searchspace)) sample = datapoint.sent_vecs.masked_select(datapoint.p_searchspace[k].bool()) # not padded sent embeddngs score = self.__call__(sample, datapoint.gold_sent_vecs) loss = self.loss_fn(score, datapoint.top_rouge[k]) else: if len(datapoint.sent_vecs) >= 3: narray = np.random.choice(len(datapoint.sent_vecs), 3, replace = False) narray.sort() sample = datapoint.sent_vecs[narray] else: continue # handle len(sent_vecs) < 3 score = self.__call__(sample, datapoint.gold_sent_vecs) loss = self.loss_fn(score, rouge(datapoint.raw_document[narray]), datapoint.raw_summary) # rouge score berechnen für negative sample => besser wäre externes berechnen und speichern? epoch_loss += loss.item() return epoch_loss / len(dataloader.dataset)
