| import torch |
| import torch.nn.functional as F |
| import numpy as np |
| import os |
| import copy |
| import logging |
| from torch import nn |
| from sklearn.metrics import confusion_matrix, f1_score, accuracy_score |
| from tqdm import trange, tqdm |
| from losses import loss_map |
| from utils.functions import save_model, restore_model, centroids_cal |
|
|
|
|
| class PretrainManager: |
| |
| def __init__(self, args, data, model, logger_name = 'Detection'): |
|
|
| self.logger = logging.getLogger(logger_name) |
|
|
| self.set_model_optimizer(args, data, model) |
| |
| self.train_dataloader = data.dataloader.train_labeled_loader |
| self.eval_dataloader = data.dataloader.eval_loader |
| self.test_dataloader = data.dataloader.test_loader |
|
|
| self.loss_fct = loss_map[args.loss_fct] |
| self.centroids = None |
| self.best_eval_score = None |
|
|
| if args.pretrain or (not os.path.exists(args.model_output_dir)): |
| self.logger.info('Pre-training Begin...') |
|
|
| if args.backbone == 'bert_disaware' or args.backbone == 'llama_disaware': |
| print("PretrainManager_disaware---------------") |
| self.train_disaware(args, data) |
| else: |
| self.train_plain(args, data) |
|
|
| self.logger.info('Pre-training finished...') |
| |
| else: |
| self.model = restore_model(self.model, args.model_output_dir) |
|
|
| def set_model_optimizer(self, args, data, model): |
| |
| self.model = model.set_model(args, 'llama') |
| self.optimizer, self.scheduler = model.set_optimizer(self.model, data.dataloader.num_train_examples, args.train_batch_size, \ |
| args.num_train_epochs, args.lr, args.warmup_proportion) |
| self.device = args.device |
|
|
| def train_plain(self, args, data): |
|
|
| wait = 0 |
| best_model = None |
| best_eval_score = 0 |
| |
| for epoch in trange(int(args.num_train_epochs), desc="Epoch"): |
| self.model.train() |
| torch.cuda.empty_cache() |
| tr_loss = 0 |
| nb_tr_examples, nb_tr_steps = 0, 0 |
| |
| for step, batch in enumerate(tqdm(self.train_dataloader, desc="Iteration")): |
| batch = tuple(t.to(self.device) for t in batch) |
| input_ids, input_mask, segment_ids, label_ids = batch |
| with torch.set_grad_enabled(True): |
| loss = self.model(input_ids, segment_ids, input_mask, labels = label_ids, mode = "train", loss_fct = self.loss_fct) |
| self.optimizer.zero_grad() |
|
|
| loss.backward() |
|
|
| self.optimizer.step() |
| self.scheduler.step() |
|
|
| tr_loss += loss.item() |
| nb_tr_examples += input_ids.size(0) |
| nb_tr_steps += 1 |
| |
| loss = tr_loss / nb_tr_steps |
| |
| y_true, y_pred = self.get_outputs(args, data, mode = 'eval') |
| eval_score = round(accuracy_score(y_true, y_pred) * 100, 2) |
|
|
| eval_results = { |
| 'train_loss': loss, |
| 'eval_score': eval_score, |
| 'best_eval_score':best_eval_score, |
| } |
| self.logger.info("***** Epoch: %s: Eval results *****", str(epoch + 1)) |
| for key in sorted(eval_results.keys()): |
| self.logger.info(" %s = %s", key, str(eval_results[key])) |
| |
| if eval_score > best_eval_score: |
| |
| best_model = copy.deepcopy(self.model) |
| wait = 0 |
| best_eval_score = eval_score |
|
|
| elif eval_score > 0: |
|
|
| wait += 1 |
| if wait >= args.wait_patient: |
| break |
|
|
| self.model = best_model |
|
|
| if args.save_model: |
| self.logger.info('Trained models are saved in %s', args.model_output_dir) |
| save_model(self.model, args.model_output_dir) |
|
|
| def train_disaware(self, args, data): |
|
|
| wait = 0 |
| best_model = None |
| best_centroids = None |
| best_eval_score = 0 |
| args.device = self.device |
| |
| for epoch in trange(int(args.num_train_epochs), desc="Epoch"): |
| self.centroids = centroids_cal(self.model, args, data, self.train_dataloader, self.device) |
| self.model.train() |
| tr_loss = 0 |
| nb_tr_examples, nb_tr_steps = 0, 0 |
|
|
| for step, batch in enumerate(tqdm(self.train_dataloader, desc="Iteration")): |
| batch = tuple(t.to(self.device) for t in batch) |
| input_ids, input_mask, segment_ids, label_ids = batch |
|
|
| with torch.set_grad_enabled(True): |
| |
| loss = self.model(input_ids, segment_ids, input_mask, label_ids, mode = "train", loss_fct = self.loss_fct, centroids = self.centroids) |
|
|
| self.optimizer.zero_grad() |
|
|
| loss.backward() |
| self.optimizer.step() |
| self.scheduler.step() |
| |
| tr_loss += loss.item() |
| nb_tr_examples += input_ids.size(0) |
| nb_tr_steps += 1 |
| |
| loss = tr_loss / nb_tr_steps |
|
|
| y_true, y_pred = self.get_outputs(args, data, mode = 'eval') |
| eval_score = round(f1_score(y_true, y_pred, average = 'macro') * 100, 2) |
|
|
| eval_results = { |
| 'train_loss': loss, |
| 'eval_score': eval_score, |
| 'best_eval_score':best_eval_score, |
| } |
| self.logger.info("***** Epoch: %s: Eval results *****", str(epoch + 1)) |
| for key in sorted(eval_results.keys()): |
| self.logger.info(" %s = %s", key, str(eval_results[key])) |
| |
| if eval_score > best_eval_score: |
| |
| best_model = copy.deepcopy(self.model) |
| best_centroids = copy.copy(self.centroids) |
| wait = 0 |
| best_eval_score = eval_score |
|
|
| elif eval_score > 0: |
|
|
| wait += 1 |
| if wait >= args.wait_patient: |
| break |
|
|
| self.model = best_model |
| self.centroids = best_centroids |
| self.best_eval_score = best_eval_score |
|
|
| if args.save_model: |
| self.logger.info('Trained models are saved in %s', args.model_output_dir) |
| save_model(self.model, args.model_output_dir) |
| |
|
|
| def get_outputs(self, args, data, mode = 'eval', get_feats = False): |
| |
| if mode == 'eval': |
| dataloader = self.eval_dataloader |
| elif mode == 'test': |
| dataloader = self.test_dataloader |
|
|
| self.model.eval() |
|
|
| total_labels = torch.empty(0,dtype=torch.long).to(self.device) |
| total_preds = torch.empty(0,dtype=torch.long).to(self.device) |
| |
| total_features = torch.empty((0,args.feat_dim)).to(self.device) |
| total_logits = torch.empty((0, data.num_labels)).to(self.device) |
| |
| for batch in tqdm(dataloader, desc="Iteration"): |
|
|
| batch = tuple(t.to(self.device) for t in batch) |
| |
| input_ids, input_mask, segment_ids, label_ids = batch |
| with torch.set_grad_enabled(False): |
|
|
| if args.backbone == 'bert_disaware' or args.backbone =='llama_disaware': |
| pooled_output, logits = self.model(input_ids, segment_ids, input_mask, centroids = self.centroids, labels = label_ids, mode = mode) |
| else: |
| pooled_output, logits = self.model(input_ids, segment_ids, input_mask, mode = mode) |
|
|
| total_labels = torch.cat((total_labels,label_ids)) |
| total_features = torch.cat((total_features, pooled_output)) |
| total_logits = torch.cat((total_logits, logits)) |
|
|
| if get_feats: |
| feats = total_features.cpu().numpy() |
| return feats |
|
|
| else: |
|
|
| total_probs = F.softmax(total_logits.detach(), dim=1) |
| total_maxprobs, total_preds = total_probs.max(dim = 1) |
|
|
| y_pred = total_preds.cpu().numpy() |
| y_true = total_labels.cpu().numpy() |
|
|
| return y_true, y_pred |
|
|
