Finetune_LMHL5_8_8_RTE.log

01/08/2021 12:06:47 1
01/08/2021 12:06:47 Launching the MT-DNN training
01/08/2021 12:06:47 Loading data/canonical_data/bert_base_uncased_lower/rte_train.json as task 1
01/08/2021 12:06:47 ####################
01/08/2021 12:06:47 {'log_file': 'checkpoints/finetune-rte-LM_loss8/log.log', 'tensorboard': False, 'tensorboard_logdir': 'tensorboard_logdir', 'init_checkpoint': 'checkpoints/mtdnn-LM_loss8/model_4.pt', 'data_dir': 'data/canonical_data/bert_base_uncased_lower', 'data_sort_on': False, 'name': 'farmer', 'task_def': 'experiments/glue/glue_task_def.yml', 'train_datasets': ['rte'], 'test_datasets': ['rte'], 'glue_format_on': False, 'mkd_opt': 0, 'do_padding': False, 'update_bert_opt': 0, 'multi_gpu_on': True, 'mem_cum_type': 'simple', 'answer_num_turn': 5, 'answer_mem_drop_p': 0.1, 'answer_att_hidden_size': 128, 'answer_att_type': 'bilinear', 'answer_rnn_type': 'gru', 'answer_sum_att_type': 'bilinear', 'answer_merge_opt': 1, 'answer_mem_type': 1, 'max_answer_len': 10, 'answer_dropout_p': 0.1, 'answer_weight_norm_on': False, 'dump_state_on': False, 'answer_opt': 1, 'mtl_opt': 0, 'ratio': 0, 'mix_opt': 0, 'max_seq_len': 512, 'init_ratio': 1, 'encoder_type': <EncoderModelType.BERT: 1>, 'num_hidden_layers': -1, 'bert_model_type': 'bert-base-uncased', 'do_lower_case': False, 'masked_lm_prob': 0.15, 'short_seq_prob': 0.2, 'max_predictions_per_seq': 128, 'bin_on': False, 'bin_size': 64, 'bin_grow_ratio': 0.5, 'cuda': True, 'log_per_updates': 500, 'save_per_updates': 10000, 'save_per_updates_on': False, 'epochs': 20, 'batch_size': 8, 'batch_size_eval': 8, 'optimizer': 'adamax', 'grad_clipping': 0.0, 'global_grad_clipping': 1.0, 'weight_decay': 0, 'learning_rate': 5e-05, 'momentum': 0, 'warmup': 0.1, 'warmup_schedule': 'warmup_linear', 'adam_eps': 1e-06, 'vb_dropout': True, 'dropout_p': 0.1, 'dropout_w': 0.0, 'bert_dropout_p': 0.1, 'model_ckpt': 'checkpoints/model_0.pt', 'resume': False, 'have_lr_scheduler': True, 'multi_step_lr': '10,20,30', 'lr_gamma': 0.5, 'scheduler_type': 'ms', 'output_dir': 'checkpoints/finetune-rte-LM_loss8', 'seed': 2018, 'grad_accumulation_step': 1, 'fp16': False, 'fp16_opt_level': 'O1', 'adv_train': False, 'adv_opt': 0, 'adv_norm_level': 0, 'adv_p_norm': 'inf', 'adv_alpha': 1, 'adv_k': 1, 'adv_step_size': 0.001, 'adv_noise_var': 1e-05, 'adv_epsilon': 1e-06, 'loss_pred': True, 'collect_uncertainty': None, 'collect_topk': 0.1, 'load_ranked_data': None, 'mc_dropout': 0, 'finetune': True, 'encode_mode': False, 'task_def_list': [{'self': '{}', 'label_vocab': '<data_utils.vocab.Vocabulary object at 0x7f5e4a9598b0>', 'n_class': '2', 'data_type': '<DataFormat.PremiseAndOneHypothesis: 2>', 'task_type': '<TaskType.Classification: 1>', 'metric_meta': '(<Metric.ACC: 0>,)', 'split_names': "['train', 'dev', 'test']", 'enable_san': 'False', 'dropout_p': 'None', 'loss': '<LossCriterion.CeCriterion: 0>', 'kd_loss': '<LossCriterion.MseCriterion: 1>', 'adv_loss': '<LossCriterion.SymKlCriterion: 7>', '__class__': "<class 'experiments.exp_def.TaskDef'>"}]}
01/08/2021 12:06:47 ####################
01/08/2021 12:06:47 ############# Gradient Accumulation Info #############
01/08/2021 12:06:47 number of step: 6240
01/08/2021 12:06:47 number of grad grad_accumulation step: 1
01/08/2021 12:06:47 adjusted number of step: 6240
01/08/2021 12:06:47 ############# Gradient Accumulation Info #############
01/08/2021 12:06:57 
############# Model Arch of MT-DNN #############
SANBertNetwork(
  (dropout_list): ModuleList(
    (0): DropoutWrapper()
    (1): DropoutWrapper()
    (2): DropoutWrapper()
    (3): DropoutWrapper()
    (4): DropoutWrapper()
    (5): DropoutWrapper()
    (6): DropoutWrapper()
    (7): DropoutWrapper()
  )
  (bert): BertModel(
    (embeddings): BertEmbeddings(
      (word_embeddings): Embedding(30522, 768, padding_idx=0)
      (position_embeddings): Embedding(512, 768)
      (token_type_embeddings): Embedding(2, 768)
      (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
      (dropout): Dropout(p=0.1, inplace=False)
    )
    (encoder): BertEncoder(
      (layer): ModuleList(
        (0): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (1): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (2): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (3): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (4): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (5): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (6): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (7): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (8): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (9): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (10): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
        (11): BertLayer(
          (attention): BertAttention(
            (self): BertSelfAttention(
              (query): Linear(in_features=768, out_features=768, bias=True)
              (key): Linear(in_features=768, out_features=768, bias=True)
              (value): Linear(in_features=768, out_features=768, bias=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (output): BertSelfOutput(
              (dense): Linear(in_features=768, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
          (intermediate): BertIntermediate(
            (dense): Linear(in_features=768, out_features=3072, bias=True)
          )
          (output): BertOutput(
            (dense): Linear(in_features=3072, out_features=768, bias=True)
            (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
            (dropout): Dropout(p=0.1, inplace=False)
          )
        )
      )
    )
    (pooler): BertPooler(
      (dense): Linear(in_features=768, out_features=768, bias=True)
      (activation): Tanh()
    )
  )
  (loss_pred_fc): Linear(in_features=768, out_features=1, bias=True)
  (scoring_list): ModuleList(
    (0): Linear(in_features=768, out_features=3, bias=True)
    (1): Linear(in_features=768, out_features=2, bias=True)
    (2): Linear(in_features=768, out_features=2, bias=True)
    (3): Linear(in_features=768, out_features=2, bias=True)
    (4): Linear(in_features=768, out_features=2, bias=True)
    (5): Linear(in_features=768, out_features=2, bias=True)
    (6): Linear(in_features=768, out_features=2, bias=True)
    (7): Linear(in_features=768, out_features=1, bias=True)
  )
)

01/08/2021 12:06:57 Total number of params: 109495313
01/08/2021 12:06:57 At epoch 0
01/08/2021 12:06:57 Task [ 1] updates[     1] train loss[0.82788] remaining[0:02:36]
01/08/2021 12:07:58 Task rte -- epoch 0 -- Dev ACC: 69.675
01/08/2021 12:08:09 [new test scores saved.]
01/08/2021 12:08:16 At epoch 1
01/08/2021 12:08:51 Task [ 1] updates[   500] train loss[0.59723] remaining[0:00:23]
01/08/2021 12:09:17 Task rte -- epoch 1 -- Dev ACC: 70.758
01/08/2021 12:09:27 [new test scores saved.]
01/08/2021 12:09:32 At epoch 2
01/08/2021 12:10:33 Task rte -- epoch 2 -- Dev ACC: 68.953
01/08/2021 12:10:44 [new test scores saved.]
01/08/2021 12:10:48 At epoch 3
01/08/2021 12:11:00 Task [ 1] updates[  1000] train loss[0.51846] remaining[0:00:47]
01/08/2021 12:11:47 Task rte -- epoch 3 -- Dev ACC: 68.953
01/08/2021 12:11:58 [new test scores saved.]
01/08/2021 12:12:02 At epoch 4
01/08/2021 12:12:49 Task [ 1] updates[  1500] train loss[0.43312] remaining[0:00:11]
01/08/2021 12:13:02 Task rte -- epoch 4 -- Dev ACC: 67.870
01/08/2021 12:13:13 [new test scores saved.]
01/08/2021 12:13:18 At epoch 5
01/08/2021 12:14:21 Task rte -- epoch 5 -- Dev ACC: 72.202
01/08/2021 12:14:32 [new test scores saved.]
01/08/2021 12:14:37 At epoch 6
01/08/2021 12:15:01 Task [ 1] updates[  2000] train loss[0.35895] remaining[0:00:34]
01/08/2021 12:15:37 Task rte -- epoch 6 -- Dev ACC: 72.563
01/08/2021 12:15:48 [new test scores saved.]
01/08/2021 12:15:53 At epoch 7
01/08/2021 12:16:54 Task rte -- epoch 7 -- Dev ACC: 72.924
01/08/2021 12:17:05 [new test scores saved.]
01/08/2021 12:17:10 At epoch 8
01/08/2021 12:17:11 Task [ 1] updates[  2500] train loss[0.30127] remaining[0:01:07]
01/08/2021 12:18:11 Task rte -- epoch 8 -- Dev ACC: 72.563
01/08/2021 12:18:22 [new test scores saved.]
01/08/2021 12:18:27 At epoch 9
01/08/2021 12:19:04 Task [ 1] updates[  3000] train loss[0.25856] remaining[0:00:23]
01/08/2021 12:19:28 Task rte -- epoch 9 -- Dev ACC: 72.202
01/08/2021 12:19:39 [new test scores saved.]
01/08/2021 12:19:43 At epoch 10
01/08/2021 12:20:42 Task rte -- epoch 10 -- Dev ACC: 73.646
01/08/2021 12:20:53 [new test scores saved.]
01/08/2021 12:20:57 At epoch 11
01/08/2021 12:21:10 Task [ 1] updates[  3500] train loss[0.22630] remaining[0:00:45]
01/08/2021 12:21:56 Task rte -- epoch 11 -- Dev ACC: 76.173
01/08/2021 12:22:07 [new test scores saved.]
01/08/2021 12:22:11 At epoch 12
01/08/2021 12:22:58 Task [ 1] updates[  4000] train loss[0.20062] remaining[0:00:10]
01/08/2021 12:23:10 Task rte -- epoch 12 -- Dev ACC: 74.729
01/08/2021 12:23:20 [new test scores saved.]
01/08/2021 12:23:25 At epoch 13
01/08/2021 12:24:24 Task rte -- epoch 13 -- Dev ACC: 74.007
01/08/2021 12:24:35 [new test scores saved.]
01/08/2021 12:24:40 At epoch 14
01/08/2021 12:25:04 Task [ 1] updates[  4500] train loss[0.17951] remaining[0:00:33]
01/08/2021 12:25:38 Task rte -- epoch 14 -- Dev ACC: 74.368
01/08/2021 12:25:49 [new test scores saved.]
01/08/2021 12:25:54 At epoch 15
01/08/2021 12:26:52 Task rte -- epoch 15 -- Dev ACC: 74.007
01/08/2021 12:27:03 [new test scores saved.]
01/08/2021 12:27:08 At epoch 16
01/08/2021 12:27:09 Task [ 1] updates[  5000] train loss[0.16215] remaining[0:01:01]
01/08/2021 12:28:06 Task rte -- epoch 16 -- Dev ACC: 74.007
01/08/2021 12:28:17 [new test scores saved.]
01/08/2021 12:28:22 At epoch 17
01/08/2021 12:28:57 Task [ 1] updates[  5500] train loss[0.14781] remaining[0:00:21]
01/08/2021 12:29:21 Task rte -- epoch 17 -- Dev ACC: 74.729
01/08/2021 12:29:32 [new test scores saved.]
01/08/2021 12:29:37 At epoch 18
01/08/2021 12:30:36 Task rte -- epoch 18 -- Dev ACC: 75.090
01/08/2021 12:30:48 [new test scores saved.]
01/08/2021 12:30:53 At epoch 19
01/08/2021 12:31:06 Task [ 1] updates[  6000] train loss[0.13606] remaining[0:00:45]
01/08/2021 12:31:51 Task rte -- epoch 19 -- Dev ACC: 75.090
01/08/2021 12:32:02 [new test scores saved.]