DLLL {"timestamp": "1689557913.603774", "datetime": "2023-07-17 09:38:33.603774", "elapsedtime": "0.006562", "type": "LOG", "step": "PARAMETER", "data": {"Config": ["Namespace(allreduce_post_accumulation=False, allreduce_post_accumulation_fp16=False, amp=True, bert_model='bert-large-uncased', checkpoint_activations=False, config_file='./bert_config.json', disable_progress_bar=False, dist_url='tcp://224.66.41.62:23456', do_train=True, fp16=True, gpus_per_node=1, gradient_accumulation_steps=1, init_checkpoint=None, init_loss_scale=1048576, input_dir='/public/software/apps/DeepLearning/Data/wikicorpus_en/lower_case_1_seq_len_128_max_pred_20_masked_lm_prob_0.15_random_seed_12345_dupe_factor_5_shard_1472_test_split_10/wikicorpus_en/training', json_summary='./output/dllogger.json', learning_rate=0.0004, local_rank=-1, log_freq=1.0, loss_scale=0.0, max_predictions_per_seq=20, max_seq_length=128, max_steps=100000.0, n_gpu=1, num_steps_per_checkpoint=20, num_train_epochs=3.0, output_dir='/public/home/hepj/outdir/torch/pre_wiki/phrase1', phase1_end_step=7038, phase2=False, resume_from_checkpoint=False, resume_step=-1, seed=12439, skip_checkpoint=False, steps_this_run=100000.0, train_batch_size=16, use_env=False, warmup_proportion=0.0, world_size=1)"]}}
DLLL {"timestamp": "1689557920.976409", "datetime": "2023-07-17 09:38:40.976409", "elapsedtime": "7.379197", "type": "LOG", "step": "PARAMETER", "data": {"SEED": 12439}}
DLLL {"timestamp": "1689557920.97655", "datetime": "2023-07-17 09:38:40.976550", "elapsedtime": "7.379338", "type": "LOG", "step": "PARAMETER", "data": {"train_start": true}}
DLLL {"timestamp": "1689557920.976623", "datetime": "2023-07-17 09:38:40.976623", "elapsedtime": "7.379411", "type": "LOG", "step": "PARAMETER", "data": {"batch_size_per_gpu": 16}}
DLLL {"timestamp": "1689557920.976687", "datetime": "2023-07-17 09:38:40.976687", "elapsedtime": "7.379475", "type": "LOG", "step": "PARAMETER", "data": {"learning_rate": 0.0004}}
DLLL {"timestamp": "1689557925.873029", "datetime": "2023-07-17 09:38:45.873029", "elapsedtime": "12.275817", "type": "LOG", "step": [0, 1], "data": {"average_loss": 11.34375, "step_loss": 11.34375, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557926.215128", "datetime": "2023-07-17 09:38:46.215128", "elapsedtime": "12.617916", "type": "LOG", "step": [0, 2], "data": {"average_loss": 11.2890625, "step_loss": 11.2890625, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557926.557062", "datetime": "2023-07-17 09:38:46.557062", "elapsedtime": "12.95985", "type": "LOG", "step": [0, 3], "data": {"average_loss": 11.3359375, "step_loss": 11.3359375, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557926.898986", "datetime": "2023-07-17 09:38:46.898986", "elapsedtime": "13.301774", "type": "LOG", "step": [0, 4], "data": {"average_loss": 11.390625, "step_loss": 11.390625, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557927.240626", "datetime": "2023-07-17 09:38:47.240626", "elapsedtime": "13.643414", "type": "LOG", "step": [0, 5], "data": {"average_loss": 11.328125, "step_loss": 11.328125, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557927.712912", "datetime": "2023-07-17 09:38:47.712912", "elapsedtime": "14.1157", "type": "LOG", "step": [0, 6], "data": {"average_loss": 11.3359375, "step_loss": 11.3359375, "learning_rate": 0.000399997999995}}
DLLL {"timestamp": "1689557928.103626", "datetime": "2023-07-17 09:38:48.103626", "elapsedtime": "14.506414", "type": "LOG", "step": [0, 7], "data": {"average_loss": 10.859375, "step_loss": 10.859375, "learning_rate": 0.0003999959999799998}}
DLLL {"timestamp": "1689557928.489023", "datetime": "2023-07-17 09:38:48.489023", "elapsedtime": "14.891811", "type": "LOG", "step": [0, 8], "data": {"average_loss": 10.625, "step_loss": 10.625, "learning_rate": 0.00039999399995499935}}
DLLL {"timestamp": "1689557928.873728", "datetime": "2023-07-17 09:38:48.873728", "elapsedtime": "15.276516", "type": "LOG", "step": [0, 9], "data": {"average_loss": 10.4140625, "step_loss": 10.4140625, "learning_rate": 0.0003999919999199984}}
DLLL {"timestamp": "1689557929.258729", "datetime": "2023-07-17 09:38:49.258729", "elapsedtime": "15.661517", "type": "LOG", "step": [0, 10], "data": {"average_loss": 10.3125, "step_loss": 10.3125, "learning_rate": 0.0003999899998749969}}
DLLL {"timestamp": "1689557929.645026", "datetime": "2023-07-17 09:38:49.645026", "elapsedtime": "16.047814", "type": "LOG", "step": [0, 11], "data": {"average_loss": 10.3515625, "step_loss": 10.3515625, "learning_rate": 0.0003999879998199946}}
DLLL {"timestamp": "1689557930.031004", "datetime": "2023-07-17 09:38:50.031004", "elapsedtime": "16.433792", "type": "LOG", "step": [0, 12], "data": {"average_loss": 9.9453125, "step_loss": 9.9453125, "learning_rate": 0.0003999859997549914}}
DLLL {"timestamp": "1689557930.417146", "datetime": "2023-07-17 09:38:50.417146", "elapsedtime": "16.819934", "type": "LOG", "step": [0, 13], "data": {"average_loss": 10.0, "step_loss": 10.0, "learning_rate": 0.00039998399967998725}}
DLLL {"timestamp": "1689557930.80166", "datetime": "2023-07-17 09:38:50.801660", "elapsedtime": "17.204448", "type": "LOG", "step": [0, 14], "data": {"average_loss": 9.96875, "step_loss": 9.96875, "learning_rate": 0.0003999819995949818}}
DLLL {"timestamp": "1689557931.18733", "datetime": "2023-07-17 09:38:51.187330", "elapsedtime": "17.590118", "type": "LOG", "step": [0, 15], "data": {"average_loss": 9.859375, "step_loss": 9.859375, "learning_rate": 0.00039997999949997504}}
DLLL {"timestamp": "1689557931.573538", "datetime": "2023-07-17 09:38:51.573538", "elapsedtime": "17.976326", "type": "LOG", "step": [0, 16], "data": {"average_loss": 9.8515625, "step_loss": 9.8515625, "learning_rate": 0.0003999779993949667}}