loaded library: /usr/lib/x86_64-linux-gnu/libibverbs.so.1 loaded library: /usr/lib/x86_64-linux-gnu/libibverbs.so.1 ------------------------ arguments ------------------------ batch_size ...................................... 65536 batch_size_per_proc ............................. 65536 data_dir ........................................ /dataset/f9f659c5/wdl_ofrecord data_part_name_suffix_length .................... 5 data_part_num ................................... 256 dataset_format .................................. ofrecord ddp ............................................. False deep_dropout_rate ............................... 0.5 deep_embedding_vec_size ......................... 16 deep_vocab_size ................................. 2322444 eval_after_training ............................. False eval_batchs ..................................... 20 eval_interval ................................... 0 execution_mode .................................. graph hidden_size ..................................... 1024 hidden_units_num ................................ 2 learning_rate ................................... 0.001 loss_print_every_n_iter ......................... 100 max_iter ........................................ 1100 model_load_dir .................................. model_save_dir .................................. ./checkpoint num_deep_sparse_fields .......................... 26 num_dense_fields ................................ 13 num_wide_sparse_fields .......................... 2 save_initial_model .............................. False save_model_after_each_eval ...................... False test_name ....................................... noname_test wide_vocab_size ................................. 2322444 -------------------- end of arguments --------------------- [rank:0] iter: 100/1100, loss: 0.4974715113639832, latency(ms): 135.0059207156300545 | 2022-04-28 10:29:00.958 timestamp, name, driver_version, utilization.gpu [%], utilization.memory [%], memory.total [MiB], memory.free [MiB], memory.used [MiB] 2022/04/28 10:29:00.979, Tesla V100-SXM2-32GB, 470.57.02, 55 %, 21 %, 32510 MiB, 29404 MiB, 3106 MiB [rank:0] iter: 200/1100, loss: 0.4676933586597443, latency(ms): 91.8811921030282974 | 2022-04-28 10:29:10.146 [rank:0] iter: 300/1100, loss: 0.4620038866996765, latency(ms): 95.0090845674276352 | 2022-04-28 10:29:19.647 [rank:0] iter: 400/1100, loss: 0.4591771662235260, latency(ms): 93.0943915992975235 | 2022-04-28 10:29:28.957 [rank:0] iter: 500/1100, loss: 0.4563112556934357, latency(ms): 93.8052801787853241 | 2022-04-28 10:29:38.337 [rank:0] iter: 600/1100, loss: 0.4524161815643311, latency(ms): 92.1023444831371307 | 2022-04-28 10:29:47.548 [rank:0] iter: 700/1100, loss: 0.4466367661952972, latency(ms): 96.0330877080559731 | 2022-04-28 10:29:57.151 [rank:0] iter: 800/1100, loss: 0.4462897479534149, latency(ms): 92.9417379572987556 | 2022-04-28 10:30:06.445 [rank:0] iter: 900/1100, loss: 0.4455785453319550, latency(ms): 93.2319718226790428 | 2022-04-28 10:30:15.768 [rank:0] iter: 1000/1100, loss: 0.4451048374176025, latency(ms): 92.2003921493887901 | 2022-04-28 10:30:24.988 [rank:0] iter: 1100/1100, loss: 0.4448628127574921, latency(ms): 95.6405724957585335 | 2022-04-28 10:30:34.552