experiments/gpt2-meg-ds-3d/meg_ds_3d_gpt2_perf_n16-ds-on.out

*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
*****************************************
using world size: 64, data-parallel-size: 1, tensor-model-parallel size: 4, pipeline-model-parallel size: 16 
using torch.float16 for parameters ...
------------------------ arguments ------------------------
  accumulate_allreduce_grads_in_fp32 .............. False
  adam_beta1 ...................................... 0.9
  adam_beta2 ...................................... 0.999
  adam_eps ........................................ 1e-08
  adlr_autoresume ................................. False
  adlr_autoresume_interval ........................ 1000
  apply_query_key_layer_scaling ................... True
  apply_residual_connection_post_layernorm ........ False
  attention_dropout ............................... 0.1
  attention_softmax_in_fp32 ....................... False
  bert_binary_head ................................ True
  bert_load ....................................... None
  bf16 ............................................ False
  bias_dropout_fusion ............................. True
  bias_gelu_fusion ................................ True
  biencoder_projection_dim ........................ 0
  biencoder_shared_query_context_model ............ False
  block_data_path ................................. None
  checkpoint_activations .......................... True
  checkpoint_in_cpu ............................... False
  checkpoint_num_layers ........................... 1
  clip_grad ....................................... 1.0
  consumed_train_samples .......................... 0
  consumed_valid_samples .......................... 0
  contigious_checkpointing ........................ False
  cpu_optimizer ................................... False
  data_impl ....................................... mmap
  data_parallel_size .............................. 1
  data_path ....................................... ['/gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document']
  dataloader_type ................................. single
  DDP_impl ........................................ local
  decoder_seq_length .............................. None
  deepscale ....................................... False
  deepscale_config ................................ None
  deepspeed ....................................... True
  deepspeed_activation_checkpointing .............. True
  deepspeed_config ................................ ./ds_config.json
  deepspeed_mpi ................................... False
  distribute_checkpointed_activations ............. False
  distributed_backend ............................. nccl
  embedding_path .................................. None
  encoder_seq_length .............................. 1024
  eod_mask_loss ................................... False
  eval_interval ................................... 100
  eval_iters ...................................... 10
  evidence_data_path .............................. None
  exit_duration_in_mins ........................... None
  exit_interval ................................... None
  ffn_hidden_size ................................. 32768
  finetune ........................................ False
  fp16 ............................................ True
  fp16_lm_cross_entropy ........................... False
  fp32_residual_connection ........................ False
  global_batch_size ............................... 1024
  hidden_dropout .................................. 0.1
  hidden_size ..................................... 8192
  hysteresis ...................................... 2
  ict_head_size ................................... None
  ict_load ........................................ None
  img_dim ......................................... 224
  indexer_batch_size .............................. 128
  indexer_log_interval ............................ 1000
  init_method_std ................................. 0.02
  init_method_xavier_uniform ...................... False
  initial_loss_scale .............................. 4294967296
  kv_channels ..................................... 256
  layernorm_epsilon ............................... 1e-05
  lazy_mpu_init ................................... None
  load ............................................ /gpfsscratch/rech/six/ura81os/checkpoints/gpt2-meg-ds
  local_rank ...................................... 0
  log_batch_size_to_tensorboard ................... False
  log_interval .................................... 1
  log_learning_rate_to_tensorboard ................ True
  log_loss_scale_to_tensorboard ................... True
  log_num_zeros_in_grad ........................... False
  log_params_norm ................................. False
  log_timers_to_tensorboard ....................... False
  log_validation_ppl_to_tensorboard ............... False
  loss_scale ...................................... 12.0
  loss_scale_window ............................... 1000
  lr .............................................. 0.00015
  lr_decay_iters .................................. 800
  lr_decay_samples ................................ None
  lr_decay_style .................................. cosine
  lr_warmup_fraction .............................. 0.01
  lr_warmup_iters ................................. 0
  lr_warmup_samples ............................... 0
  make_vocab_size_divisible_by .................... 128
  mask_prob ....................................... 0.15
  masked_softmax_fusion ........................... True
  max_position_embeddings ......................... 1024
  merge_file ...................................... /gpfswork/rech/six/commun/models-custom/megatron-gpt2/megatron_lm_345m_v0.0/release/gpt2-merges.txt
  micro_batch_size ................................ 4
  min_loss_scale .................................. 1.0
  min_lr .......................................... 1e-05
  mmap_warmup ..................................... False
  no_load_optim ................................... None
  no_load_rng ..................................... None
  no_save_optim ................................... None
  no_save_rng ..................................... None
  num_attention_heads ............................. 32
  num_channels .................................... 3
  num_classes ..................................... 1000
  num_layers ...................................... 64
  num_layers_per_virtual_pipeline_stage ........... None
  num_workers ..................................... 2
  onnx_safe ....................................... None
  openai_gelu ..................................... False
  optimizer ....................................... adam
  override_lr_scheduler ........................... False
  params_dtype .................................... torch.float16
  partition_activations ........................... False
  patch_dim ....................................... 16
  pipeline_model_parallel_size .................... 16
  profile_backward ................................ False
  query_in_block_prob ............................. 0.1
  rampup_batch_size ............................... None
  rank ............................................ 0
  remote_device ................................... none
  reset_attention_mask ............................ False
  reset_position_ids .............................. False
  retriever_report_topk_accuracies ................ []
  retriever_score_scaling ......................... False
  retriever_seq_length ............................ 256
  sample_rate ..................................... 1.0
  save ............................................ /gpfsscratch/rech/six/ura81os/checkpoints/gpt2-meg-ds
  save_interval ................................... 500
  scatter_gather_tensors_in_pipeline .............. True
  seed ............................................ 1234
  seq_length ...................................... 1024
  sgd_momentum .................................... 0.9
  short_seq_prob .................................. 0.1
  split ........................................... 949,50,1
  synchronize_each_layer .......................... False
  tensor_model_parallel_size ...................... 4
  tensorboard_dir ................................. None
  tensorboard_log_interval ........................ 1
  tensorboard_queue_size .......................... 1000
  titles_data_path ................................ None
  tokenizer_type .................................. GPT2BPETokenizer
  train_iters ..................................... 1000
  train_samples ................................... None
  use_checkpoint_lr_scheduler ..................... False
  use_contiguous_buffers_in_ddp ................... False
  use_cpu_initialization .......................... None
  use_one_sent_docs ............................... False
  virtual_pipeline_model_parallel_size ............ None
  vocab_extra_ids ................................. 0
  vocab_file ...................................... /gpfswork/rech/six/commun/models-custom/megatron-gpt2/megatron_lm_345m_v0.0/release/gpt2-vocab.json
  weight_decay .................................... 0.01
  world_size ...................................... 64
  zero_stage ...................................... 0
-------------------- end of arguments ---------------------
setting number of micro-batches to constant 256
> building GPT2BPETokenizer tokenizer ...
 > padded vocab (size: 50257) with 431 dummy tokens (new size: 50688)
> initializing torch distributed ...
> initializing tensor model parallel with size 4
> initializing pipeline model parallel with size 16
> setting random seeds to 1234 ...
[2021-06-10 20:47:37,205] [INFO] [checkpointing.py:226:model_parallel_cuda_manual_seed] > initializing model parallel cuda seeds on global rank 0, model parallel rank 0, and data parallel rank 0 with model parallel seed: 3952 and data parallel seed: 1234
> compiling dataset index builder ...
make: Entering directory '/gpfsdswork/projects/rech/six/ura81os/stas/code/megatron-jeffra/megatron/data'
make: Nothing to be done for 'default'.
make: Leaving directory '/gpfsdswork/projects/rech/six/ura81os/stas/code/megatron-jeffra/megatron/data'
>>> done with dataset index builder. Compilation time: 0.106 seconds
> compiling and loading fused kernels ...
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
Detected CUDA files, patching ldflags
Emitting ninja build file /gpfsdswork/projects/rech/six/ura81os/stas/code/megatron-jeffra/megatron/fused_kernels/build/build.ninja...
Building extension module scaled_upper_triang_masked_softmax_cuda...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
ninja: no work to do.
Loading extension module scaled_upper_triang_masked_softmax_cuda...
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
Detected CUDA files, patching ldflags
Emitting ninja build file /gpfsdswork/projects/rech/six/ura81os/stas/code/megatron-jeffra/megatron/fused_kernels/build/build.ninja...
Building extension module scaled_masked_softmax_cuda...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
ninja: no work to do.
Loading extension module scaled_masked_softmax_cuda...
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
Detected CUDA files, patching ldflags
Emitting ninja build file /gpfsdswork/projects/rech/six/ura81os/stas/code/megatron-jeffra/megatron/fused_kernels/build/build.ninja...
Building extension module fused_mix_prec_layer_norm_cuda...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
ninja: no work to do.
Loading extension module fused_mix_prec_layer_norm_cuda...
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/utils/cpp_extension.py:283: UserWarning: 

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler (c++) is not compatible with the compiler Pytorch was
built with for this platform, which is g++ on linux. Please
use g++ to to compile your extension. Alternatively, you may
compile PyTorch from source using c++, and then you can also use
c++ to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!

  warnings.warn(WRONG_COMPILER_WARNING.format(
>>> done with compiling and loading fused kernels. Compilation time: 12.900 seconds
time to initialize megatron (seconds): -41.720
[after megatron is initialized] datetime: 2021-06-10 20:47:50 
building GPT model ...
[2021-06-10 20:47:50,326] [INFO] [utils.py:627:see_memory_usage] Before Building Model
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/cuda/memory.py:373: FutureWarning: torch.cuda.memory_cached has been renamed to torch.cuda.memory_reserved
  warnings.warn(
/gpfswork/rech/six/commun/conda/stas/lib/python3.8/site-packages/torch/cuda/memory.py:381: FutureWarning: torch.cuda.max_memory_cached has been renamed to torch.cuda.max_memory_reserved
  warnings.warn(
[2021-06-10 20:47:50,329] [INFO] [utils.py:628:see_memory_usage] MA 0.0 GB         Max_MA 0.0 GB         CA 0.0 GB         Max_CA 0 GB 
[2021-06-10 20:47:50,329] [INFO] [utils.py:636:see_memory_usage] CPU Virtual Memory:  used = 38.96 GB, percent = 20.8%
SEED_LAYERS=False BASE_SEED=1234 SEED_FN=None
Using topology: {ProcessCoord(pipe=0, data=0, model=0): 0, ProcessCoord(pipe=0, data=0, model=1): 1, ProcessCoord(pipe=0, data=0, model=2): 2, ProcessCoord(pipe=0, data=0, model=3): 3, ProcessCoord(pipe=1, data=0, model=0): 4, ProcessCoord(pipe=1, data=0, model=1): 5, ProcessCoord(pipe=1, data=0, model=2): 6, ProcessCoord(pipe=1, data=0, model=3): 7, ProcessCoord(pipe=2, data=0, model=0): 8, ProcessCoord(pipe=2, data=0, model=1): 9, ProcessCoord(pipe=2, data=0, model=2): 10, ProcessCoord(pipe=2, data=0, model=3): 11, ProcessCoord(pipe=3, data=0, model=0): 12, ProcessCoord(pipe=3, data=0, model=1): 13, ProcessCoord(pipe=3, data=0, model=2): 14, ProcessCoord(pipe=3, data=0, model=3): 15, ProcessCoord(pipe=4, data=0, model=0): 16, ProcessCoord(pipe=4, data=0, model=1): 17, ProcessCoord(pipe=4, data=0, model=2): 18, ProcessCoord(pipe=4, data=0, model=3): 19, ProcessCoord(pipe=5, data=0, model=0): 20, ProcessCoord(pipe=5, data=0, model=1): 21, ProcessCoord(pipe=5, data=0, model=2): 22, ProcessCoord(pipe=5, data=0, model=3): 23, ProcessCoord(pipe=6, data=0, model=0): 24, ProcessCoord(pipe=6, data=0, model=1): 25, ProcessCoord(pipe=6, data=0, model=2): 26, ProcessCoord(pipe=6, data=0, model=3): 27, ProcessCoord(pipe=7, data=0, model=0): 28, ProcessCoord(pipe=7, data=0, model=1): 29, ProcessCoord(pipe=7, data=0, model=2): 30, ProcessCoord(pipe=7, data=0, model=3): 31, ProcessCoord(pipe=8, data=0, model=0): 32, ProcessCoord(pipe=8, data=0, model=1): 33, ProcessCoord(pipe=8, data=0, model=2): 34, ProcessCoord(pipe=8, data=0, model=3): 35, ProcessCoord(pipe=9, data=0, model=0): 36, ProcessCoord(pipe=9, data=0, model=1): 37, ProcessCoord(pipe=9, data=0, model=2): 38, ProcessCoord(pipe=9, data=0, model=3): 39, ProcessCoord(pipe=10, data=0, model=0): 40, ProcessCoord(pipe=10, data=0, model=1): 41, ProcessCoord(pipe=10, data=0, model=2): 42, ProcessCoord(pipe=10, data=0, model=3): 43, ProcessCoord(pipe=11, data=0, model=0): 44, ProcessCoord(pipe=11, data=0, model=1): 45, ProcessCoord(pipe=11, data=0, model=2): 46, ProcessCoord(pipe=11, data=0, model=3): 47, ProcessCoord(pipe=12, data=0, model=0): 48, ProcessCoord(pipe=12, data=0, model=1): 49, ProcessCoord(pipe=12, data=0, model=2): 50, ProcessCoord(pipe=12, data=0, model=3): 51, ProcessCoord(pipe=13, data=0, model=0): 52, ProcessCoord(pipe=13, data=0, model=1): 53, ProcessCoord(pipe=13, data=0, model=2): 54, ProcessCoord(pipe=13, data=0, model=3): 55, ProcessCoord(pipe=14, data=0, model=0): 56, ProcessCoord(pipe=14, data=0, model=1): 57, ProcessCoord(pipe=14, data=0, model=2): 58, ProcessCoord(pipe=14, data=0, model=3): 59, ProcessCoord(pipe=15, data=0, model=0): 60, ProcessCoord(pipe=15, data=0, model=1): 61, ProcessCoord(pipe=15, data=0, model=2): 62, ProcessCoord(pipe=15, data=0, model=3): 63}
[2021-06-10 20:47:51,179] [INFO] [module.py:360:_partition_layers] Partitioning pipeline stages with method type:transformer
stage=0 layers=7
     0: _to_float16
     1: EmbeddingPipe
     2: <lambda>
     3: ParallelTransformerLayerPipe
     4: ParallelTransformerLayerPipe
     5: ParallelTransformerLayerPipe
     6: ParallelTransformerLayerPipe
stage=1 layers=4
     7: ParallelTransformerLayerPipe
     8: ParallelTransformerLayerPipe
     9: ParallelTransformerLayerPipe
    10: ParallelTransformerLayerPipe
stage=2 layers=4
    11: ParallelTransformerLayerPipe
    12: ParallelTransformerLayerPipe
    13: ParallelTransformerLayerPipe
    14: ParallelTransformerLayerPipe
stage=3 layers=4
    15: ParallelTransformerLayerPipe
    16: ParallelTransformerLayerPipe
    17: ParallelTransformerLayerPipe
    18: ParallelTransformerLayerPipe
stage=4 layers=4
    19: ParallelTransformerLayerPipe
    20: ParallelTransformerLayerPipe
    21: ParallelTransformerLayerPipe
    22: ParallelTransformerLayerPipe
stage=5 layers=4
    23: ParallelTransformerLayerPipe
    24: ParallelTransformerLayerPipe
    25: ParallelTransformerLayerPipe
    26: ParallelTransformerLayerPipe
stage=6 layers=4
    27: ParallelTransformerLayerPipe
    28: ParallelTransformerLayerPipe
    29: ParallelTransformerLayerPipe
    30: ParallelTransformerLayerPipe
stage=7 layers=4
    31: ParallelTransformerLayerPipe
    32: ParallelTransformerLayerPipe
    33: ParallelTransformerLayerPipe
    34: ParallelTransformerLayerPipe
stage=8 layers=4
    35: ParallelTransformerLayerPipe
    36: ParallelTransformerLayerPipe
    37: ParallelTransformerLayerPipe
    38: ParallelTransformerLayerPipe
stage=9 layers=4
    39: ParallelTransformerLayerPipe
    40: ParallelTransformerLayerPipe
    41: ParallelTransformerLayerPipe
    42: ParallelTransformerLayerPipe
stage=10 layers=4
    43: ParallelTransformerLayerPipe
    44: ParallelTransformerLayerPipe
    45: ParallelTransformerLayerPipe
    46: ParallelTransformerLayerPipe
stage=11 layers=4
    47: ParallelTransformerLayerPipe
    48: ParallelTransformerLayerPipe
    49: ParallelTransformerLayerPipe
    50: ParallelTransformerLayerPipe
stage=12 layers=4
    51: ParallelTransformerLayerPipe
    52: ParallelTransformerLayerPipe
    53: ParallelTransformerLayerPipe
    54: ParallelTransformerLayerPipe
stage=13 layers=4
    55: ParallelTransformerLayerPipe
    56: ParallelTransformerLayerPipe
    57: ParallelTransformerLayerPipe
    58: ParallelTransformerLayerPipe
stage=14 layers=4
    59: ParallelTransformerLayerPipe
    60: ParallelTransformerLayerPipe
    61: ParallelTransformerLayerPipe
    62: ParallelTransformerLayerPipe
stage=15 layers=8
    63: ParallelTransformerLayerPipe
    64: ParallelTransformerLayerPipe
    65: ParallelTransformerLayerPipe
    66: ParallelTransformerLayerPipe
    67: <lambda>
    68: MixedFusedLayerNorm
    69: EmbeddingPipe
    70: float16_to_fp32
  loss: CrossEntropy
 > number of parameters on (tensor, pipeline) model parallel rank (2, 12): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 7): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 8): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 8): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 8): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 12): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 12): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 12): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 8): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 7): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 7): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 7): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 9): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 9): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 9): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 1): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 4): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 4): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 4): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 11): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 11): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 11): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 11): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 3): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 3): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 3): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 3): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 9): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 14): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 1): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 1): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 1): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 4): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 6): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 6): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 6): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 14): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 14): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 14): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 6): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 5): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 5): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 5): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 5): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 13): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 13): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 13): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 13): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 2): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 2): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 2): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 2): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (0, 10): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (3, 10): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (1, 10): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 10): 805560320
 > number of parameters on (tensor, pipeline) model parallel rank (2, 15): 917774336
 > number of parameters on (tensor, pipeline) model parallel rank (0, 15): 917774336
 > number of parameters on (tensor, pipeline) model parallel rank (2, 0): 917757952
 > number of parameters on (tensor, pipeline) model parallel rank (3, 15): 917774336 > number of parameters on (tensor, pipeline) model parallel rank (1, 15): 917774336

 > number of parameters on (tensor, pipeline) model parallel rank (3, 0): 917757952
 > number of parameters on (tensor, pipeline) model parallel rank (1, 0): 917757952
[2021-06-10 20:47:51,720] [INFO] [utils.py:627:see_memory_usage] After Building Model
[2021-06-10 20:47:51,721] [INFO] [utils.py:628:see_memory_usage] MA 1.73 GB         Max_MA 1.73 GB         CA 1.75 GB         Max_CA 2 GB 
[2021-06-10 20:47:51,721] [INFO] [utils.py:636:see_memory_usage] CPU Virtual Memory:  used = 39.14 GB, percent = 20.9%
 > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 917757952
> learning rate decay style: cosine
DeepSpeed is enabled.
[2021-06-10 20:47:51,724] [INFO] [logging.py:60:log_dist] [Rank 0] DeepSpeed info: version=0.4.0+407ff0f, git-hash=407ff0f, git-branch=megatron2.4-3d
[2021-06-10 20:47:51,764] [INFO] [engine.py:172:__init__] DeepSpeed Flops Profiler Enabled: False
[2021-06-10 20:47:51,764] [INFO] [engine.py:682:_configure_optimizer] Removing param_group that has no 'params' in the client Optimizer
[2021-06-10 20:47:51,765] [INFO] [engine.py:687:_configure_optimizer] Using client Optimizer as basic optimizer
[2021-06-10 20:47:51,765] [INFO] [engine.py:696:_configure_optimizer] DeepSpeed Basic Optimizer = FusedAdam
[2021-06-10 20:47:51,765] [INFO] [logging.py:60:log_dist] [Rank 0] Creating fp16 unfused optimizer with dynamic loss scale
[2021-06-10 20:47:51,765] [INFO] [unfused_optimizer.py:37:__init__] Fused Lamb Legacy : False 
[2021-06-10 20:47:51,885] [INFO] [logging.py:60:log_dist] [Rank 0] DeepSpeed Final Optimizer = FusedAdam
[2021-06-10 20:47:51,885] [INFO] [engine.py:509:_configure_lr_scheduler] DeepSpeed using client LR scheduler
[2021-06-10 20:47:51,885] [INFO] [logging.py:60:log_dist] [Rank 0] DeepSpeed LR Scheduler = <megatron.learning_rates.AnnealingLR object at 0x1533710dffd0>
[2021-06-10 20:47:51,885] [INFO] [logging.py:60:log_dist] [Rank 0] step=0, skipped=0, lr=[0.0, 0.0], mom=[(0.9, 0.999), (0.9, 0.999)]
[2021-06-10 20:47:51,885] [INFO] [config.py:900:print] DeepSpeedEngine configuration:
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   activation_checkpointing_config  {
    "partition_activations": false, 
    "contiguous_memory_optimization": false, 
    "cpu_checkpointing": false, 
    "number_checkpoints": null, 
    "synchronize_checkpoint_boundary": false, 
    "profile": false
}
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   aio_config ................... {'block_size': 1048576, 'queue_depth': 8, 'thread_count': 1, 'single_submit': False, 'overlap_events': True}
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   allreduce_always_fp32 ........ False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   amp_enabled .................. False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   amp_params ................... False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   checkpoint_tag_validation_enabled  True
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   checkpoint_tag_validation_fail  False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   disable_allgather ............ False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   dump_state ................... False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   dynamic_loss_scale_args ...... {'init_scale': 4096, 'scale_window': 500, 'delayed_shift': 2, 'min_scale': 1}
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   eigenvalue_enabled ........... False
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   eigenvalue_gas_boundary_resolution  1
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   eigenvalue_layer_name ........ bert.encoder.layer
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   eigenvalue_layer_num ......... 0
[2021-06-10 20:47:51,885] [INFO] [config.py:904:print]   eigenvalue_max_iter .......... 100
10 20:47:51,886] [INFO] [config.py:904:print]   optimizer_params ............. None
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   pipeline ..................... {'stages': 'auto', 'partition': 'best', 'seed_layers': False, 'activation_checkpoint_interval': 0}
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   pld_enabled .................. False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   pld_params ................... False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   prescale_gradients ........... True
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_change_rate ......... 0.001
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_groups .............. 1
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_offset .............. 1000
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_period .............. 1000
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_rounding ............ 0
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_start_bits .......... 16
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_target_bits ......... 8
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_training_enabled .... False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_type ................ 0
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   quantize_verbose ............. False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   scheduler_name ............... None
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   scheduler_params ............. None
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   sparse_attention ............. None
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   sparse_gradients_enabled ..... False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   steps_per_print .............. 2000
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   tensorboard_enabled .......... False
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   tensorboard_job_name ......... DeepSpeedJobName
[2021-06-10 20:47:51,886] [INFO] [config.py:904:print]   tensorboard_output_path ...... 
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   train_batch_size ............. 1024
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   train_micro_batch_size_per_gpu  4
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   use_quantizer_kernel ......... False
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   wall_clock_breakdown ......... False
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   world_size ................... 1
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   zero_allow_untested_optimizer  False
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   zero_config .................. {
    "stage": 0, 
    "contiguous_gradients": false, 
    "reduce_scatter": true, 
    "reduce_bucket_size": 5.000000e+08, 
    "allgather_partitions": true, 
    "allgather_bucket_size": 5.000000e+08, 
    "overlap_comm": false, 
    "load_from_fp32_weights": true, 
    "elastic_checkpoint": true, 
    "offload_param": null, 
    "offload_optimizer": null, 
    "sub_group_size": 1.000000e+12, 
    "prefetch_bucket_size": 5.000000e+07, 
    "param_persistence_threshold": 1.000000e+05, 
    "max_live_parameters": 1.000000e+09, 
    "max_reuse_distance": 1.000000e+09, 
    "gather_fp16_weights_on_model_save": false, 
    "ignore_unused_parameters": true, 
    "legacy_stage1": false
}
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   zero_enabled ................. False
[2021-06-10 20:47:51,887] [INFO] [config.py:904:print]   zero_optimization_stage ...... 0
[2021-06-10 20:47:51,887] [INFO] [config.py:906:print]   json = {
    "train_micro_batch_size_per_gpu": 4, 
    "gradient_accumulation_steps": 256, 
    "gradient_clipping": 1.0, 
    "prescale_gradients": true, 
    "zero_optimization": {
        "stage": 0
    }, 
    "fp16": {
        "enabled": true, 
        "loss_scale": 0, 
        "loss_scale_window": 500, 
        "hysteresis": 2, 
        "min_loss_scale": 1, 
        "initial_scale_power": 12
    }, 
    "steps_per_print": 2.000000e+03, 
    "wall_clock_breakdown": false
}
[2021-06-10 20:47:51,888] [INFO] [engine.py:76:__init__] CONFIG: micro_batches=256 micro_batch_size=4
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=0 STAGE=0 LAYERS=7 [0, 7) STAGE_PARAMS=917757952 (917.758M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=1 STAGE=0 LAYERS=7 [0, 7) STAGE_PARAMS=917757952 (917.758M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=2 STAGE=0 LAYERS=7 [0, 7) STAGE_PARAMS=917757952 (917.758M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=3 STAGE=0 LAYERS=7 [0, 7) STAGE_PARAMS=917757952 (917.758M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=32 STAGE=8 LAYERS=4 [35, 39) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=34 STAGE=8 LAYERS=4 [35, 39) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=33 STAGE=8 LAYERS=4 [35, 39) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=35 STAGE=8 LAYERS=4 [35, 39) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=17 STAGE=4 LAYERS=4 [19, 23) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=49 STAGE=12 LAYERS=4 [51, 55) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=51 STAGE=12 LAYERS=4 [51, 55) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=48 STAGE=12 LAYERS=4 [51, 55) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=50 STAGE=12 LAYERS=4 [51, 55) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=16 STAGE=4 LAYERS=4 [19, 23) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=19 STAGE=4 LAYERS=4 [19, 23) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=18 STAGE=4 LAYERS=4 [19, 23) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=26 STAGE=6 LAYERS=4 [27, 31) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=24 STAGE=6 LAYERS=4 [27, 31) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=25 STAGE=6 LAYERS=4 [27, 31) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=27 STAGE=6 LAYERS=4 [27, 31) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=43 STAGE=10 LAYERS=4 [43, 47) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=40 STAGE=10 LAYERS=4 [43, 47) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=41 STAGE=10 LAYERS=4 [43, 47) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=42 STAGE=10 LAYERS=4 [43, 47) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=10 STAGE=2 LAYERS=4 [11, 15) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=8 STAGE=2 LAYERS=4 [11, 15) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=11 STAGE=2 LAYERS=4 [11, 15) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=4 STAGE=1 LAYERS=4 [7, 11) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=6 STAGE=1 LAYERS=4 [7, 11) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=7 STAGE=1 LAYERS=4 [7, 11) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=5 STAGE=1 LAYERS=4 [7, 11) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=58 STAGE=14 LAYERS=4 [59, 63) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=57 STAGE=14 LAYERS=4 [59, 63) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=56 STAGE=14 LAYERS=4 [59, 63) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=59 STAGE=14 LAYERS=4 [59, 63) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=61 STAGE=15 LAYERS=8 [63, 71) STAGE_PARAMS=917774336 (917.774M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=60 STAGE=15 LAYERS=8 [63, 71) STAGE_PARAMS=917774336 (917.774M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=63 STAGE=15 LAYERS=8 [63, 71) STAGE_PARAMS=917774336 (917.774M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=37 STAGE=9 LAYERS=4 [39, 43) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=36 STAGE=9 LAYERS=4 [39, 43) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=39 STAGE=9 LAYERS=4 [39, 43) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=38 STAGE=9 LAYERS=4 [39, 43) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=44 STAGE=11 LAYERS=4 [47, 51) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=47 STAGE=11 LAYERS=4 [47, 51) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=46 STAGE=11 LAYERS=4 [47, 51) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=45 STAGE=11 LAYERS=4 [47, 51) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=28 STAGE=7 LAYERS=4 [31, 35) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=30 STAGE=7 LAYERS=4 [31, 35) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=31 STAGE=7 LAYERS=4 [31, 35) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=29 STAGE=7 LAYERS=4 [31, 35) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=54 STAGE=13 LAYERS=4 [55, 59) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=55 STAGE=13 LAYERS=4 [55, 59) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=52 STAGE=13 LAYERS=4 [55, 59) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=12 STAGE=3 LAYERS=4 [15, 19) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=13 STAGE=3 LAYERS=4 [15, 19) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=14 STAGE=3 LAYERS=4 [15, 19) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=9 STAGE=2 LAYERS=4 [11, 15) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=53 STAGE=13 LAYERS=4 [55, 59) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=15 STAGE=3 LAYERS=4 [15, 19) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=62 STAGE=15 LAYERS=8 [63, 71) STAGE_PARAMS=917774336 (917.774M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=20 STAGE=5 LAYERS=4 [23, 27) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=23 STAGE=5 LAYERS=4 [23, 27) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=22 STAGE=5 LAYERS=4 [23, 27) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
[2021-06-10 20:47:52,226] [INFO] [engine.py:134:__init__] RANK=21 STAGE=5 LAYERS=4 [23, 27) STAGE_PARAMS=805560320 (805.560M) TOTAL_PARAMS=52453507072 (52453.507M) UNIQUE_PARAMS=52004716544 (52004.717M)
WARNING: could not find the metadata file /gpfsscratch/rech/six/ura81os/checkpoints/gpt2-meg-ds/latest_checkpointed_iteration.txt 
    will not load any checkpoints and will start from random
time (ms) | load-checkpoint: 11.96
[after model, optimizer, and learning rate scheduler are built] datetime: 2021-06-10 20:47:53 
> building train, validation, and test datasets ...
 > datasets target sizes (minimum size):
    train:      1024000
    validation: 112640
    test:       10240
> building train, validation, and test datasets for GPT ...
 > building dataset index ...
    reading sizes...
    reading pointers...
    reading document index...
    creating numpy buffer of mmap...
    creating memory view of numpy buffer...
 > finished creating indexed dataset in 0.032667 seconds
    number of documents: 10000
 > dataset split:
    train:
     document indices in [0, 9490) total of 9490 documents
    validation:
     document indices in [9490, 9990) total of 500 documents
    test:
     document indices in [9990, 10000) total of 10 documents
 > loading doc-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_train_indexmap_1024000ns_1024sl_1234s_doc_idx.npy
 > loading sample-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_train_indexmap_1024000ns_1024sl_1234s_sample_idx.npy
 > loading shuffle-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_train_indexmap_1024000ns_1024sl_1234s_shuffle_idx.npy
    loaded indexed file in 0.115 seconds
    total number of samples: 1024856
    total number of epochs: 99
 > loading doc-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_valid_indexmap_112640ns_1024sl_1234s_doc_idx.npy
 > loading sample-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_valid_indexmap_112640ns_1024sl_1234s_sample_idx.npy
 > loading shuffle-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_valid_indexmap_112640ns_1024sl_1234s_shuffle_idx.npy
    loaded indexed file in 0.050 seconds
    total number of samples: 113200
    total number of epochs: 182
 > loading doc-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_test_indexmap_10240ns_1024sl_1234s_doc_idx.npy
 > loading sample-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_test_indexmap_10240ns_1024sl_1234s_sample_idx.npy
 > loading shuffle-idx mapping from /gpfswork/rech/six/commun/datasets-custom/openwebtext-10k/meg-gpt2_text_document_test_indexmap_10240ns_1024sl_1234s_shuffle_idx.npy
    loaded indexed file in 0.023 seconds
    total number of samples: 10255
    total number of epochs: 672
> finished creating GPT datasets ...
[after dataloaders are built] datetime: 2021-06-10 20:47:54 
time (ms) | model-and-optimizer-setup: 2744.35 | train/valid/test-data-iterators-setup: 815.33
done with setup ...
training ...
[before the start of training step] datetime: 2021-06-10 20:47:54 
[2021-06-10 20:47:54,339] [INFO] [checkpointing.py:408:forward] Activation Checkpointing Information
[2021-06-10 20:47:54,339] [INFO] [checkpointing.py:409:forward] ----Partition Activations False, CPU CHECKPOINTING False
[2021-06-10 20:47:54,339] [INFO] [checkpointing.py:412:forward] ----contiguous Memory Checkpointing False with 64 total layers
[2021-06-10 20:47:54,339] [INFO] [checkpointing.py:415:forward] ----Synchronization False
[2021-06-10 20:47:54,339] [INFO] [checkpointing.py:416:forward] ----Profiling time in checkpointing False
[Rank 1] (after 1 iterations) memory (MB) | allocated: 12337.45654296875 | max allocated: 19961.072265625 | reserved: 23288.0 | max reserved: 23288.0
[Rank 61] (after 1 iterations) memory (MB) | allocated: 12923.83251953125 | max allocated: 18175.37841796875 | reserved: 19286.0 | max reserved: 19286.0
[Rank 5] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17461.94775390625 | reserved: 20002.0 | max reserved: 20002.0
[Rank 9] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17189.947265625 | reserved: 19824.0 | max reserved: 19824.0
[Rank 17] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16645.9462890625 | reserved: 19216.0 | max reserved: 19216.0
[Rank 13] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16917.94677734375 | reserved: 19456.0 | max reserved: 19456.0
[Rank 25] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16101.9453125 | reserved: 18640.0 | max reserved: 18640.0
[Rank 29] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15829.94482421875 | reserved: 18384.0 | max reserved: 18384.0
[Rank 21] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16373.94580078125 | reserved: 18882.0 | max reserved: 18882.0
[Rank 33] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15557.9443359375 | reserved: 18654.0 | max reserved: 18654.0
[Rank 62] (after 1 iterations) memory (MB) | allocated: 12923.83251953125 | max allocated: 18175.37841796875 | reserved: 19286.0 | max reserved: 19286.0
[Rank 6] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17461.94775390625 | reserved: 20002.0 | max reserved: 20002.0
[Rank 2] (after 1 iterations) memory (MB) | allocated: 12337.45654296875 | max allocated: 19961.072265625 | reserved: 23204.0 | max reserved: 23204.0
[Rank 10] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17189.947265625 | reserved: 19760.0 | max reserved: 19760.0
[Rank 18] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16645.9462890625 | reserved: 19218.0 | max reserved: 19218.0
[Rank 14] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16917.94677734375 | reserved: 19424.0 | max reserved: 19424.0
[Rank 0] (after 1 iterations) memory (MB) | allocated: 12337.45654296875 | max allocated: 19961.072265625 | reserved: 22892.0 | max reserved: 22892.0
 iteration        1/    1000 | consumed samples:         1024 | elapsed time per iteration (ms): 159778.9 | learning rate: 1.875E-05 | global batch size:  1024 | lm-loss: 1.244238E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
[Rank 22] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16373.94580078125 | reserved: 18882.0 | max reserved: 18882.0
[Rank 26] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16101.9453125 | reserved: 19024.0 | max reserved: 19024.0
[Rank 41] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18064.0 | max reserved: 18064.0
[Rank 4] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17461.94775390625 | reserved: 20018.0 | max reserved: 20018.0
[Rank 8] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17189.947265625 | reserved: 19794.0 | max reserved: 19794.0
[Rank 45] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18302.0 | max reserved: 18302.0
[Rank 30] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15829.94482421875 | reserved: 18384.0 | max reserved: 18384.0
[Rank 16] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16645.9462890625 | reserved: 19200.0 | max reserved: 19200.0
[Rank 34] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15557.9443359375 | reserved: 18622.0 | max reserved: 18622.0
[Rank 12] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16917.94677734375 | reserved: 19504.0 | max reserved: 19504.0
[Rank 53] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 49] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17710.0 | max reserved: 17710.0
[Rank 60] (after 1 iterations) memory (MB) | allocated: 12923.83251953125 | max allocated: 18175.37841796875 | reserved: 19286.0 | max reserved: 19286.0
[Rank 24] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16101.9453125 | reserved: 19040.0 | max reserved: 19040.0
[Rank 37] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18304.0 | max reserved: 18304.0
[Rank 57] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 16926.0 | max reserved: 16926.0
[Rank 28] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15829.94482421875 | reserved: 18368.0 | max reserved: 18368.0
[Rank 32] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15557.9443359375 | reserved: 18606.0 | max reserved: 18606.0
[Rank 3] (after 1 iterations) memory (MB) | allocated: 12337.45654296875 | max allocated: 19961.072265625 | reserved: 23270.0 | max reserved: 23270.0
[Rank 7] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17461.94775390625 | reserved: 20002.0 | max reserved: 20002.0
[Rank 11] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 17189.947265625 | reserved: 19744.0 | max reserved: 19744.0
[Rank 63] (after 1 iterations) memory (MB) | allocated: 12923.83251953125 | max allocated: 18175.37841796875 | reserved: 19286.0 | max reserved: 19286.0
[Rank 20] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16373.94580078125 | reserved: 18962.0 | max reserved: 18962.0
[Rank 15] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16917.94677734375 | reserved: 19536.0 | max reserved: 19536.0
[Rank 40] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18480.0 | max reserved: 18480.0
[Rank 44] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17934.0 | max reserved: 17934.0
[Rank 19] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16645.9462890625 | reserved: 19200.0 | max reserved: 19200.0
[Rank 23] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16373.94580078125 | reserved: 18912.0 | max reserved: 18912.0
time (ms)
[Rank 48] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17710.0 | max reserved: 17710.0
[Rank 31] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15829.94482421875 | reserved: 18384.0 | max reserved: 18384.0
[Rank 27] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 16101.9453125 | reserved: 18640.0 | max reserved: 18640.0
[Rank 42] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18304.0 | max reserved: 18304.0
[Rank 35] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15557.9443359375 | reserved: 18654.0 | max reserved: 18654.0
[Rank 56] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 52] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 36] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18494.0 | max reserved: 18494.0
[Rank 39] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18304.0 | max reserved: 18304.0[Rank 38] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18304.0 | max reserved: 18304.0

[Rank 54] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 46] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18222.0 | max reserved: 18222.0
[Rank 43] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18224.0 | max reserved: 18224.0
[Rank 50] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17710.0 | max reserved: 17710.0
[Rank 47] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 18302.0 | max reserved: 18302.0
[Rank 58] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 55] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
[Rank 51] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17710.0 | max reserved: 17710.0
[Rank 59] (after 1 iterations) memory (MB) | allocated: 10837.39501953125 | max allocated: 15446.84716796875 | reserved: 17182.0 | max reserved: 17182.0
 iteration        2/    1000 | consumed samples:         2048 | elapsed time per iteration (ms): 141096.8 | learning rate: 3.750E-05 | global batch size:  1024 | lm-loss: 1.244502E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        3/    1000 | consumed samples:         3072 | elapsed time per iteration (ms): 137138.4 | learning rate: 5.625E-05 | global batch size:  1024 | lm-loss: 4.103157E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        4/    1000 | consumed samples:         4096 | elapsed time per iteration (ms): 138928.9 | learning rate: 7.500E-05 | global batch size:  1024 | lm-loss: 4.305696E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        5/    1000 | consumed samples:         5120 | elapsed time per iteration (ms): 137805.9 | learning rate: 9.375E-05 | global batch size:  1024 | lm-loss: 3.814122E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        6/    1000 | consumed samples:         6144 | elapsed time per iteration (ms): 139183.6 | learning rate: 1.125E-04 | global batch size:  1024 | lm-loss: 3.368778E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        7/    1000 | consumed samples:         7168 | elapsed time per iteration (ms): 138604.6 | learning rate: 1.312E-04 | global batch size:  1024 | lm-loss: 3.123441E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        8/    1000 | consumed samples:         8192 | elapsed time per iteration (ms): 137448.5 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 2.563856E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration        9/    1000 | consumed samples:         9216 | elapsed time per iteration (ms): 134118.7 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 2.213366E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       10/    1000 | consumed samples:        10240 | elapsed time per iteration (ms): 136533.1 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.981217E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       11/    1000 | consumed samples:        11264 | elapsed time per iteration (ms): 139544.9 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.872394E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       12/    1000 | consumed samples:        12288 | elapsed time per iteration (ms): 138324.6 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.740661E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       13/    1000 | consumed samples:        13312 | elapsed time per iteration (ms): 134446.2 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.575262E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       14/    1000 | consumed samples:        14336 | elapsed time per iteration (ms): 137764.0 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.397998E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       15/    1000 | consumed samples:        15360 | elapsed time per iteration (ms): 137041.8 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.245603E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       16/    1000 | consumed samples:        16384 | elapsed time per iteration (ms): 139143.0 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.082751E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       17/    1000 | consumed samples:        17408 | elapsed time per iteration (ms): 139118.9 | learning rate: 1.500E-04 | global batch size:  1024 | lm-loss: 1.204085E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       18/    1000 | consumed samples:        18432 | elapsed time per iteration (ms): 138928.4 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 1.150506E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       19/    1000 | consumed samples:        19456 | elapsed time per iteration (ms): 139037.8 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 1.115988E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       20/    1000 | consumed samples:        20480 | elapsed time per iteration (ms): 138096.1 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 9.714051E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       21/    1000 | consumed samples:        21504 | elapsed time per iteration (ms): 139033.1 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 9.586049E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       22/    1000 | consumed samples:        22528 | elapsed time per iteration (ms): 136872.8 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 9.537881E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       23/    1000 | consumed samples:        23552 | elapsed time per iteration (ms): 137788.2 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 9.239707E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       24/    1000 | consumed samples:        24576 | elapsed time per iteration (ms): 137068.7 | learning rate: 1.499E-04 | global batch size:  1024 | lm-loss: 8.807950E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       25/    1000 | consumed samples:        25600 | elapsed time per iteration (ms): 139326.6 | learning rate: 1.498E-04 | global batch size:  1024 | lm-loss: 9.411034E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       26/    1000 | consumed samples:        26624 | elapsed time per iteration (ms): 138753.7 | learning rate: 1.498E-04 | global batch size:  1024 | lm-loss: 1.019738E+01 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       27/    1000 | consumed samples:        27648 | elapsed time per iteration (ms): 135832.6 | learning rate: 1.498E-04 | global batch size:  1024 | lm-loss: 8.967265E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       28/    1000 | consumed samples:        28672 | elapsed time per iteration (ms): 137159.8 | learning rate: 1.498E-04 | global batch size:  1024 | lm-loss: 8.756670E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       29/    1000 | consumed samples:        29696 | elapsed time per iteration (ms): 135068.0 | learning rate: 1.498E-04 | global batch size:  1024 | lm-loss: 8.835566E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       30/    1000 | consumed samples:        30720 | elapsed time per iteration (ms): 135619.2 | learning rate: 1.497E-04 | global batch size:  1024 | lm-loss: 8.811040E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       31/    1000 | consumed samples:        31744 | elapsed time per iteration (ms): 137837.0 | learning rate: 1.497E-04 | global batch size:  1024 | lm-loss: 8.659844E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       32/    1000 | consumed samples:        32768 | elapsed time per iteration (ms): 135370.3 | learning rate: 1.497E-04 | global batch size:  1024 | lm-loss: 8.494865E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       33/    1000 | consumed samples:        33792 | elapsed time per iteration (ms): 132840.5 | learning rate: 1.497E-04 | global batch size:  1024 | lm-loss: 8.415603E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       34/    1000 | consumed samples:        34816 | elapsed time per iteration (ms): 135995.0 | learning rate: 1.496E-04 | global batch size:  1024 | lm-loss: 8.276673E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       35/    1000 | consumed samples:        35840 | elapsed time per iteration (ms): 130121.0 | learning rate: 1.496E-04 | global batch size:  1024 | lm-loss: 8.076686E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       36/    1000 | consumed samples:        36864 | elapsed time per iteration (ms): 134088.0 | learning rate: 1.496E-04 | global batch size:  1024 | lm-loss: 7.927558E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       37/    1000 | consumed samples:        37888 | elapsed time per iteration (ms): 132751.8 | learning rate: 1.495E-04 | global batch size:  1024 | lm-loss: 8.049387E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       38/    1000 | consumed samples:        38912 | elapsed time per iteration (ms): 137618.8 | learning rate: 1.495E-04 | global batch size:  1024 | lm-loss: 8.101182E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       39/    1000 | consumed samples:        39936 | elapsed time per iteration (ms): 136129.3 | learning rate: 1.495E-04 | global batch size:  1024 | lm-loss: 8.031030E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       40/    1000 | consumed samples:        40960 | elapsed time per iteration (ms): 125643.3 | learning rate: 1.494E-04 | global batch size:  1024 | lm-loss: 8.032815E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       41/    1000 | consumed samples:        41984 | elapsed time per iteration (ms): 137845.6 | learning rate: 1.494E-04 | global batch size:  1024 | lm-loss: 8.030648E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       42/    1000 | consumed samples:        43008 | elapsed time per iteration (ms): 136653.4 | learning rate: 1.494E-04 | global batch size:  1024 | lm-loss: 7.932028E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       43/    1000 | consumed samples:        44032 | elapsed time per iteration (ms): 133720.0 | learning rate: 1.493E-04 | global batch size:  1024 | lm-loss: 7.879141E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       44/    1000 | consumed samples:        45056 | elapsed time per iteration (ms): 134441.1 | learning rate: 1.493E-04 | global batch size:  1024 | lm-loss: 7.791877E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       45/    1000 | consumed samples:        46080 | elapsed time per iteration (ms): 137502.0 | learning rate: 1.492E-04 | global batch size:  1024 | lm-loss: 7.738390E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       46/    1000 | consumed samples:        47104 | elapsed time per iteration (ms): 131717.1 | learning rate: 1.492E-04 | global batch size:  1024 | lm-loss: 7.792564E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       47/    1000 | consumed samples:        48128 | elapsed time per iteration (ms): 134668.9 | learning rate: 1.492E-04 | global batch size:  1024 | lm-loss: 7.803430E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       48/    1000 | consumed samples:        49152 | elapsed time per iteration (ms): 134516.4 | learning rate: 1.491E-04 | global batch size:  1024 | lm-loss: 7.790527E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)
 iteration       49/    1000 | consumed samples:        50176 | elapsed time per iteration (ms): 136328.8 | learning rate: 1.491E-04 | global batch size:  1024 | lm-loss: 7.747273E+00 | loss scale: -1.0 | grad norm: 0.000 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
time (ms)