peacock-data-public-datasets-idc-bigscience
/
experiments
/gpt2-meg-ds-3d
/meg_ds_3d_gpt2_perf_n16-ds-off.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 ....................................... False | |
| deepspeed_activation_checkpointing .............. False | |
| deepspeed_config ................................ None | |
| 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 ...................................... 1 | |
| -------------------- 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 ... | |
| > 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: 0.099 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( | |
| /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: 16.555 seconds | |
| time to initialize megatron (seconds): 70.184 | |
| [after megatron is initialized] datetime: 2021-06-10 23:06:14 | |
| building GPT model ... | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 2): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 3): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 6): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 8): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 2): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 4): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 10): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 14): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 10): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 6): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 11): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 8): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 13): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 2): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 8): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 2): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 1): 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, 12): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 12): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 3): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 3): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 6): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 10): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 3): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 10): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 6): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 13): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 8): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 9): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 1): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 1): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 1): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 5): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 5): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 5): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 14): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 5): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 14): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 14): 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 (1, 9): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 9): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 4): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 9): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 12): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 12): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 11): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 11): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 11): 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 (0, 7): 805560320 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 7): 805560320 | |
| [2021-06-10 23:06:14,218] [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 23:06:14,219] [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 23:06:14,220] [INFO] [utils.py:636:see_memory_usage] CPU Virtual Memory: used = 39.0 GB, percent = 20.8% | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 15): 909385728 | |
| > number of parameters on (tensor, pipeline) model parallel rank (1, 0): 917757952 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 15): 909385728 | |
| > number of parameters on (tensor, pipeline) model parallel rank (2, 0): 917757952 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 15): 909385728 | |
| > number of parameters on (tensor, pipeline) model parallel rank (3, 0): 917757952 | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 15): 909385728 | |
| [2021-06-10 23:06:14,491] [INFO] [utils.py:627:see_memory_usage] After Building Model | |
| [2021-06-10 23:06:14,491] [INFO] [utils.py:628:see_memory_usage] MA 1.69 GB Max_MA 1.69 GB CA 1.7 GB Max_CA 2 GB | |
| [2021-06-10 23:06:14,492] [INFO] [utils.py:636:see_memory_usage] CPU Virtual Memory: used = 39.17 GB, percent = 20.9% | |
| > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 917757952 | |
| > learning rate decay style: cosine | |
| 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: 0.23 | |
| [after model, optimizer, and learning rate scheduler are built] datetime: 2021-06-10 23:06:14 | |
| > 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 0.000764 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 0.012 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 0.002 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 0.001 seconds | |
| total number of samples: 10255 | |
| total number of epochs: 672 | |
| > finished creating GPT datasets ... | |
| [after dataloaders are built] datetime: 2021-06-10 23:06:15 | |
| time (ms) | model-and-optimizer-setup: 336.26 | train/valid/test-data-iterators-setup: 662.92 | |
| done with setup ... | |
| training ... | |
| [before the start of training step] datetime: 2021-06-10 23:06:15 | |
| [Rank 43] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18738.0 | max reserved: 18738.0 | |
| [Rank 21] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20344.0 | max reserved: 20344.0 | |
| [Rank 23] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20278.0 | max reserved: 20278.0 | |
| [Rank 22] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20344.0 | max reserved: 20344.0 | |
| [Rank 25] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20022.0 | max reserved: 20022.0 | |
| [Rank 59] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17454.0 | max reserved: 17454.0 | |
| [Rank 41] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18672.0 | max reserved: 18672.0 | |
| [Rank 13] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20986.0 | max reserved: 20986.0 | |
| [Rank 15] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20986.0 | max reserved: 20986.0 | |
| [Rank 14] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20920.0 | max reserved: 20920.0 | |
| [Rank 26] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20022.0 | max reserved: 20022.0 | |
| [Rank 58] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17454.0 | max reserved: 17454.0 | |
| [Rank 12] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20920.0 | max reserved: 20920.0 | |
| [Rank 40] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18672.0 | max reserved: 18672.0 | |
| [Rank 37] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18994.0 | max reserved: 18994.0 | |
| [Rank 24] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20022.0 | max reserved: 20022.0 | |
| [Rank 20] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20278.0 | max reserved: 20278.0 | |
| [Rank 27] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20022.0 | max reserved: 20022.0 | |
| [Rank 56] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17454.0 | max reserved: 17454.0 | |
| [Rank 36] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19060.0 | max reserved: 19060.0 | |
| [Rank 57] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17454.0 | max reserved: 17454.0 | |
| [Rank 29] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19702.0 | max reserved: 19702.0 | |
| [Rank 30] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19702.0 | max reserved: 19702.0 | |
| [Rank 39] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19060.0 | max reserved: 19060.0 | |
| [Rank 38] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19060.0 | max reserved: 19060.0 | |
| [Rank 42] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18672.0 | max reserved: 18672.0 | |
| [Rank 31] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19702.0 | max reserved: 19702.0 | |
| [Rank 11] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21306.0 | max reserved: 21306.0 | |
| [Rank 9] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21306.0 | max reserved: 21306.0 | |
| [Rank 10] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21306.0 | max reserved: 21306.0 | |
| [Rank 19] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20598.0 | max reserved: 20598.0 | |
| [Rank 54] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17774.0 | max reserved: 17774.0 | |
| [Rank 17] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20664.0 | max reserved: 20664.0 | |
| [Rank 8] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21306.0 | max reserved: 21306.0 | |
| [Rank 48] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18096.0 | max reserved: 18096.0 | |
| [Rank 53] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17776.0 | max reserved: 17776.0 | |
| [Rank 49] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18030.0 | max reserved: 18030.0 | |
| [Rank 16] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20598.0 | max reserved: 20598.0 | |
| [Rank 52] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17776.0 | max reserved: 17776.0 | |
| [Rank 28] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19702.0 | max reserved: 19702.0 | |
| [Rank 18] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 20598.0 | max reserved: 20598.0 | |
| [Rank 51] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18096.0 | max reserved: 18096.0 | |
| [Rank 7] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21628.0 | max reserved: 21628.0 | |
| [Rank 44] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18418.0 | max reserved: 18418.0 | |
| [Rank 4] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21562.0 | max reserved: 21562.0 | |
| [Rank 6] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21628.0 | max reserved: 21628.0 | |
| [Rank 5] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 21562.0 | max reserved: 21562.0 | |
| [Rank 45] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18352.0 | max reserved: 18352.0[Rank 47] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18418.0 | max reserved: 18418.0 | |
| [Rank 55] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 17776.0 | max reserved: 17776.0 | |
| [Rank 50] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18096.0 | max reserved: 18096.0 | |
| [Rank 46] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 18384.0 | max reserved: 18384.0 | |
| [Rank 32] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19380.0 | max reserved: 19380.0 | |
| [Rank 34] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19380.0 | max reserved: 19380.0 | |
| [Rank 35] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19380.0 | max reserved: 19380.0 | |
| [Rank 33] (after 1 iterations) memory (MB) | allocated: 15492.84619140625 | max allocated: 15492.84619140625 | reserved: 19378.0 | max reserved: 19378.0 | |
| iteration 1/ 1000 | consumed samples: 1024 | elapsed time per iteration (ms): 144915.6 | learning rate: 1.875E-05 | global batch size: 1024 | lm loss: 1.244238E+01 | loss scale: 12.0 | grad norm: 67.593 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| [Rank 60] (after 1 iterations) memory (MB) | allocated: 17473.15966796875 | max allocated: 17473.28466796875 | reserved: 18882.0 | max reserved: 18882.0 | |
| [Rank 61] (after 1 iterations) memory (MB) | allocated: 17473.15966796875 | max allocated: 17473.28466796875 | reserved: 18882.0 | max reserved: 18882.0 | |
| [Rank 62] (after 1 iterations) memory (MB) | allocated: 17473.15966796875 | max allocated: 17473.28466796875 | reserved: 18882.0 | max reserved: 18882.0 | |
| [Rank 63] (after 1 iterations) memory (MB) | allocated: 17473.15966796875 | max allocated: 17473.28466796875 | reserved: 18882.0 | max reserved: 18882.0 | |
| time (ms) | forward-compute: 29189.57 | forward-recv: 15653.89 | backward-compute: 76786.07 | backward-send: 3.27 | backward-send-forward-recv: 18493.89 | backward-params-all-reduce: 26.63 | backward-embedding-all-reduce: 4254.14 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 317.20 | optimizer-clip-main-grad: 55.00 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 470.39 | batch-generator: 206.47 | |
| [Rank 3] (after 1 iterations) memory (MB) | allocated: 17504.84619140625 | max allocated: 17504.84619140625 | reserved: 24410.0 | max reserved: 24410.0 | |
| [Rank 2] (after 1 iterations) memory (MB) | allocated: 17504.84619140625 | max allocated: 17504.84619140625 | reserved: 24460.0 | max reserved: 24460.0 | |
| [Rank 1] (after 1 iterations) memory (MB) | allocated: 17504.84619140625 | max allocated: 17504.84619140625 | reserved: 24428.0 | max reserved: 24428.0 | |
| [Rank 0] (after 1 iterations) memory (MB) | allocated: 17504.84619140625 | max allocated: 17504.84619140625 | reserved: 24444.0 | max reserved: 24444.0 | |
| iteration 2/ 1000 | consumed samples: 2048 | elapsed time per iteration (ms): 125536.8 | learning rate: 3.750E-05 | global batch size: 1024 | lm loss: 1.244502E+01 | loss scale: 12.0 | grad norm: 68.180 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28373.70 | forward-recv: 1502.84 | backward-compute: 75963.21 | backward-send: 2.93 | backward-send-forward-recv: 15307.54 | backward-params-all-reduce: 26.30 | backward-embedding-all-reduce: 4249.11 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 9.92 | optimizer-clip-main-grad: 14.46 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.25 | batch-generator: 180.53 | |
| iteration 3/ 1000 | consumed samples: 3072 | elapsed time per iteration (ms): 123997.6 | learning rate: 5.625E-05 | global batch size: 1024 | lm loss: 4.424266E+01 | loss scale: 12.0 | grad norm: 77.479 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28224.18 | forward-recv: 1498.15 | backward-compute: 75322.70 | backward-send: 2.89 | backward-send-forward-recv: 14230.93 | backward-params-all-reduce: 26.78 | backward-embedding-all-reduce: 4582.24 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.01 | optimizer-clip-main-grad: 14.41 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.19 | batch-generator: 161.65 | |
| iteration 4/ 1000 | consumed samples: 4096 | elapsed time per iteration (ms): 124018.6 | learning rate: 7.500E-05 | global batch size: 1024 | lm loss: 4.814127E+01 | loss scale: 12.0 | grad norm: 62.352 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28185.77 | forward-recv: 1496.78 | backward-compute: 74978.14 | backward-send: 3.14 | backward-send-forward-recv: 15078.20 | backward-params-all-reduce: 17.28 | backward-embedding-all-reduce: 4149.49 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 9.96 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.17 | batch-generator: 159.81 | |
| iteration 5/ 1000 | consumed samples: 5120 | elapsed time per iteration (ms): 126993.5 | learning rate: 9.375E-05 | global batch size: 1024 | lm loss: 4.750028E+01 | loss scale: 12.0 | grad norm: 62.615 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28137.30 | forward-recv: 1496.07 | backward-compute: 74904.71 | backward-send: 3.37 | backward-send-forward-recv: 17419.10 | backward-params-all-reduce: 17.31 | backward-embedding-all-reduce: 4905.64 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 9.87 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.27 | optimizer: 76.12 | batch-generator: 160.78 | |
| iteration 6/ 1000 | consumed samples: 6144 | elapsed time per iteration (ms): 124457.6 | learning rate: 1.125E-04 | global batch size: 1024 | lm loss: 4.659282E+01 | loss scale: 12.0 | grad norm: 62.860 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28160.31 | forward-recv: 1498.15 | backward-compute: 74913.02 | backward-send: 3.13 | backward-send-forward-recv: 15599.28 | backward-params-all-reduce: 17.22 | backward-embedding-all-reduce: 4156.74 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 9.94 | optimizer-clip-main-grad: 14.40 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.20 | batch-generator: 161.36 | |
| iteration 7/ 1000 | consumed samples: 7168 | elapsed time per iteration (ms): 126538.9 | learning rate: 1.312E-04 | global batch size: 1024 | lm loss: 4.565659E+01 | loss scale: 12.0 | grad norm: 62.898 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28153.19 | forward-recv: 1495.03 | backward-compute: 74902.44 | backward-send: 2.96 | backward-send-forward-recv: 17497.13 | backward-params-all-reduce: 17.25 | backward-embedding-all-reduce: 4361.06 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 9.96 | optimizer-clip-main-grad: 14.41 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.22 | batch-generator: 163.53 | |
| iteration 8/ 1000 | consumed samples: 8192 | elapsed time per iteration (ms): 124177.3 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 4.428070E+01 | loss scale: 12.0 | grad norm: 62.715 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28162.01 | forward-recv: 1503.69 | backward-compute: 74904.75 | backward-send: 3.03 | backward-send-forward-recv: 15319.15 | backward-params-all-reduce: 17.48 | backward-embedding-all-reduce: 4157.46 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 9.94 | optimizer-clip-main-grad: 14.51 | optimizer-copy-main-to-model-params: 8.26 | optimizer: 76.22 | batch-generator: 163.17 | |
| iteration 9/ 1000 | consumed samples: 9216 | elapsed time per iteration (ms): 129137.8 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 4.274238E+01 | loss scale: 12.0 | grad norm: 62.498 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28118.75 | forward-recv: 1500.66 | backward-compute: 74864.14 | backward-send: 2.74 | backward-send-forward-recv: 20369.06 | backward-params-all-reduce: 17.34 | backward-embedding-all-reduce: 4155.15 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 10.11 | optimizer-clip-main-grad: 14.51 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.38 | batch-generator: 165.20 | |
| iteration 10/ 1000 | consumed samples: 10240 | elapsed time per iteration (ms): 126697.3 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 4.105743E+01 | loss scale: 12.0 | grad norm: 62.250 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28147.09 | forward-recv: 1500.89 | backward-compute: 74930.89 | backward-send: 2.83 | backward-send-forward-recv: 17827.00 | backward-params-all-reduce: 17.28 | backward-embedding-all-reduce: 4161.53 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.00 | optimizer-clip-main-grad: 14.47 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.20 | batch-generator: 165.69 | |
| iteration 11/ 1000 | consumed samples: 11264 | elapsed time per iteration (ms): 126607.6 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.924586E+01 | loss scale: 12.0 | grad norm: 62.065 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28133.59 | forward-recv: 1505.41 | backward-compute: 74865.88 | backward-send: 2.91 | backward-send-forward-recv: 17814.19 | backward-params-all-reduce: 17.27 | backward-embedding-all-reduce: 4158.48 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.04 | optimizer-clip-main-grad: 14.46 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.40 | batch-generator: 165.96 | |
| iteration 12/ 1000 | consumed samples: 12288 | elapsed time per iteration (ms): 123082.9 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.742614E+01 | loss scale: 12.0 | grad norm: 61.519 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28172.96 | forward-recv: 1506.30 | backward-compute: 74955.83 | backward-send: 2.73 | backward-send-forward-recv: 14168.14 | backward-params-all-reduce: 17.34 | backward-embedding-all-reduce: 4149.68 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.01 | optimizer-clip-main-grad: 14.49 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.39 | batch-generator: 166.31 | |
| iteration 13/ 1000 | consumed samples: 13312 | elapsed time per iteration (ms): 127414.3 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.567815E+01 | loss scale: 12.0 | grad norm: 58.588 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28139.33 | forward-recv: 1504.03 | backward-compute: 74917.06 | backward-send: 3.15 | backward-send-forward-recv: 18568.81 | backward-params-all-reduce: 17.36 | backward-embedding-all-reduce: 4154.19 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 10.05 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.28 | batch-generator: 172.49 | |
| iteration 14/ 1000 | consumed samples: 14336 | elapsed time per iteration (ms): 129181.9 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.400079E+01 | loss scale: 12.0 | grad norm: 48.799 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28111.34 | forward-recv: 1504.04 | backward-compute: 74867.14 | backward-send: 2.96 | backward-send-forward-recv: 20418.93 | backward-params-all-reduce: 17.23 | backward-embedding-all-reduce: 4150.18 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 9.94 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.24 | batch-generator: 166.23 | |
| iteration 15/ 1000 | consumed samples: 15360 | elapsed time per iteration (ms): 124965.5 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.260079E+01 | loss scale: 12.0 | grad norm: 42.450 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28124.74 | forward-recv: 1506.52 | backward-compute: 74931.71 | backward-send: 3.15 | backward-send-forward-recv: 16107.40 | backward-params-all-reduce: 17.34 | backward-embedding-all-reduce: 4165.04 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 9.93 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.04 | batch-generator: 164.90 | |
| iteration 16/ 1000 | consumed samples: 16384 | elapsed time per iteration (ms): 125984.1 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 3.100228E+01 | loss scale: 12.0 | grad norm: 42.998 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28141.38 | forward-recv: 1507.71 | backward-compute: 74923.88 | backward-send: 2.92 | backward-send-forward-recv: 17107.86 | backward-params-all-reduce: 17.48 | backward-embedding-all-reduce: 4172.59 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 10.02 | optimizer-clip-main-grad: 14.54 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.37 | batch-generator: 166.42 | |
| iteration 17/ 1000 | consumed samples: 17408 | elapsed time per iteration (ms): 130254.5 | learning rate: 1.500E-04 | global batch size: 1024 | lm loss: 2.948225E+01 | loss scale: 12.0 | grad norm: 44.652 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28109.24 | forward-recv: 1509.03 | backward-compute: 74834.83 | backward-send: 2.86 | backward-send-forward-recv: 21514.62 | backward-params-all-reduce: 17.36 | backward-embedding-all-reduce: 4153.74 | optimizer-copy-to-main-grad: 8.21 | optimizer-unscale-and-check-inf: 12.77 | optimizer-clip-main-grad: 14.51 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 79.20 | batch-generator: 166.22 | |
| iteration 18/ 1000 | consumed samples: 18432 | elapsed time per iteration (ms): 124895.5 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.778518E+01 | loss scale: 12.0 | grad norm: 44.022 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28157.13 | forward-recv: 1509.98 | backward-compute: 74940.15 | backward-send: 3.05 | backward-send-forward-recv: 15997.64 | backward-params-all-reduce: 17.42 | backward-embedding-all-reduce: 4157.55 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 12.26 | optimizer-clip-main-grad: 14.51 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 78.65 | batch-generator: 171.84 | |
| iteration 19/ 1000 | consumed samples: 19456 | elapsed time per iteration (ms): 125352.3 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.632198E+01 | loss scale: 12.0 | grad norm: 38.373 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28160.31 | forward-recv: 1510.96 | backward-compute: 74958.33 | backward-send: 2.96 | backward-send-forward-recv: 16432.75 | backward-params-all-reduce: 17.44 | backward-embedding-all-reduce: 4156.55 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 12.63 | optimizer-clip-main-grad: 14.47 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 78.89 | batch-generator: 172.95 | |
| iteration 20/ 1000 | consumed samples: 20480 | elapsed time per iteration (ms): 124730.9 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.485645E+01 | loss scale: 12.0 | grad norm: 35.316 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28159.21 | forward-recv: 1510.71 | backward-compute: 74958.75 | backward-send: 3.02 | backward-send-forward-recv: 15816.65 | backward-params-all-reduce: 17.41 | backward-embedding-all-reduce: 4152.87 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 12.10 | optimizer-clip-main-grad: 14.48 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 78.39 | batch-generator: 170.32 | |
| iteration 21/ 1000 | consumed samples: 21504 | elapsed time per iteration (ms): 125326.2 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.330399E+01 | loss scale: 12.0 | grad norm: 34.645 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28139.58 | forward-recv: 1511.38 | backward-compute: 74969.60 | backward-send: 2.86 | backward-send-forward-recv: 16425.18 | backward-params-all-reduce: 17.44 | backward-embedding-all-reduce: 4147.49 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 12.32 | optimizer-clip-main-grad: 14.50 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 78.70 | batch-generator: 170.82 | |
| iteration 22/ 1000 | consumed samples: 22528 | elapsed time per iteration (ms): 124804.8 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.197279E+01 | loss scale: 12.0 | grad norm: 31.805 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28150.72 | forward-recv: 1510.22 | backward-compute: 74962.34 | backward-send: 3.10 | backward-send-forward-recv: 15891.01 | backward-params-all-reduce: 17.46 | backward-embedding-all-reduce: 4156.29 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 13.29 | optimizer-clip-main-grad: 14.48 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 79.65 | batch-generator: 170.69 | |
| iteration 23/ 1000 | consumed samples: 23552 | elapsed time per iteration (ms): 122173.3 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 2.054678E+01 | loss scale: 12.0 | grad norm: 30.377 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28178.65 | forward-recv: 1512.05 | backward-compute: 74968.11 | backward-send: 2.95 | backward-send-forward-recv: 13228.55 | backward-params-all-reduce: 17.46 | backward-embedding-all-reduce: 4151.52 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 14.11 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.27 | optimizer: 80.40 | batch-generator: 171.66 | |
| iteration 24/ 1000 | consumed samples: 24576 | elapsed time per iteration (ms): 127877.7 | learning rate: 1.499E-04 | global batch size: 1024 | lm loss: 1.917008E+01 | loss scale: 12.0 | grad norm: 33.208 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28127.78 | forward-recv: 1509.79 | backward-compute: 74915.57 | backward-send: 2.92 | backward-send-forward-recv: 19039.93 | backward-params-all-reduce: 17.33 | backward-embedding-all-reduce: 4153.33 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.98 | optimizer-clip-main-grad: 14.49 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 77.27 | batch-generator: 172.01 | |
| iteration 25/ 1000 | consumed samples: 25600 | elapsed time per iteration (ms): 120406.6 | learning rate: 1.498E-04 | global batch size: 1024 | lm loss: 1.783947E+01 | loss scale: 12.0 | grad norm: 35.653 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28194.44 | forward-recv: 1511.80 | backward-compute: 75008.97 | backward-send: 2.85 | backward-send-forward-recv: 11398.39 | backward-params-all-reduce: 17.37 | backward-embedding-all-reduce: 4161.05 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 11.59 | optimizer-clip-main-grad: 14.51 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 77.95 | batch-generator: 169.42 | |
| iteration 26/ 1000 | consumed samples: 26624 | elapsed time per iteration (ms): 125256.9 | learning rate: 1.498E-04 | global batch size: 1024 | lm loss: 1.633506E+01 | loss scale: 12.0 | grad norm: 35.190 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28162.63 | forward-recv: 1511.53 | backward-compute: 74942.58 | backward-send: 2.91 | backward-send-forward-recv: 15998.76 | backward-params-all-reduce: 17.25 | backward-embedding-all-reduce: 4510.63 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 10.53 | optimizer-clip-main-grad: 14.49 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.79 | batch-generator: 170.65 | |
| iteration 27/ 1000 | consumed samples: 27648 | elapsed time per iteration (ms): 125520.6 | learning rate: 1.498E-04 | global batch size: 1024 | lm loss: 1.496973E+01 | loss scale: 12.0 | grad norm: 29.656 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28170.75 | forward-recv: 1513.24 | backward-compute: 74952.75 | backward-send: 2.87 | backward-send-forward-recv: 16599.92 | backward-params-all-reduce: 17.29 | backward-embedding-all-reduce: 4153.18 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.46 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.70 | batch-generator: 170.60 | |
| iteration 28/ 1000 | consumed samples: 28672 | elapsed time per iteration (ms): 130263.9 | learning rate: 1.498E-04 | global batch size: 1024 | lm loss: 1.371372E+01 | loss scale: 12.0 | grad norm: 20.988 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28122.58 | forward-recv: 1506.03 | backward-compute: 74880.39 | backward-send: 2.98 | backward-send-forward-recv: 21471.84 | backward-params-all-reduce: 17.37 | backward-embedding-all-reduce: 4152.40 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.22 | optimizer-clip-main-grad: 14.44 | optimizer-copy-main-to-model-params: 8.28 | optimizer: 76.50 | batch-generator: 172.90 | |
| iteration 29/ 1000 | consumed samples: 29696 | elapsed time per iteration (ms): 126795.7 | learning rate: 1.498E-04 | global batch size: 1024 | lm loss: 1.267445E+01 | loss scale: 12.0 | grad norm: 15.112 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28141.51 | forward-recv: 1508.34 | backward-compute: 74939.67 | backward-send: 2.82 | backward-send-forward-recv: 17884.13 | backward-params-all-reduce: 17.40 | backward-embedding-all-reduce: 4191.10 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.76 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.95 | batch-generator: 171.74 | |
| iteration 30/ 1000 | consumed samples: 30720 | elapsed time per iteration (ms): 127372.5 | learning rate: 1.497E-04 | global batch size: 1024 | lm loss: 1.187400E+01 | loss scale: 12.0 | grad norm: 8.336 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28127.22 | forward-recv: 1515.11 | backward-compute: 74916.47 | backward-send: 3.06 | backward-send-forward-recv: 18527.14 | backward-params-all-reduce: 17.45 | backward-embedding-all-reduce: 4156.30 | optimizer-copy-to-main-grad: 8.15 | optimizer-unscale-and-check-inf: 10.06 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.27 | optimizer: 76.19 | batch-generator: 172.96 | |
| iteration 31/ 1000 | consumed samples: 31744 | elapsed time per iteration (ms): 128454.5 | learning rate: 1.497E-04 | global batch size: 1024 | lm loss: 1.164951E+01 | loss scale: 12.0 | grad norm: 6.203 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28115.20 | forward-recv: 1506.38 | backward-compute: 74858.54 | backward-send: 2.92 | backward-send-forward-recv: 19688.48 | backward-params-all-reduce: 17.20 | backward-embedding-all-reduce: 4154.96 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.72 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.33 | optimizer: 76.97 | batch-generator: 173.55 | |
| iteration 32/ 1000 | consumed samples: 32768 | elapsed time per iteration (ms): 126586.1 | learning rate: 1.497E-04 | global batch size: 1024 | lm loss: 1.183907E+01 | loss scale: 12.0 | grad norm: 7.559 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28113.34 | forward-recv: 1505.21 | backward-compute: 74877.84 | backward-send: 2.97 | backward-send-forward-recv: 17810.71 | backward-params-all-reduce: 17.23 | backward-embedding-all-reduce: 4148.96 | optimizer-copy-to-main-grad: 8.19 | optimizer-unscale-and-check-inf: 9.97 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.24 | batch-generator: 171.62 | |
| iteration 33/ 1000 | consumed samples: 33792 | elapsed time per iteration (ms): 126448.8 | learning rate: 1.497E-04 | global batch size: 1024 | lm loss: 1.232106E+01 | loss scale: 12.0 | grad norm: 7.904 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28162.67 | forward-recv: 1505.48 | backward-compute: 74926.76 | backward-send: 2.86 | backward-send-forward-recv: 17561.21 | backward-params-all-reduce: 17.34 | backward-embedding-all-reduce: 4161.60 | optimizer-copy-to-main-grad: 8.19 | optimizer-unscale-and-check-inf: 10.92 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 77.28 | batch-generator: 173.12 | |
| iteration 34/ 1000 | consumed samples: 34816 | elapsed time per iteration (ms): 129470.1 | learning rate: 1.496E-04 | global batch size: 1024 | lm loss: 1.280134E+01 | loss scale: 12.0 | grad norm: 7.885 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28125.41 | forward-recv: 1508.74 | backward-compute: 74935.94 | backward-send: 2.86 | backward-send-forward-recv: 20614.69 | backward-params-all-reduce: 17.47 | backward-embedding-all-reduce: 4154.99 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.02 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.32 | batch-generator: 172.22 | |
| iteration 35/ 1000 | consumed samples: 35840 | elapsed time per iteration (ms): 127702.0 | learning rate: 1.496E-04 | global batch size: 1024 | lm loss: 1.324976E+01 | loss scale: 12.0 | grad norm: 7.859 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28109.27 | forward-recv: 1506.62 | backward-compute: 74912.59 | backward-send: 3.04 | backward-send-forward-recv: 18883.61 | backward-params-all-reduce: 17.17 | backward-embedding-all-reduce: 4157.66 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 11.84 | optimizer-clip-main-grad: 14.44 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 78.13 | batch-generator: 173.23 | |
| iteration 36/ 1000 | consumed samples: 36864 | elapsed time per iteration (ms): 125749.8 | learning rate: 1.496E-04 | global batch size: 1024 | lm loss: 1.368772E+01 | loss scale: 12.0 | grad norm: 7.905 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28142.76 | forward-recv: 1509.66 | backward-compute: 74946.29 | backward-send: 2.96 | backward-send-forward-recv: 16867.87 | backward-params-all-reduce: 17.28 | backward-embedding-all-reduce: 4152.92 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 9.97 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.21 | batch-generator: 173.67 | |
| iteration 37/ 1000 | consumed samples: 37888 | elapsed time per iteration (ms): 127395.5 | learning rate: 1.495E-04 | global batch size: 1024 | lm loss: 1.418495E+01 | loss scale: 12.0 | grad norm: 7.839 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28116.48 | forward-recv: 1508.57 | backward-compute: 74932.39 | backward-send: 2.84 | backward-send-forward-recv: 18552.82 | backward-params-all-reduce: 17.47 | backward-embedding-all-reduce: 4154.04 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 10.70 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.97 | batch-generator: 171.40 | |
| iteration 38/ 1000 | consumed samples: 38912 | elapsed time per iteration (ms): 126640.6 | learning rate: 1.495E-04 | global batch size: 1024 | lm loss: 1.455190E+01 | loss scale: 12.0 | grad norm: 7.850 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28126.21 | forward-recv: 1505.76 | backward-compute: 74914.70 | backward-send: 2.80 | backward-send-forward-recv: 17811.30 | backward-params-all-reduce: 17.11 | backward-embedding-all-reduce: 4152.47 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.26 | optimizer-clip-main-grad: 14.39 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.43 | batch-generator: 171.98 | |
| iteration 39/ 1000 | consumed samples: 39936 | elapsed time per iteration (ms): 125915.8 | learning rate: 1.495E-04 | global batch size: 1024 | lm loss: 1.497495E+01 | loss scale: 12.0 | grad norm: 7.889 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28169.51 | forward-recv: 1510.71 | backward-compute: 74975.56 | backward-send: 2.96 | backward-send-forward-recv: 16975.06 | backward-params-all-reduce: 17.40 | backward-embedding-all-reduce: 4154.63 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 9.97 | optimizer-clip-main-grad: 14.41 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.24 | batch-generator: 171.44 | |
| iteration 40/ 1000 | consumed samples: 40960 | elapsed time per iteration (ms): 125625.3 | learning rate: 1.494E-04 | global batch size: 1024 | lm loss: 1.537068E+01 | loss scale: 12.0 | grad norm: 7.901 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28174.54 | forward-recv: 1506.25 | backward-compute: 74935.33 | backward-send: 3.12 | backward-send-forward-recv: 16725.04 | backward-params-all-reduce: 17.42 | backward-embedding-all-reduce: 4151.71 | optimizer-copy-to-main-grad: 8.14 | optimizer-unscale-and-check-inf: 10.03 | optimizer-clip-main-grad: 14.47 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.32 | batch-generator: 177.88 | |
| iteration 41/ 1000 | consumed samples: 41984 | elapsed time per iteration (ms): 126693.8 | learning rate: 1.494E-04 | global batch size: 1024 | lm loss: 1.567975E+01 | loss scale: 12.0 | grad norm: 7.874 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28155.24 | forward-recv: 1507.36 | backward-compute: 74940.74 | backward-send: 2.91 | backward-send-forward-recv: 17795.36 | backward-params-all-reduce: 17.58 | backward-embedding-all-reduce: 4157.95 | optimizer-copy-to-main-grad: 8.19 | optimizer-unscale-and-check-inf: 11.27 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 77.58 | batch-generator: 196.62 | |
| iteration 42/ 1000 | consumed samples: 43008 | elapsed time per iteration (ms): 125836.8 | learning rate: 1.494E-04 | global batch size: 1024 | lm loss: 1.602291E+01 | loss scale: 12.0 | grad norm: 7.956 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28172.41 | forward-recv: 1508.20 | backward-compute: 74952.39 | backward-send: 2.87 | backward-send-forward-recv: 15944.05 | backward-params-all-reduce: 17.45 | backward-embedding-all-reduce: 5123.10 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 11.64 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 77.95 | batch-generator: 195.91 | |
| iteration 43/ 1000 | consumed samples: 44032 | elapsed time per iteration (ms): 128515.1 | learning rate: 1.493E-04 | global batch size: 1024 | lm loss: 1.632536E+01 | loss scale: 12.0 | grad norm: 7.877 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28138.85 | forward-recv: 1509.12 | backward-compute: 74928.86 | backward-send: 2.80 | backward-send-forward-recv: 19649.06 | backward-params-all-reduce: 17.57 | backward-embedding-all-reduce: 4152.01 | optimizer-copy-to-main-grad: 8.19 | optimizer-unscale-and-check-inf: 11.72 | optimizer-clip-main-grad: 14.46 | optimizer-copy-main-to-model-params: 8.33 | optimizer: 78.13 | batch-generator: 196.68 | |
| iteration 44/ 1000 | consumed samples: 45056 | elapsed time per iteration (ms): 126234.3 | learning rate: 1.493E-04 | global batch size: 1024 | lm loss: 1.656669E+01 | loss scale: 12.0 | grad norm: 7.903 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28117.72 | forward-recv: 1507.85 | backward-compute: 74904.93 | backward-send: 3.14 | backward-send-forward-recv: 17411.21 | backward-params-all-reduce: 17.44 | backward-embedding-all-reduce: 4155.93 | optimizer-copy-to-main-grad: 8.19 | optimizer-unscale-and-check-inf: 10.91 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 77.21 | batch-generator: 196.97 | |
| iteration 45/ 1000 | consumed samples: 46080 | elapsed time per iteration (ms): 128778.4 | learning rate: 1.492E-04 | global batch size: 1024 | lm loss: 1.695541E+01 | loss scale: 12.0 | grad norm: 7.994 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28131.03 | forward-recv: 1506.69 | backward-compute: 74917.48 | backward-send: 2.99 | backward-send-forward-recv: 19923.71 | backward-params-all-reduce: 17.48 | backward-embedding-all-reduce: 4163.18 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.92 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.33 | optimizer: 77.21 | batch-generator: 196.63 | |
| iteration 46/ 1000 | consumed samples: 47104 | elapsed time per iteration (ms): 127798.5 | learning rate: 1.492E-04 | global batch size: 1024 | lm loss: 1.719514E+01 | loss scale: 12.0 | grad norm: 7.913 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28148.95 | forward-recv: 1505.02 | backward-compute: 74924.96 | backward-send: 2.84 | backward-send-forward-recv: 18926.61 | backward-params-all-reduce: 17.53 | backward-embedding-all-reduce: 4156.85 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.63 | optimizer-clip-main-grad: 14.47 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.94 | batch-generator: 196.02 | |
| iteration 47/ 1000 | consumed samples: 48128 | elapsed time per iteration (ms): 126873.0 | learning rate: 1.492E-04 | global batch size: 1024 | lm loss: 1.741152E+01 | loss scale: 12.0 | grad norm: 7.909 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28160.07 | forward-recv: 1502.62 | backward-compute: 74977.97 | backward-send: 3.04 | backward-send-forward-recv: 17936.91 | backward-params-all-reduce: 17.66 | backward-embedding-all-reduce: 4159.89 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.00 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 76.30 | batch-generator: 195.48 | |
| iteration 48/ 1000 | consumed samples: 49152 | elapsed time per iteration (ms): 127674.6 | learning rate: 1.491E-04 | global batch size: 1024 | lm loss: 1.768794E+01 | loss scale: 12.0 | grad norm: 7.879 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28128.53 | forward-recv: 1504.40 | backward-compute: 74884.15 | backward-send: 2.92 | backward-send-forward-recv: 18860.09 | backward-params-all-reduce: 17.45 | backward-embedding-all-reduce: 4161.86 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 10.03 | optimizer-clip-main-grad: 14.44 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 76.41 | batch-generator: 195.68 | |
| iteration 49/ 1000 | consumed samples: 50176 | elapsed time per iteration (ms): 126896.6 | learning rate: 1.491E-04 | global batch size: 1024 | lm loss: 1.792036E+01 | loss scale: 12.0 | grad norm: 7.910 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28135.60 | forward-recv: 1501.27 | backward-compute: 74912.60 | backward-send: 2.99 | backward-send-forward-recv: 18051.57 | backward-params-all-reduce: 17.56 | backward-embedding-all-reduce: 4159.74 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 9.95 | optimizer-clip-main-grad: 14.47 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.33 | batch-generator: 193.44 | |
| iteration 50/ 1000 | consumed samples: 51200 | elapsed time per iteration (ms): 126914.2 | learning rate: 1.490E-04 | global batch size: 1024 | lm loss: 1.818993E+01 | loss scale: 12.0 | grad norm: 8.010 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28112.45 | forward-recv: 1500.33 | backward-compute: 74891.05 | backward-send: 2.90 | backward-send-forward-recv: 17847.70 | backward-params-all-reduce: 17.68 | backward-embedding-all-reduce: 4426.43 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.27 | optimizer-clip-main-grad: 14.41 | optimizer-copy-main-to-model-params: 8.33 | optimizer: 76.57 | batch-generator: 197.73 | |
| iteration 51/ 1000 | consumed samples: 52224 | elapsed time per iteration (ms): 129434.9 | learning rate: 1.490E-04 | global batch size: 1024 | lm loss: 1.835458E+01 | loss scale: 12.0 | grad norm: 7.958 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28125.14 | forward-recv: 1502.55 | backward-compute: 74893.74 | backward-send: 2.88 | backward-send-forward-recv: 20614.42 | backward-params-all-reduce: 17.56 | backward-embedding-all-reduce: 4163.45 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.20 | optimizer-clip-main-grad: 14.45 | optimizer-copy-main-to-model-params: 8.29 | optimizer: 76.48 | batch-generator: 197.12 | |
| iteration 52/ 1000 | consumed samples: 53248 | elapsed time per iteration (ms): 124920.5 | learning rate: 1.489E-04 | global batch size: 1024 | lm loss: 1.865323E+01 | loss scale: 12.0 | grad norm: 7.982 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28131.92 | forward-recv: 1503.25 | backward-compute: 74921.83 | backward-send: 2.96 | backward-send-forward-recv: 16059.25 | backward-params-all-reduce: 17.61 | backward-embedding-all-reduce: 4167.66 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.96 | optimizer-clip-main-grad: 14.48 | optimizer-copy-main-to-model-params: 8.32 | optimizer: 77.35 | batch-generator: 193.26 | |
| iteration 53/ 1000 | consumed samples: 54272 | elapsed time per iteration (ms): 127742.1 | learning rate: 1.489E-04 | global batch size: 1024 | lm loss: 1.887249E+01 | loss scale: 12.0 | grad norm: 7.968 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28129.03 | forward-recv: 1502.93 | backward-compute: 74897.21 | backward-send: 3.10 | backward-send-forward-recv: 18917.75 | backward-params-all-reduce: 17.56 | backward-embedding-all-reduce: 4157.41 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 11.62 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 77.91 | batch-generator: 194.19 | |
| iteration 54/ 1000 | consumed samples: 55296 | elapsed time per iteration (ms): 129973.5 | learning rate: 1.488E-04 | global batch size: 1024 | lm loss: 1.903958E+01 | loss scale: 12.0 | grad norm: 7.962 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28079.12 | forward-recv: 1502.86 | backward-compute: 74860.08 | backward-send: 2.96 | backward-send-forward-recv: 21231.49 | backward-params-all-reduce: 17.51 | backward-embedding-all-reduce: 4163.70 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 10.56 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.88 | batch-generator: 192.44 | |
| iteration 55/ 1000 | consumed samples: 56320 | elapsed time per iteration (ms): 127636.3 | learning rate: 1.488E-04 | global batch size: 1024 | lm loss: 1.920096E+01 | loss scale: 12.0 | grad norm: 8.005 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28097.78 | forward-recv: 1499.86 | backward-compute: 74889.06 | backward-send: 3.01 | backward-send-forward-recv: 18857.67 | backward-params-all-reduce: 17.64 | backward-embedding-all-reduce: 4155.96 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.02 | optimizer-clip-main-grad: 14.44 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.28 | batch-generator: 194.35 | |
| iteration 56/ 1000 | consumed samples: 57344 | elapsed time per iteration (ms): 127092.9 | learning rate: 1.487E-04 | global batch size: 1024 | lm loss: 1.939602E+01 | loss scale: 12.0 | grad norm: 7.939 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28119.17 | forward-recv: 1503.30 | backward-compute: 74908.49 | backward-send: 2.94 | backward-send-forward-recv: 18269.93 | backward-params-all-reduce: 17.58 | backward-embedding-all-reduce: 4156.14 | optimizer-copy-to-main-grad: 8.18 | optimizer-unscale-and-check-inf: 9.95 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.27 | batch-generator: 194.21 | |
| iteration 57/ 1000 | consumed samples: 58368 | elapsed time per iteration (ms): 125650.0 | learning rate: 1.487E-04 | global batch size: 1024 | lm loss: 1.963987E+01 | loss scale: 12.0 | grad norm: 7.994 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28128.79 | forward-recv: 1503.76 | backward-compute: 74883.32 | backward-send: 2.97 | backward-send-forward-recv: 16837.12 | backward-params-all-reduce: 17.60 | backward-embedding-all-reduce: 4161.37 | optimizer-copy-to-main-grad: 8.16 | optimizer-unscale-and-check-inf: 10.01 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.40 | batch-generator: 196.59 | |
| iteration 58/ 1000 | consumed samples: 59392 | elapsed time per iteration (ms): 128251.4 | learning rate: 1.486E-04 | global batch size: 1024 | lm loss: 1.979258E+01 | loss scale: 12.0 | grad norm: 7.987 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28128.73 | forward-recv: 1506.26 | backward-compute: 74901.21 | backward-send: 2.85 | backward-send-forward-recv: 18980.76 | backward-params-all-reduce: 17.43 | backward-embedding-all-reduce: 4598.90 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 10.17 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 76.59 | batch-generator: 194.26 | |
| iteration 59/ 1000 | consumed samples: 60416 | elapsed time per iteration (ms): 128585.3 | learning rate: 1.486E-04 | global batch size: 1024 | lm loss: 1.998816E+01 | loss scale: 12.0 | grad norm: 7.995 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28107.46 | forward-recv: 1505.32 | backward-compute: 74920.44 | backward-send: 3.34 | backward-send-forward-recv: 19342.07 | backward-params-all-reduce: 17.60 | backward-embedding-all-reduce: 4572.24 | optimizer-copy-to-main-grad: 8.21 | optimizer-unscale-and-check-inf: 11.58 | optimizer-clip-main-grad: 14.44 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 77.93 | batch-generator: 193.94 | |
| iteration 60/ 1000 | consumed samples: 61440 | elapsed time per iteration (ms): 126533.1 | learning rate: 1.485E-04 | global batch size: 1024 | lm loss: 2.011507E+01 | loss scale: 12.0 | grad norm: 7.934 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28100.85 | forward-recv: 1518.02 | backward-compute: 74862.41 | backward-send: 3.28 | backward-send-forward-recv: 17608.70 | backward-params-all-reduce: 17.48 | backward-embedding-all-reduce: 4304.46 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 12.61 | optimizer-clip-main-grad: 14.43 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 78.88 | batch-generator: 194.19 | |
| iteration 61/ 1000 | consumed samples: 62464 | elapsed time per iteration (ms): 128838.5 | learning rate: 1.485E-04 | global batch size: 1024 | lm loss: 2.028895E+01 | loss scale: 12.0 | grad norm: 7.992 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28125.05 | forward-recv: 1509.59 | backward-compute: 74898.52 | backward-send: 3.29 | backward-send-forward-recv: 20011.24 | backward-params-all-reduce: 17.44 | backward-embedding-all-reduce: 4157.11 | optimizer-copy-to-main-grad: 8.20 | optimizer-unscale-and-check-inf: 11.07 | optimizer-clip-main-grad: 14.46 | optimizer-copy-main-to-model-params: 8.31 | optimizer: 77.43 | batch-generator: 194.49 | |
| iteration 62/ 1000 | consumed samples: 63488 | elapsed time per iteration (ms): 125247.8 | learning rate: 1.484E-04 | global batch size: 1024 | lm loss: 2.043871E+01 | loss scale: 12.0 | grad norm: 8.017 | number of skipped iterations: 0 | number of nan iterations: 0 | | |
| time (ms) | forward-compute: 28117.27 | forward-recv: 1503.24 | backward-compute: 74875.03 | backward-send: 3.35 | backward-send-forward-recv: 16454.95 | backward-params-all-reduce: 17.33 | backward-embedding-all-reduce: 4164.99 | optimizer-copy-to-main-grad: 8.17 | optimizer-unscale-and-check-inf: 10.69 | optimizer-clip-main-grad: 14.42 | optimizer-copy-main-to-model-params: 8.30 | optimizer: 76.94 | batch-generator: 175.93 | |