[03/11 00:22:39 libai]: Rank of current process: 0. World size: 4
[03/11 00:22:39 libai]: Command line arguments: Namespace(config_file='configs/swin_imagenet.py', eval_only=False, fast_dev_run=False, opts=['model.cfg.hidden_dropout_prob=0.1', 'model.cfg.attention_probs_dropout_prob=0.1', 'model.cfg.bias_dropout_fusion=true', 'model.cfg.hidden_layers=12', 'model.cfg.hidden_size=768', 'model.cfg.num_attention_heads=12', 'model.cfg.intermediate_size=3072', 'model.cfg.ffn_hidden_size=3072', 'model.cfg.head_size=64', 'graph.enabled=true', 'train.dist.pipeline_num_layers=12', 'train.train_micro_batch_size=128', 'train.global_batch_size=1024', 'train.dist.tensor_parallel_size=2', 'train.dist.pipeline_parallel_size=2', 'train.amp.enabled=true', 'train.activation_checkpoint.enabled=true', 'train.num_accumulation_steps=8', 'train.evaluation.enabled=false', 'train.train_iter=220', 'train.train_epoch=0', 'train.log_period=100', 'train.zero_optimization.enabled=true', 'train.zero_optimization.stage=2', 'train.load_weight=', 'train.output_dir=test_logs/oneflow-28/NVIDIA_GeForce_RTX_3080_Ti/1ea2bb7/LibAI_swin_imagenet_graph_nl12_nah12_hs768_FP16_actrue_DP1_MP2_PP2_zerotrue_stage2_mbs128_gbs1024_acc8_1n4g'], resume=False)
[03/11 00:22:39 libai]: Contents of args.config_file=configs/swin_imagenet.py:
from libai.config import LazyCall
from .common.models.swin.swin_tiny_patch4_window7_224 import model
from .common.models.graph import graph
from .common.train import train
from .common.optim import optim
from .common.data.imagenet import dataloader

from flowvision.data import Mixup
from flowvision.loss.cross_entropy import SoftTargetCrossEntropy

# Refine data path to imagenet
dataloader.train.dataset[0].root = "/ssd/dataset/ImageNet/extract"
dataloader.test[0].dataset.root = "/ssd/dataset/ImageNet/extract"

# Add Mixup Func
dataloader.train.mixup_func = LazyCall(Mixup)(
    mixup_alpha=0.8,
    cutmix_alpha=1.0,
    prob=1.0,
    switch_prob=0.5,
    mode="batch",
    num_classes=1000,
)

# Refine model cfg for vit training on imagenet
model.cfg.num_classes = 1000
model.cfg.loss_func = SoftTargetCrossEntropy()
# Refine optimizer cfg for vit model
optim.lr = 1e-3
optim.eps = 1e-8
optim.weight_decay = 0.05
optim.params.clip_grad_max_norm = None
optim.params.clip_grad_norm_type = None

# Refine train cfg for vit model
train.train_micro_batch_size = 128
train.test_micro_batch_size = 128
train.train_epoch = 300
train.warmup_ratio = 20 / 300
train.eval_period = 1562
train.log_period = 100

# Scheduler
train.scheduler.warmup_factor = 0.001
train.scheduler.alpha = 0.01
train.scheduler.warmup_method = "linear"

# Set fp16 ON
train.amp.enabled = True

[03/11 00:22:39 libai]: Full config saved to test_logs/oneflow-28/NVIDIA_GeForce_RTX_3080_Ti/1ea2bb7/LibAI_swin_imagenet_graph_nl12_nah12_hs768_FP16_actrue_DP1_MP2_PP2_zerotrue_stage2_mbs128_gbs1024_acc8_1n4g/config.yaml
[03/11 00:22:39 lb.engine.default]: > compiling dataset index builder ...
make: Entering directory '/ssd/home/ouyangyu/libai_week_test/libai/libai/data/data_utils'
make: Nothing to be done for 'default'.
make: Leaving directory '/ssd/home/ouyangyu/libai_week_test/libai/libai/data/data_utils'
[03/11 00:22:39 lb.engine.default]: >>> done with dataset index builder. Compilation time: 0.055 seconds
[03/11 00:22:39 lb.engine.default]: >>> done with compiling. Compilation time: 0.057 seconds
[03/11 00:22:39 lb.engine.default]: Prepare training, validating, testing set
[03/11 00:22:43 lb.engine.default]: Prepare testing set
[03/11 00:22:44 lb.engine.default]: Auto-scaling the config to train.train_iter=220, train.warmup_iter=15
[03/11 00:22:44 libai]: > Start building model...
W20230311 00:22:47.479815 3180410 eager_local_op_interpreter.cpp:256] Casting a local tensor to a global tensor with Broadcast sbp will modify the data of input! If you want to keep the input local tensor unchanged, please set the arg copy to True.
[03/11 00:22:50 lb.engine.default]: Model:
SwinTransformer(
  (patch_embed): PatchEmbed(
    (proj): Conv2d(3, 96, kernel_size=(4, 4), stride=(4, 4))
    (norm): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
  )
  (pos_drop): Dropout(p=0.0, inplace=False)
  (layers): ModuleList(
    (0): BasicLayer(
      (blocks): ModuleList(
        (0): SwinTransformerBlock(
          (norm1): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=96, out_features=288, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=96, out_features=96, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): Identity()
          (norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=96, out_features=384, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=384, out_features=96, bias=True, parallel=row)
          )
        )
        (1): SwinTransformerBlock(
          (norm1): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=96, out_features=288, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=96, out_features=96, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((96,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=96, out_features=384, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=384, out_features=96, bias=True, parallel=row)
          )
        )
      )
      (downsample): PatchMerging(
        (reduction): Linear1D(in_features=384, out_features=192, bias=False, parallel=data)
        (norm): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
      )
    )
    (1): BasicLayer(
      (blocks): ModuleList(
        (0): SwinTransformerBlock(
          (norm1): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=192, out_features=576, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=192, out_features=192, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=192, out_features=768, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=768, out_features=192, bias=True, parallel=row)
          )
        )
        (1): SwinTransformerBlock(
          (norm1): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=192, out_features=576, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=192, out_features=192, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((192,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=192, out_features=768, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=768, out_features=192, bias=True, parallel=row)
          )
        )
      )
      (downsample): PatchMerging(
        (reduction): Linear1D(in_features=768, out_features=384, bias=False, parallel=data)
        (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
      )
    )
    (2): BasicLayer(
      (blocks): ModuleList(
        (0): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
        (1): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
        (2): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
        (3): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
        (4): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
        (5): SwinTransformerBlock(
          (norm1): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=384, out_features=1152, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=384, out_features=384, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=384, out_features=1536, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=1536, out_features=384, bias=True, parallel=row)
          )
        )
      )
      (downsample): PatchMerging(
        (reduction): Linear1D(in_features=1536, out_features=768, bias=False, parallel=data)
        (norm): LayerNorm((1536,), eps=1e-05, elementwise_affine=True)
      )
    )
    (3): BasicLayer(
      (blocks): ModuleList(
        (0): SwinTransformerBlock(
          (norm1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=768, out_features=2304, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=768, out_features=768, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=768, out_features=3072, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=3072, out_features=768, bias=True, parallel=row)
          )
        )
        (1): SwinTransformerBlock(
          (norm1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
          (attn): WindowAttention(
            (qkv): Linear1D(in_features=768, out_features=2304, bias=True, parallel=data)
            (attn_drop): Dropout(p=0.0, inplace=False)
            (proj): Linear1D(in_features=768, out_features=768, bias=True, parallel=data)
            (proj_drop): Dropout(p=0.0, inplace=False)
            (softmax): Softmax(dim=-1)
          )
          (drop_path): DropPath()
          (norm2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
          (mlp): MLP(
            bias_gelu_fusion=True, bias_dropout_fusion=True, dropout=0.0
            (dense_h_to_4h): Linear1D(in_features=768, out_features=3072, bias=True, parallel=col)
            (dense_4h_to_h): Linear1D(in_features=3072, out_features=768, bias=True, parallel=row)
          )
        )
      )
    )
  )
  (norm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
  (avgpool): AdaptiveAvgPool1d()
  (head): Linear1D(in_features=768, out_features=1000, bias=True, parallel=data)
  (loss_func): SoftTargetCrossEntropy()
)
[03/11 00:22:50 libai]: >>> done with building model. Building time: 5.397 seconds
[03/11 00:22:50 lb.engine.trainer]: Starting training from iteration 0
[03/11 00:22:53 lb.models.utils.graph_base]: Start compiling the train graph which may take some time. Please wait for a moment ...
W20230311 00:23:05.712817 3180413 insert_nccl_logical_op_pass.cpp:1088]  In Graph: GraphBase_0 Placement: cuda-@0:0-@1:1 the total_op_num = 1125 and has 2 different nccl stream which is possible to trigger cuda stream kernel launch upper limit. So the nccl logical kernel will from async to sync exec, which may affect performance.
W20230311 00:23:06.133960 3180411 insert_nccl_logical_op_pass.cpp:1088]  In Graph: GraphBase_0 Placement: cuda-@0:0-@1:1 the total_op_num = 1125 and has 2 different nccl stream which is possible to trigger cuda stream kernel launch upper limit. So the nccl logical kernel will from async to sync exec, which may affect performance.
W20230311 00:23:06.332463 3180410 insert_nccl_logical_op_pass.cpp:1088]  In Graph: GraphBase_0 Placement: cuda-@0:0-@1:1 the total_op_num = 1125 and has 2 different nccl stream which is possible to trigger cuda stream kernel launch upper limit. So the nccl logical kernel will from async to sync exec, which may affect performance.
W20230311 00:23:06.404680 3180412 insert_nccl_logical_op_pass.cpp:1088]  In Graph: GraphBase_0 Placement: cuda-@0:0-@1:1 the total_op_num = 1125 and has 2 different nccl stream which is possible to trigger cuda stream kernel launch upper limit. So the nccl logical kernel will from async to sync exec, which may affect performance.
timestamp, name, driver_version, utilization.gpu [%], utilization.memory [%], memory.total [MiB], memory.free [MiB], memory.used [MiB]
2023/03/11 00:30:21.868, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 2895 MiB, 9158 MiB
2023/03/11 00:30:21.869, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 22 %, 0 %, 12288 MiB, 2969 MiB, 9084 MiB
2023/03/11 00:30:21.871, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 8307 MiB, 3746 MiB
2023/03/11 00:30:21.874, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 6 %, 0 %, 12288 MiB, 8315 MiB, 3738 MiB
2023/03/11 00:30:21.875, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.876, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.877, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.879, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
timestamp, name, driver_version, utilization.gpu [%], utilization.memory [%], memory.total [MiB], memory.free [MiB], memory.used [MiB]
2023/03/11 00:30:21.888, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 2895 MiB, 9158 MiB
2023/03/11 00:30:21.889, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 22 %, 0 %, 12288 MiB, 2969 MiB, 9084 MiB
2023/03/11 00:30:21.890, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 42 %, 0 %, 12288 MiB, 8307 MiB, 3746 MiB
2023/03/11 00:30:21.898, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 6 %, 0 %, 12288 MiB, 8315 MiB, 3738 MiB
2023/03/11 00:30:21.900, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.901, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.902, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.903, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
timestamp, name, driver_version, utilization.gpu [%], utilization.memory [%], memory.total [MiB], memory.free [MiB], memory.used [MiB]
2023/03/11 00:30:21.909, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 2895 MiB, 9158 MiB
2023/03/11 00:30:21.910, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 22 %, 0 %, 12288 MiB, 2969 MiB, 9084 MiB
2023/03/11 00:30:21.911, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 42 %, 0 %, 12288 MiB, 8307 MiB, 3746 MiB
2023/03/11 00:30:21.912, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 6 %, 0 %, 12288 MiB, 8315 MiB, 3738 MiB
2023/03/11 00:30:21.913, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.914, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.915, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:21.916, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
timestamp, name, driver_version, utilization.gpu [%], utilization.memory [%], memory.total [MiB], memory.free [MiB], memory.used [MiB]
2023/03/11 00:30:24.852, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 98 %, 54 %, 12288 MiB, 2895 MiB, 9158 MiB
2023/03/11 00:30:24.853, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 68 %, 38 %, 12288 MiB, 2969 MiB, 9084 MiB
2023/03/11 00:30:24.854, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 8307 MiB, 3746 MiB
2023/03/11 00:30:24.855, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 17 %, 5 %, 12288 MiB, 8315 MiB, 3738 MiB
2023/03/11 00:30:24.856, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:24.857, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:24.858, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
2023/03/11 00:30:24.859, NVIDIA GeForce RTX 3080 Ti, 515.65.01, 0 %, 0 %, 12288 MiB, 12051 MiB, 2 MiB
[03/11 00:30:29 lb.utils.events]:  eta: 0:08:50  iteration: 99/220  consumed_samples: 102400  total_loss: 6.918  time: 4.3961 s/iter  data_time: 1.7716 s/iter total_throughput: 232.93 samples/s lr: 5.82e-04  
[03/11 00:37:43 lb.utils.events]:  eta: 0:01:27  iteration: 199/220  consumed_samples: 204800  total_loss: 6.898  time: 4.3695 s/iter  data_time: 1.7027 s/iter total_throughput: 234.35 samples/s lr: 3.21e-05  
[03/11 00:39:07 lb.utils.events]:  eta: 0:00:00  iteration: 219/220  consumed_samples: 225280  total_loss: 6.891  time: 4.3544 s/iter  data_time: 1.6240 s/iter total_throughput: 235.16 samples/s lr: 1.01e-05  
[03/11 00:39:07 lb.engine.hooks]: Overall training speed: 218 iterations in 0:15:49 (4.3545 s / it)
[03/11 00:39:07 lb.engine.hooks]: Total training time: 0:15:49 (0:00:00 on hooks)
*****************************************
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. 
*****************************************
oneflow-version(git_commit)=0.9.1.dev20230309+cu117
oneflow-commit(git_commit)=1ea2bb7
oneflow-libai(git_commit)=50a973dc5de635b8613ad7666c073c763e238850