W0325 16:55:16.395000 538513 site-packages/torch/distributed/run.py:793] 
W0325 16:55:16.395000 538513 site-packages/torch/distributed/run.py:793] *****************************************
W0325 16:55:16.395000 538513 site-packages/torch/distributed/run.py:793] 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. 
W0325 16:55:16.395000 538513 site-packages/torch/distributed/run.py:793] *****************************************
时间网格：t_c=0.75, 步数 (1→t_c)=100, (t_c→0)=50
Total number of images that will be sampled: 40192

sampling:   0%|          | 0/157 [00:00<?, ?it/s][rank2]:[W325 16:56:59.533804859 ProcessGroupNCCL.cpp:4115] [PG ID 0 PG GUID 0 Rank 2]  using GPU 2 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect.Specify device_ids in barrier() to force use of a particular device,or call init_process_group() with a device_id.
[rank3]:[W325 16:57:00.799344818 ProcessGroupNCCL.cpp:4115] [PG ID 0 PG GUID 0 Rank 3]  using GPU 3 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect.Specify device_ids in barrier() to force use of a particular device,or call init_process_group() with a device_id.
[rank1]:[W325 16:57:00.847229448 ProcessGroupNCCL.cpp:4115] [PG ID 0 PG GUID 0 Rank 1]  using GPU 1 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect.Specify device_ids in barrier() to force use of a particular device,or call init_process_group() with a device_id.
[rank0]:[W325 16:57:01.326116049 ProcessGroupNCCL.cpp:4115] [PG ID 0 PG GUID 0 Rank 0]  using GPU 0 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect.Specify device_ids in barrier() to force use of a particular device,or call init_process_group() with a device_id.

sampling:   1%|          | 1/157 [00:39<1:43:36, 39.85s/it]
sampling:   1%|▏         | 2/157 [01:17<1:39:23, 38.48s/it]
sampling:   2%|▏         | 3/157 [01:54<1:37:23, 37.94s/it]
sampling:   3%|▎         | 4/157 [02:32<1:36:11, 37.72s/it]
sampling:   3%|▎         | 5/157 [03:09<1:35:18, 37.62s/it]
sampling:   4%|▍         | 6/157 [03:46<1:34:33, 37.57s/it]
sampling:   4%|▍         | 7/157 [04:24<1:33:43, 37.49s/it]
sampling:   5%|▌         | 8/157 [05:01<1:32:54, 37.42s/it]
sampling:   6%|▌         | 9/157 [05:38<1:32:15, 37.40s/it]
sampling:   6%|▋         | 10/157 [06:16<1:31:34, 37.38s/it]
sampling:   7%|▋         | 11/157 [06:53<1:30:56, 37.37s/it]