RuntimeError: NCCL error in: ../torch/lib/c10d/ProcessGroupNCCL.cpp:859, invalid usage, NCCL version

Traceback (most recent call last):
  File "tools/train_net.py", line 209, in <module>
    launch(
  File "/home/e300/code/detectron2/detectron2/engine/launch.py", line 67, in launch
    mp.spawn(
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 247, in spawn
    return start_processes(fn, args, nprocs, join, daemon, start_method='spawn')
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 205, in start_processes
    while not context.join():
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 166, in join
    raise ProcessRaisedException(msg, error_index, failed_process.pid)
torch.multiprocessing.spawn.ProcessRaisedException: 

-- Process 3 terminated with the following error:
Traceback (most recent call last):
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 59, in _wrap
    fn(i, *args)
  File "/home/e300/code/detectron2/detectron2/engine/launch.py", line 108, in _distributed_worker
    raise e
  File "/home/e300/code/detectron2/detectron2/engine/launch.py", line 98, in _distributed_worker
    dist.init_process_group(
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 467, in init_process_group
    barrier()
  File "/home/e300/anaconda3/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 2145, in barrier
    work = _default_pg.barrier()
RuntimeError: NCCL error in: ../torch/lib/c10d/ProcessGroupNCCL.cpp:859, invalid usage, NCCL version 2.7.8
ncclInvalidUsage: This usually reflects invalid usage of NCCL library (such as too many async ops, too many collectives at once, mixing streams in a group, etc).

错误原因：
用了多个gpu，根据自己电脑配置修改下gpu个数即可