Error caused by parameter num_clients

[ghost]

When I run main_oneDS.py and set the parameter num_clients equal to 10, the program can run successfully, but when I set num_clients equal to 40, the program reports an error.The error is following：

Done
Wrote to ./outputs/seqLen20/oneDS-nonOverlap/IMDB-BINARY-40clients/eps_0.07_0.2/repeats/1_stats_trainData.csv

Done setting up devices.
Self-training ...
  > 0-IMDB-BINARY done.
  > 1-IMDB-BINARY done.
  > 2-IMDB-BINARY done.
  > 3-IMDB-BINARY done.
  > 4-IMDB-BINARY done.
  > 5-IMDB-BINARY done.
  > 6-IMDB-BINARY done.
  > 7-IMDB-BINARY done.
  > 8-IMDB-BINARY done.
  > 9-IMDB-BINARY done.
  > 10-IMDB-BINARY done.
  > 11-IMDB-BINARY done.
  > 12-IMDB-BINARY done.
  > 13-IMDB-BINARY done.
  > 14-IMDB-BINARY done.
  > 15-IMDB-BINARY done.
  > 16-IMDB-BINARY done.
  > 17-IMDB-BINARY done.
  > 18-IMDB-BINARY done.
  > 19-IMDB-BINARY done.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [0,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [1,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [2,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [3,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [4,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [5,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [6,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [7,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [8,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [9,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [10,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [11,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [12,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [13,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [14,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [15,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [16,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [17,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [18,0,0] Assertion `t >= 0 && t < n_classes` failed.
/pytorch/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [19,0,0] Assertion `t >= 0 && t < n_classes` failed.
Traceback (most recent call last):
  File "main_oneDS.py", line 204, in <module>
    process_selftrain(clients=copy.deepcopy(init_clients), server=copy.deepcopy(init_server), local_epoch=50)
  File "main_oneDS.py", line 15, in process_selftrain
    allAccs = run_selftrain_GC(clients, server, local_epoch)
  File "/data/slwang/GCFL-main/training.py", line 13, in run_selftrain_GC
    client.local_train(local_epoch)
  File "/data/slwang/GCFL-main/client.py", line 41, in local_train
    train_stats = train_gc(self.model, self.dataLoader, self.optimizer, local_epoch, self.args.device)
  File "/data/slwang/GCFL-main/client.py", line 136, in train_gc
    loss.backward()
  File "/opt/anaconda3/envs/torch1.9text/lib/python3.7/site-packages/torch/_tensor.py", line 255, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph, inputs=inputs)
  File "/opt/anaconda3/envs/torch1.9text/lib/python3.7/site-packages/torch/autograd/__init__.py", line 149, in backward
    allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag
RuntimeError: CUDA error: device-side assert triggered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

[ZhouAugustus]

Have you found a solution? I also encountered a similar problem.