mxnet.base.MXNetError: [08:13:05] src/ndarray/ndarray.cc:1137: Check failed: to.IsDefaultData()

[Samirakamali]

Hi,
I am trying to feed my tabular dataset to a federated Asynchronous algorithm written by MXNet. but after executing the code I face the following error:

warm up
[Epoch 0] validation: acc-top1=0.631837 acc-top5=0.949816, loss=1.210562, lr=0.100000, rho=0.010000, alpha=0.800000, max_delay=12, mean_delay=0.000000, time=2.378084
Traceback (most recent call last):
File "/content/drive/MyDrive/scada_totorial_FL/data-privacy-main/experiments/async_fl-main/Copy_train__mxnet.py", line 216, in
npx.waitall()
File "/usr/local/lib/python3.10/dist-packages/mxnet/ndarray/ndarray.py", line 200, in waitall
check_call(_LIB.MXNDArrayWaitAll())
File "/usr/local/lib/python3.10/dist-packages/mxnet/base.py", line 255, in check_call
raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: [08:13:05] src/ndarray/ndarray.cc:1137: Check failed: to.IsDefaultData():
Stack trace:
[bt] (0) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(+0x313f4b) [0x7d090db13f4b]
[bt] (1) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(void mxnet::CopyFromToDnsImpl<mshadow::cpu, mshadow::cpu>(mxnet::NDArray const&, mxnet::NDArray const&, mxnet::RunContext)+0xa02) [0x7d0910ebf342]
[bt] (2) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(void mxnet::CopyFromToImpl<mshadow::cpu, mshadow::cpu>(mxnet::NDArray const&, mxnet::NDArray const&, mxnet::RunContext, std::vector<mxnet::Resource, std::allocatormxnet::Resource > const&)+0x6ce) [0x7d0910ec727e]
[bt] (3) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(+0x36c742d) [0x7d0910ec742d]
[bt] (4) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(+0x34ca0d1) [0x7d0910cca0d1]
[bt] (5) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(+0x34cd404) [0x7d0910ccd404]
[bt] (6) /usr/local/lib/python3.10/dist-packages/mxnet/libmxnet.so(+0x34c8884) [0x7d0910cc8884]
[bt] (7) /lib/x86_64-linux-gnu/libstdc++.so.6(+0xdc3ec) [0x7d091a8c83ec]
[bt] (8) /lib/x86_64-linux-gnu/libc.so.6(+0x94b43) [0x7d0921763b43]

if model_name == 'default':
net = gluon.nn.Sequential()
with net.name_scope():
# First convolutional layer
net.add(gluon.nn.Conv1D(channels=8, kernel_size=3, padding= 1, activation='relu'))
net.add(gluon.nn.MaxPool1D(pool_size=2))
net.add(gluon.nn.Conv1D(channels=16, kernel_size=3, padding=1, activation='relu'))
net.add(gluon.nn.AvgPool1D(pool_size=2))
net.add(gluon.nn.Flatten())
net.add(gluon.nn.Dense(classes))
# net.add(gluon.nn.Dropout(rate=0.25))
# net.add(gluon.nn.Dense(classes))
else:
model_kwargs = {'ctx': context, 'pretrained': False, 'classes': classes}
net = get_model(model_name, **model_kwargs)
....
for epoch in range(args.epochs):

# alpha decay
if epoch in alpha_decay_epoch:
    alpha = alpha * args.alpha_decay

# obtain previous model
model_idx = random.randint(0, len(params_prev_list)-1)
params_prev = params_prev_list[model_idx]
for param, param_prev in zip(net.collect_params().values(), params_prev):
    if param.grad_req != 'null':
        weight = param.data()
        weight[:] = param_prev

params_prev = [param.data().copy() for param in net.collect_params().values()]
if rho > 0:
    # param_prev is the initial model, pulled at the beginning of the epoch
    # param_prev is rescaled by rho
    for param, param_prev in zip(net.collect_params().values(), params_prev):
        if param.grad_req != 'null':
            param_prev[:] = param_prev * rho
nd.waitall()

could you possibly help me? I need it urgently.