train with gpu but the computer always restarts

[liubamboo]

When I train the ronin_resnet with gpu, the computer always restart Compilation is falling back to object mode WITH looplifting enabled because Function "change_cf" failed type inference due to: No implementation of function Function(<function repeat at 0x7f8817b93378>) found for signature:

repeat(list(array(float32, 1d, A)), int64, axis=Literalint)

There are 2 candidate implementations:

• Of which 2 did not match due to: Overload in function 'np_repeat': File: numba/np/arrayobj.py: Line 1925. With argument(s): '(list(array(float32, 1d, A)), int64, axis=int64)': Rejected as the implementation raised a specific error: TypeError: np_repeat() got an unexpected keyword argument 'axis' raised from /home/lwt/anaconda3/envs/py36-for-ronin/lib/python3.6/site-packages/numba/core/typing/templates.py:710

During: resolving callee type: Function(<function repeat at 0x7f8817b93378>) During: typing of call at /mnt/mdisk/lwt/project/ronin/source/transformations.py (21)

File "transformations.py", line 21: def change_cf(ori, vectors):

if len(ori.shape) == 1:
    ori = np.repeat([ori], vectors.shape[0], axis=0)
    ^

@jit('f8[:,:](f4[:],f4[:, :])') /home/lwt/anaconda3/envs/py36-for-ronin/lib/python3.6/site-packages/numba/core/object_mode_passes.py:178: NumbaWarning: Function "change_cf" was compiled in object mode without forceobj=True.

File "transformations.py", line 10: @jit('f8[:,:](f4[:],f4[:, :])') def change_cf(ori, vectors): ^

state.func_ir.loc)) /home/lwt/anaconda3/envs/py36-for-ronin/lib/python3.6/site-packages/numba/core/object_mode_passes.py:188: NumbaDeprecationWarning: Fall-back from the nopython compilation path to the object mode compilation path has been detected, this is deprecated behaviour.

[Sachini]

Thank you. I have removed the numba signature to be compatiable with latest versions