Build mxnet-tensorrt on jetson but interface changed?

#1

I build mxnet-tensorrt from source(v1.15.x) using USE_TENSORRT=ON flag.

but when I run the script blow:

Execute with TensorRT

print(‘Building TensorRT engine’)
os.environ[‘MXNET_USE_TENSORRT’] = ‘1’
arg_params.update(aux_params)
all_params = dict([(k, v.as_in_context(mx.gpu(0))) for k, v in arg_params.items()])
executor = mx.contrib.tensorrt.tensorrt_bind(sym, ctx=mx.gpu(0), all_params=all_params,
data=batch_shape, grad_req=‘null’, force_rebind=True)

but can’t find tensorrt_bind function.

Did the interface of tensorrt changed?

How can I fix this error?

#2

It seems like the API has changed in current master:

The tensorrt API is being reworked to support more operator and float16 inference in tensorrt.

#3

@ThomasDelteil thank you. I change to the branch where the interface is not changed.

1 Like