Hello guys !
I am getting this error, I have checked out all previous answers but none actually helped me.
My environment :
Miniconda3
Gtx 1060
python 3.5
Tensorrt 5.0.2-1+cuda9.0
Cudnn 7
My network is Pix2Pix GAN model, which has just Convolution and Deconvolution layers. Here is my code
import tensorrt as trt
import common
TRT_LOGGER = trt.Logger(trt.Logger.WARNING)
import os
# Now importing onnx Model
model_path = "generator.onnx"
builder = trt.Builder(TRT_LOGGER)
network = builder.create_network()
parser = trt.OnnxParser(network, TRT_LOGGER)
with open(model_path, 'rb') as model:
parser.parse(model.read())
print("Parsing done")
builder.int8_mode = False
builder.max_workspace_size = common.GiB(1)
builder.max_batch_size = 1
builder = trt.Builder(TRT_LOGGER)
engine = builder.build_cuda_engine(network)
# Do inference here.
But the error, is :
[b]
[TensorRT] ERROR: Internal error: could not find any implementation for node (Unnamed Layer* 25) [Deconvolution], try increasing the workspace size with IBuilder::setMaxWorkspaceSize()
[TensorRT] ERROR: …/builder/tacticOptimizer.cpp (1230) - OutOfMemory Error in computeCosts: 0
[/b]
Please help me !
Thanks