DNNC throws error " terminate called after throwing an instance of 'invalid_random_param' " - deep-learning

Question

I am working on a Xilinx DPU project and was able to train, freeze and quantize my model with the given tools and tensorflow. Now I wanted to use the dnnc-dpu1.4.0 compiler to generate an *.elf file. When executing this command:
dnnc-dpu1.4.0 \
--parser=tensorflow \
--frozen_pb=./quantize_results/deploy_model.pb \
--dpu=1152FA \
--cpu_arch=arm64 \
--output_dir=compile \
--save_kernel \
--mode normal \
--net_name=mnist
it throws this error:
terminate called after throwing an instance of 'invalid_random_param'
what(): Random Param Exception Happened
./compile.sh: line 22: 7602 Aborted (core dumped) dnnc-dpu1.4.0 --parser=tensorflow
--frozen_pb=./quantize_results/deploy_model.pb --dpu=1152FA --cpu_arch=arm64 --output_dir=compile --save_kernel --mode normal --net_name=mnist
I am not sure what to do because this error is very cryptic. It doesn't tell me where those parameters were used or occur, and I won't be able to do something about it. The model works perfectly fine, at least the layers and so on, as I was able to deploy the model itself before. The only thing I changed was the data set and output node count. What could this be? Why is there no extensive error description anywhere?
This is my output for dnnc-dpu1.4.0 --version:
dnnc version v2.05
DPU Target : v1.4.0
Build Label: Jun 24 2019 13:40:08
Copyright #2019 Xilinx Inc. All Rights Reserved.
So I should be up-to-date.
Could anybody help me? This is the last step before I'd be able to deploy my model on the Xilinx DPU. Thank you very much!
Update 1
I posted the same question in the Xilinx Community Forum already.
Update 2
Also, I already found this post on the Xilinx Community Forum, but this does not seem to have a solution to my problem. I can't update the DNNDK, an since I already compiled my model successfully and did no change anything with it, this "solution" seems to be non-applicable to my problem.