Open MMRohe opened 6 years ago
This is the setup in case it helps find the problem Selected GPU[0] Tesla V100-PCIE-16GB as the process wide default device. Build info:
Built time: Dec 5 2017 18:24:40
Last modified date: Mon Nov 13 23:06:21 2017
Build type: Release
Build target: GPU
With 1bit-SGD: no
With ASGD: yes
Math lib: mkl
CUDA version: 9.0.0
CUDNN version: 6.0.21
Build Branch: HEAD
Build SHA1: ce72a557b02b30a52c965a237b124688035f720f
MPI distribution: Microsoft MPI
MPI version: 7.0.12437.6
I came across with the same error when I am trying to run run_fast_rcnn.py example. ANy idea what might be the reason?
I hit the same issue when running the CNTK tutorials, CNTK_103B and CNTK_103C. Crashes when it gets to trainer = C.Trainer(...). 100% repro. CTNK 2.5 CPU only.
Are you using CNTK for Python ? CNTK 2.5 release requires users installing OpenCV, MKL and CUDA. Crashes can happen if they are not properly configured. Check https://docs.microsoft.com/en-us/cognitive-toolkit/Setup-Windows-Python?tabs=cntkpy25
I had encountered the same issue with CPU-Only CNTK2.5, but once MKL was separately installed, this crash was resolved.
Hey,
I am running CNTK on python on windows 10. I try to run a code (that used to run correctly) on a new machine. When the code goes to this line:
trainer = C.Trainer(y, (loss_function, eval_function), learner, progress_writers)
Python crashes with the following message: Python has stopped working A problem caused the program to stop working correctly....
with no further info.
Any idea what that might be ? The code is running correctly on other hardware so I do not really understand. And also why does it stop at this specific line.