You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I saw this error as well on Ubuntu when trying to iterate through forecasts for a DeepAR model.
My gluonts version was 0.4.2 and mxnet was 1.5.1.post0.
I patched it by rolling back mxnet from 1.5.1.post0 to 1.4.1 and pyarrow from 0.16.0 to 0.10.0.
I ran the same code on my macOS without any errors. My hunch is that the difference is my Linux machine has a GPU and has CUDA installed, although the models aren't using the GPU (I see INFO:root:Using CPU).
Update: I have configured Gluon to use the GPU, and I still need mxnet-cu100 1.4.1 and pyarrow 0.10.0.
The problem is with mxnet-1.5.x I believe: there is a bug there that prevents sampling operators from working correctly on Linux, see apache/mxnet#16135
We look forward to mxnet-1.6 being released to solve issue. For now, I think sticking with mxnet-1.4.1 is the solution.
Hello.
When I ran the sample code in this Readme, there was always "MXNetError: vector::_M_range_insert
" errors.
The text was updated successfully, but these errors were encountered: