Tlt-converter for CUDA 11.1 missing in TLT 3.0

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

In the TLT converter matrix
https://docs.nvidia.com/metropolis/TLT/tlt-user-guide/text/overview.html
the following download link is broken:
11.1/8.0 7.2 cuda111-cudnn80-trt72

I hope someone from NVidia will pick up on this and fix this download - thanks!

2

Thanks for the catching! I am syncing with internal team.
Will update here if I have.

3

Internal team is stilling working on that.

For workaround, you can copy the tlt-converter inside the docker to your own folder.

$ docker run --runtime=nvidia -it -v yourfolder:/tmp/ nvcr.io/nvidia/tlt-streamanalytics:v3.0-dp-py3 /bin/bash
# which tlt-converter
/opt/nvidia/tools/tlt-converter
# cp /opt/nvidia/tools/tlt-converter /tmp/

4

Thanks, found it. It would make sense for me to expose this in the new tlt CLI, eg.

tlt convert ...
5

Please use below command in your host PC.
Firstly, check the container ID, then copy its tlt-converter into your local folder.

$ docker ps
CONTAINER ID IMAGE
cf3ec0545257 nvcr.io/nvidia/tlt-streamanalytics:v3.0-dp-py3

$ docker cp cf3ec0545257:/opt/nvidia/tools/tlt-converter .

6

@radim.tylecek
https://developer.nvidia.com/cuda111-cudnn80-trt72 is available now.