• Hardware:Workstation 2xRTXA6000ada
• Network Type: Detectnet_v2
• TLT Version: TAO5.5.0-api
¿TAO 5.5 API kills/deprecate detectnet_v2?
After fight with the last TAO update to implement and use the new autolabeling functions, find that the rest of the teams lost the possibility to continue using and improving the open projects with the detectnet_v2 networks, based in the TF1 containerd.
Also to mention the documentation about the migration…to cry and lose time.
https://docs.nvidia.com/tao/tao-toolkit/text/migration_guides/migrating_to_tao_5.3_toolkit_apis.html
Where are all the /org/ calls? and the new upload datasets methods?, where is explained if we don’t use the AWS file storage?.. and the workspace new entity?
Any suggestion to support to the detectnet_v2 or downgrade to 5.2?
Try to launch a “convert” action from a new dataset with a “kitti” format, with the detectnet configuration, and the TAO api launch TF2 instead TF1 container… any suggestion?
Thanks in advance.
edit: I forget to mention, that the next references not exist:
This files are not updated to 5.5.
https://docs.nvidia.com/tao/tao-toolkit/text/tao_toolkit_api/api_deployment.html
→ helm fetch https://helm.ngc.nvidia.com/nvidia/tao/charts/tao-api-5.5.0.tgz --username=‘$oauthtoken’ --password=
Not exist, the filename is tao-toolkit-api-5.5.0.tgz
https://docs.nvidia.com/tao/tao-toolkit/text/tao_toolkit_api/api_rest_api.html#examples
->ngc registry resource download-version “nvidia/tao/tao-getting-started:5.5.0”
NOT EXIST!
With that the people can’t work…
5 posts - 2 participants