Liam Li

It should work after you make the directory.

can you check you made ./tmp/arch and not /tmp/arch

This code is provided in a repository linked from the README: https://github.com/liamcli/darts_asha Note that you won't be able to get exactly reproducible results for the search phase due to asynchronous...

The random search baseline that takes 4 GPU days is from the original DARTS paper. You can just use the ASHA codebase with no early-stopping to run random search for...

I don't have the capacity now to fix this but will try to get to it once I have time.

1. Our checkpoint also stores information about RNG state and other things needed to resume training whereas the mmdetection checkpoint only has model weights. 2. Where is the 92.2% that...

closing PR since we will support better hf transformers integration with a core API callback soon

Could you update the ptb_final.sh script to use the architecture and hyperparameters you used to get to the 56.3 perplexity reported in the paper? Thanks!

Two flows to build deepspeed and gpt-neox images succeeded in circleci: https://app.circleci.com/pipelines/github/determined-ai/environments?branch=gpt-neox2