Skip to main content

Environment Variables

You can find in the below table the Environment variables that AIchor lets you use and some examples.

"Slugs" are strings without special characters except hyphens - . They are suitable for use in urls, docker images or tags.

Environment Variable NameDescriptionExample value
AICHOR_USER_NAMEExperiment actor name
AICHOR_USER_EMAILExperiment actor email address
AICHOR_PROJECT_IDAIchor project IDe7be88af-7621-4574-aa61-e810679b69d4
AICHOR_ORGANIZATION_NAMECurrent organization nameinstadeep
AICHOR_ORGANIZATION_IDCurrent organization IDorg_wa4GMsQSU1ZrZ921
AICHOR_CLUSTER_NAMECurrent cluster nameAIchor
AICHOR_CLUSTER_IDCurrent cluster ID31ed1de8-c937-40ea-b5c8-d17eecb12ec1
AICHOR_EXPERIMENT_IDAIchor experiment IDe2b53b98-975d-4be7-9f5b-8dfd6b765d32
AICHOR_INPUT_PATHInput bucket paths3://project-name-e7be88af76214574-inputs/
AICHOR_OUTPUT_PATHOutput recommended path (in the output bucket)s3://project-name-e7be88af76214574-outputs/output/e2b53b98-975d-4be7-9f5b-8dfd6b765d32/
AICHOR_TENSORBOARD_PATHTensorboard records recommended path (in the output bucket) only if tensorboad is enabled in your manifest files3://project-name-e7be88af76214574-outputs/output/e2b53b98-975d-4be7-9f5b-8dfd6b765d32/tensorboard/
VCS_TYPEVCS Repository typegitlab, github or bitbucket
VCS_URLUrl of the remote git repository
VCS_REPOSITORYFull repository path, with owner, groups and nameinstadeep/group/subgroup/repo
VCS_REPOSITORY_SLUGSlugified VCS_REPOSITORYinstadeep-group-subgroup-repo
VCS_REPOSITORY_OWNERFirst subpath of the repository url, usually a usernameinstadeep
VCS_REPOSITORY_OWNER_SLUGSlugified VCS_REPOSITORY_OWNERinstadeep
VCS_REPOSITORY_GROUPSIntermediate subpaths of the repository urlgroup/subgroup
VCS_REPOSITORY_GROUPS_SLUGSlugified VCS_REPOSITORY_GROUPSgroup-subgroup
VCS_REPOSITORY_NAMELast subpath of the repository urlrepo
VCS_REPOSITORY_NAME_SLUGSlugified VCS_REPOSITORY_NAMErepo
VCS_REFGit reference of branch or tagrefs/heads/current-branch-name
VCS_REF_SLUGSlugified VCS_REFrefs-heads-current-branch-name
VCS_REF_NAMEName of branch or tagcurrent-branch-name
VCS_REF_NAME_SLUGSlugified VCS_REF_NAMEcurrent-branch-name
VCS_SHAGit commit shaffac537e6cbbf934b08745a378932722df287a53
VCS_SHA_SHORTGit shortened commit shaffac537e
VCS_AUTHOR_NAMEGit commit author name
VCS_AUTHOR_EMAILGit commit author email address
VCS_COMMIT_MESSAGEFull Git commit message
VCS_COMMIT_MESSAGE_SHORTFirst line of the Git commit message
AWS_ACCESS_KEY_IDAWS key for configuring S3 automatically
AWS_SECRET_ACCESS_KEYAWS secret for configuring S3 automatically
CURRENT_POD_IPip of the pod in the cluster10.110.20.26
CPU_REQUESTCPU requested12
NODE_NAMEname of the node on which the pod is runningmy-nvidia-gb200
AWS_REQUEST_CHECKSUM_CALCULATIONSpecify when a checksum is calculated for request payloadswhen_required
AWS_RESPONSE_CHECKSUM_VALIDATIONSpecify when checksum validation is performed for response payloadswhen_required

Additional points

  • You can add env variables via AIchor UI on the project view. Those variables will be available when running an experiment on AIchor:
    1. In your AIchor project, go the “Variables” tab > “Create new variable”
    2. Only the following predefined environment variables can be overwritten: AWS_SECRET_ACCESS_KEY, AWS_ACCESS_KEY_ID, AWS_ENDPOINT_URL. If you attempt to overwrite any other predefined variable, you will receive an error during experiment execution.
  • If you want to use tensorboard, you can save all the logs into the directory specified via AICHOR_TENSORBOARD_PATH env variable and then use the tensorboard's integration from AIchor:
    1. go to your experiment page
    2. "View Tensorboard" button on the top right.