skypilot-org / skypilot

SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 12+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
https://skypilot.readthedocs.io
Apache License 2.0
6.82k stars 513 forks source link

Add Lambda's GH200 instance type #4377

Closed cbrownstein-lambda closed 5 days ago

cbrownstein-lambda commented 5 days ago

This pull request adds GH200 to the list of instance types offered by Lambda.

Without the change in this pull request, python fetch_lambda_cloud.py --api-key $CLOUD_API_KEY returns:

Traceback (most recent call last):
  File "/home/boxer/devel/skypilot/sky/clouds/service_catalog/data_fetchers/fetch_lambda_cloud.py", line 144, in <module>
    create_catalog(get_api_key(args), 'lambda/vms.csv')
  File "/home/boxer/devel/skypilot/sky/clouds/service_catalog/data_fetchers/fetch_lambda_cloud.py", line 103, in create_catalog
    'SizeInMiB': GPU_TO_MEMORY[gpu]
KeyError: 'GH200'

Closes: https://github.com/skypilot-org/skypilot/issues/4375