create_endpoint should have parameters to allow end-user to run models on AWS spot instances. This would make it significantly cheaper and more practical to run models like Cohere Instant Large (cohere-gpt-xlarge) which require ml.p4d.24xlarge instances.
create_endpointshould have parameters to allow end-user to run models on AWS spot instances. This would make it significantly cheaper and more practical to run models like Cohere Instant Large (cohere-gpt-xlarge) which requireml.p4d.24xlargeinstances.