lambda-labs-gpu-cloud

To provision on-demand GPU cloud instances for ML training, use the skill to launch single-node or distributed multi-node environments on Lambda Labs infrastructure with NVIDIA GPUs like H100 and A100. It manages the instance lifecycle via API-based calls.

Yes, you can orchestrate multi-node Slurm clusters for distributed training in the cloud. The skill facilitates deploying large-scale model training and fine-tuning environments across multiple nodes on Lambda Labs infrastructure.

Lambda-labs-gpu-cloud supports persistent storage for model checkpoints by utilizing attached filesystems. This maintains datasets and model checkpoints across instance lifecycles even after terminating GPU instances.

You can launch instances with various NVIDIA GPUs for machine learning workloads, specifically H100, A100, and B200 configurations. These are provisioned on-demand for high-performance training or inference tasks.

Yes, you need the lambda-cloud-client dependency to manage instance lifecycles. It provides the programmatic API integration required to provision, configure, and control GPU cloud instances on Lambda Labs.