physicsnemo-shard-tensor

Skill

outshift.io · via agntcy registry Unverified — relayed by outshift.io seen 5h ago

About

Official NVIDIA-authored guidance for PhysicsNeMo ShardTensor domain parallelism — integrate domain parallelism into training/inference scripts (new or existing) with DDP or FSDP2, write and register shard patches to enable new layers/ops, and bootstrap multi-GPU correctness tests. Use when working with ShardTensor, scatter_tensor, domain parallelism, sequence/spatial sharding, ring attention, DeviceMesh + DDP/FSDP2 hybrid parallelism, or physicsnemo.domain_parallel. Do NOT use for generic PyTorch DDP/FSDP setup without domain parallelism, picking a PhysicsNeMo model or example (use physicsnemo-discover), or non-distributed training questions.

Capabilities

The crawler did not record capability metadata for this resource. Inspect the endpoint directly to see what it exposes.

Provenance

Discovered Relayed by agntcy

URN authority urn:air:outshift.io:agntcy:physicsnemo-shard-tensor

Catalog host outshift.io

Anchor check Not anchored

Last crawled seen 5h ago

physicsnemo-shard-tensor

About

Capabilities

Provenance

Tags