Quickstart

To launch a fault-tolerant job, run the following on all nodes.

shell

torchrun
   --nnodes=NUM_NODES
   --nproc-per-node=TRAINERS_PER_NODE
   --max-restarts=NUM_ALLOWED_FAILURES
   --rdzv-id=JOB_ID
   --rdzv-backend=c10d
   --rdzv-endpoint=HOST_NODE_ADDR
   YOUR_TRAINING_SCRIPT.py (--arg1 ... train script args...)

To launch an elastic job, run the following on at least MIN_SIZE nodes and at most MAX_SIZE nodes.

shell

torchrun
    --nnodes=MIN_SIZE:MAX_SIZE
    --nproc-per-node=TRAINERS_PER_NODE
    --max-restarts=NUM_ALLOWED_FAILURES_OR_MEMBERSHIP_CHANGES
    --rdzv-id=JOB_ID
    --rdzv-backend=c10d
    --rdzv-endpoint=HOST_NODE_ADDR
    YOUR_TRAINING_SCRIPT.py (--arg1 ... train script args...)

{note}

TorchElastic models failures as membership changes. When a node fails,
this is treated as a "scale down" event. When the failed node is replaced by
the scheduler, it is a "scale up" event. Hence for both fault tolerant
and elastic jobs, `--max-restarts` is used to control the total number of
restarts before giving up, regardless of whether the restart was caused
due to a failure or a scaling event.

HOST_NODE_ADDR, in form <host>[:<port>] (e.g. node1.example.com:29400), specifies the node and the port on which the C10d rendezvous backend should be instantiated and hosted. It can be any node in your training cluster, but ideally you should pick a node that has a high bandwidth.

{note}

If no port number is specified `HOST_NODE_ADDR` defaults to 29400.

{note}

The `--standalone` option can be passed to launch a single node job with a
sidecar rendezvous backend. You don't have to pass `--rdzv-id`, `--rdzv-endpoint`, and `--rdzv-backend` when the `--standalone` option
is used.

{note}

Learn more about writing your distributed training script {doc}`here <train_script>`.

If torchrun does not meet your requirements you may use our APIs directly for more powerful customization. Start by taking a look at the {doc}elastic agent <agent> API.