jax-ml/jax · error · RuntimeError

Backend {name} already initialized

Error message

Backend {name} already initialized

What it means

Error "Backend {name} already initialized" thrown in jax-ml/jax.

Source

Thrown at jax/_src/xla_bridge.py:307

# Send a PR if you would like to be added.
#
# It is fine for a plugin not to implement every feature that JAX uses, provided
# that a reasonable feature set is implemented and the plugin fails gracefully
# for unimplemented features. Wrong outputs are not acceptable.
_nonexperimental_plugins: set[str] = {'cuda', 'rocm'}

# The set of known experimental plugins that have registrations in JAX codebase.
_experimental_plugins: set[str] = {"oneapi"}

def register_backend_factory(name: str, factory: BackendFactory, *,
                             priority: int = 0,
                             fail_quietly: bool = True,
                             experimental: bool = False,
                             make_topology: TopologyFactory | None = None,
                             c_api: Any | None = None) -> None:
  with _backend_lock:
    if name in _backends:
      raise RuntimeError(f"Backend {name} already initialized")
  _backend_factories[name] = BackendRegistration(
    factory, priority, fail_quietly, experimental, c_api)
  if make_topology is not None:
    _topology_factories[name] = make_topology


def make_cpu_client(
    collectives: _jax.CpuCollectives | None = None,
) -> xla_client.Client:
  """Creates a CPU client with the requested collectives implementation.

  The implementation of CPU collectives used by the client is determined by the
  flag `--jax_cpu_collectives_implementation` - unless `collectives` is
  provided, in which case the flag is overridden and `collectives` is used.

  Args:
    collectives: An optional CPU collectives implementation, used by the client
      if provided.

View on GitHub (pinned to 1e1c6a8fc0)

When it happens

Trigger: Thrown at jax/_src/xla_bridge.py:307 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/d0594bd30c4de243. Report an issue: GitHub.