api.py

"""Top-level user API."""
from typing import Callable, Optional, Sequence, Union

from jax import linear_util as lu
from jax._src import api
from jax._src.util import HashableFunction
from jax._src.traceback_util import api_boundary
from jax.api_util import (argnums_partial, donation_vector,
                          flatten_fun_nokwargs, rebase_donate_argnums)
from jax.core import AbstractValue
from jax.experimental.maps import FrozenDict
from jax.tree_util import tree_flatten, tree_unflatten, PyTreeDef

from alpa.device_mesh import init_global_cluster, shutdown_global_cluster
from alpa.parallel_method import ParallelMethod, ShardParallel
from alpa.pipeline_parallel.primitive_def import mark_gradient
from alpa.util import (auto_donate_argnums, auto_static_argnums,
                       abstractify_with_aval, GradFuncTransformContext)

is_initialized = False


def init(cluster: str = "ray"):
    """Initialize the global environment.

    Args:
      cluster: The distributed cluster.
        Possible choices: {"local", "ray"}.
        "local" means using all local devices on a single node.
        "ray" means using all devices in a ray cluster.
    """
    global is_initialized

    if is_initialized:
        return
    is_initialized = True

    init_global_cluster(cluster)


def shutdown():
    """Shutdown the global environment."""
    global is_initialized
    assert is_initialized is True
    is_initialized = False
    shutdown_global_cluster()


def parallelize(fun: Optional[Callable] = None,
                *,
                static_argnums: Union[Sequence[int], str] = "auto",
                donate_argnums: Union[Sequence[int], str] = "auto",
                batch_argnums: Union[Sequence[int], str] = (1,),
                method: Optional[ParallelMethod] = None):
    """
    Parallelize a jax function.

    Args:
        fun: The function to be parallelized.
        static_argnums: The same as the static_argnums argument of jax.jit.
          If it is "auto", alpa uses heuristic rules to infer this.
        donate_argnums: The same as the donate_argnums argument of jax.jit.
          If it is "auto", alpa uses heuristic rules to infer this.
        batch_argnums: The indices of arguments that are the data batch.
          This information is used to split the original data batch into micro
          batches to perform gradient accumulation or pipeline parallelism.
          Alpa assumes the 0-th dimension of the tensor is the batch dimension.
        method: The parallelization method.
    """

    def decorate_fun(fun):
        api._check_callable(fun)  # pylint: disable=protected-access
        nonlocal method
        method = method or ShardParallel()
        return ParallelizedFunc(fun, static_argnums, donate_argnums,
                                batch_argnums, method)

    if fun is None:
        return decorate_fun
    return decorate_fun(fun)


class ParallelizedFunc:
    """The function after being transformed by alpa.parallelize."""

    def __init__(
        self,
        fun: Callable,
        static_argnums: Union[Sequence[int], str],
        donate_argnums: Union[Sequence[int], str],
        batch_argnums: Union[Sequence[int], str],
        method: ParallelMethod,
    ):
        self.fun = fun
        self.static_argnums = static_argnums
        self.donate_argnums = donate_argnums
        self.batch_argnums = batch_argnums
        self.method = method

        self.last_executable = None

    @api_boundary
    def __call__(self, *args):
        """Launch the computation on the driver."""
        executable, _, out_tree, args_flat = (
            self._decode_args_and_get_executable(*args))
        out = executable.launch_on_driver(*args_flat)
        return tree_unflatten(out_tree(), out)

    def get_executable(self, *args):
        """Get the compiled exectuable."""
        executable, _, _, _ = self._decode_args_and_get_executable(*args)
        return executable

    def preshard_dynamic_args(self, *args):
        """Shard the dynamic arguments."""
        executable, in_tree, _, args_flat = (
            self._decode_args_and_get_executable(*args))
        sharded_args = executable.preshard_dynamic_args(*args_flat)
        return tree_unflatten(in_tree, sharded_args)

    def get_last_executable(self):
        """Return the last compiled executable for this function."""
        return self.last_executable

    def _decode_args_and_get_executable(self, *args):
        """Flatten PyTree arguments and get the executable."""
        static_argnums, donate_argnums, batch_argnums = (self.static_argnums,
                                                         self.donate_argnums,
                                                         self.batch_argnums)
        kwargs = {}

        f = lu.wrap_init(self.fun)

        # Deal with static arguments and extract dynamic arguments
        if static_argnums == "auto":
            static_argnums = auto_static_argnums(args)

        if static_argnums:
            dyn_argnums = [
                i for i in range(len(args)) if i not in static_argnums
            ]
            # Freeze static dict to make it hashable
            frozen_args = []
            for i, arg in enumerate(args):
                if i in static_argnums and isinstance(arg, dict):
                    frozen_args.append(FrozenDict(arg))
                else:
                    frozen_args.append(arg)
            f, dyn_args = argnums_partial(f, dyn_argnums, frozen_args)
        else:
            dyn_args = args

        # Flatten pytree arguments
        args_flat, in_tree = tree_flatten(dyn_args)
        f, out_tree = flatten_fun_nokwargs(f, in_tree)
        # pylint: disable=unnecessary-lambda
        out_tree_hashable = HashableFunction(lambda: out_tree(), closure=None)

        # Deal with donate argnums
        if donate_argnums == "auto":
            donate_argnums = auto_donate_argnums(args)

        donate_tuple = rebase_donate_argnums(donate_argnums, static_argnums)
        if donate_tuple:
            donated_invars = donation_vector(donate_tuple, dyn_args, kwargs)
        else:
            donated_invars = (False,) * len(args_flat)

        # Deal with batch argnums
        batch_tuple = rebase_donate_argnums(batch_argnums, static_argnums)
        batch_invars = donation_vector(batch_tuple, dyn_args, kwargs)

        # Compile
        abstract_args = map(abstractify_with_aval, args_flat)
        executable = _compile_parallel_executable(f, in_tree, out_tree_hashable,
                                                  static_argnums,
                                                  donated_invars, batch_invars,
                                                  self.method, *abstract_args)

        self.last_executable = executable
        return executable, in_tree, out_tree, args_flat


@lu.cache
def _compile_parallel_executable(
    fun: lu.WrappedFun,
    in_tree: PyTreeDef,
    out_tree_thunk: Callable[[], PyTreeDef],
    static_argnums: Sequence[int],
    donated_invars: Sequence[bool],
    batch_invars: Sequence[bool],
    method: ParallelMethod,
    *avals: Sequence[AbstractValue],
):
    """Cached parallelized callable."""
    # Clean stores for the next call
    for store in fun.stores:
        if store:
            store.reset()

    # Compile a callable
    return method.compile_executable(fun, in_tree, out_tree_thunk,
                                     static_argnums, donated_invars,
                                     batch_invars, *avals)


def clear_executable_cache():
    """Clear all cached executables."""
    _compile_parallel_executable.cache_clear()


def grad(*args, **kwargs):
    """The same as jax.grad, but inserts a gradient marker after the gradient
    computation.

    This function annotates all gradient tensors. This information is used to
    perform gradient accumulation transformation.
    If any auxiliary tensors are returned, they are averaged over mini batches
    in the same way as how the gradients are averaged.
    """

    def ret(*call_args, **call_kwargs):
        # Apply transformations (e.g., layer construction, rematerialization)
        # to the forward func
        arg_list = list(args)
        for transform in GradFuncTransformContext.transforms:
            arg_list[0] = transform(arg_list[0])
        grad_func = api.grad(*arg_list, **kwargs)

        grads = grad_func(*call_args, **call_kwargs)
        return mark_gradient(grads)

    return ret


def value_and_grad(*args, **kwargs):
    """The same as jax.value_and_grad, but inserts a gradient marker after the
    gradient computation.

    This function annotates all gradient tensors. This information is used to
    perform gradient accumulation transformation.
    If any auxiliary tensors are returned, they are averaged over mini batches
    in the same way as how the gradients are averaged.
    """

    def ret(*call_args, **call_kwargs):
        # Apply transformations (e.g., layer construction, rematerialization)
        # to the forward func
        arg_list = list(args)
        for transform in GradFuncTransformContext.transforms:
            arg_list[0] = transform(arg_list[0])
        grad_func = api.value_and_grad(*arg_list, **kwargs)

        val, grads = grad_func(*call_args, **call_kwargs)
        return mark_gradient((val, grads))

    return ret