Python typing.Deque() Examples

def __init__(self, bot: Bot):
        super().__init__()

        self.bot = bot

        # Categories
        self.available_category: discord.CategoryChannel = None
        self.in_use_category: discord.CategoryChannel = None
        self.dormant_category: discord.CategoryChannel = None

        # Queues
        self.channel_queue: asyncio.Queue[discord.TextChannel] = None
        self.name_queue: t.Deque[str] = None

        self.name_positions = self.get_names()
        self.last_notification: t.Optional[datetime] = None

        # Asyncio stuff
        self.queue_tasks: t.List[asyncio.Task] = []
        self.ready = asyncio.Event()
        self.on_message_lock = asyncio.Lock()
        self.init_task = self.bot.loop.create_task(self.init_cog()) 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def backward(ctx: Context,
                 *grad_output: Tensor,
                 ) -> Tuple[Optional[Tensor], ...]:
        prev_stream = ctx.prev_stream
        next_stream = ctx.next_stream

        grad_input: Deque[Tensor] = deque(maxlen=len(grad_output))
        input_stream = current_stream(get_device(prev_stream))

        with use_stream(prev_stream), use_stream(next_stream):
            for x in reversed(grad_output):
                y = x.to(get_device(prev_stream))
                grad_input.appendleft(y)

                # 'next_stream' is not where 'x' has been allocated.
                record_stream(x, next_stream)
                # 'y' has been allocated on 'prev_stream'.
                # It might be used on the current stream captured as 'input_stream'.
                record_stream(y, input_stream)

        grad_streams: Tuple[Optional[Tensor], ...] = (None, None)
        return grad_streams + tuple(grad_input) 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> TensorOrTensors:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        save_rng_states(input[0].device, ctx.rng_states)

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        with torch.no_grad(), enable_checkpointing():
            output = function(input[0] if input_atomic else input)

        return output 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> Tensor:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        return phony 

def backward(ctx: Context,
                 *grad_output: Tensor,
                 ) -> Tuple[Optional[Tensor], ...]:
        prev_stream = ctx.prev_stream
        next_stream = ctx.next_stream

        grad_input: Deque[Tensor] = deque(maxlen=len(grad_output))
        input_stream = current_stream(get_device(prev_stream))

        with use_stream(prev_stream), use_stream(next_stream):
            for x in reversed(grad_output):
                y = x.to(get_device(prev_stream))
                grad_input.appendleft(y)

                # 'next_stream' is not where 'x' has been allocated.
                record_stream(x, next_stream)
                # 'y' has been allocated on 'prev_stream'.
                # It might be used on the current stream captured as 'input_stream'.
                record_stream(y, input_stream)

        grad_streams: Tuple[Optional[Tensor], ...] = (None, None)
        return grad_streams + tuple(grad_input) 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> TensorOrTensors:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        save_rng_states(input[0].device, ctx.rng_states)

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        with torch.no_grad(), enable_checkpointing():
            output = function(input[0] if input_atomic else input)

        return output 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> Tensor:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        return phony 

def backward(ctx: Context,
                 *grad_output: Tensor,
                 ) -> Tuple[Optional[Tensor], ...]:
        prev_stream = ctx.prev_stream
        next_stream = ctx.next_stream

        grad_input: Deque[Tensor] = deque(maxlen=len(grad_output))
        input_stream = current_stream(get_device(prev_stream))

        with use_stream(prev_stream), use_stream(next_stream):
            for x in reversed(grad_output):
                y = x.to(get_device(prev_stream))
                grad_input.appendleft(y)

                # 'next_stream' is not where 'x' has been allocated.
                record_stream(x, next_stream)
                # 'y' has been allocated on 'prev_stream'.
                # It might be used on the current stream captured as 'input_stream'.
                record_stream(y, input_stream)

        grad_streams: Tuple[Optional[Tensor], ...] = (None, None)
        return grad_streams + tuple(grad_input) 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> TensorOrTensors:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        save_rng_states(input[0].device, ctx.rng_states)

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        with torch.no_grad(), enable_checkpointing():
            output = function(input[0] if input_atomic else input)

        return output 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> Tensor:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        return phony 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> TensorOrTensors:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        save_rng_states(input[0].device, ctx.rng_states)

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        with torch.no_grad(), enable_checkpointing():
            output = function(input[0] if input_atomic else input)

        return output 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> Tensor:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        return phony 

def backward(ctx: Context,
                 *grad_output: Tensor,
                 ) -> Tuple[Optional[Tensor], ...]:
        prev_stream = ctx.prev_stream
        next_stream = ctx.next_stream

        grad_input: Deque[Tensor] = deque(maxlen=len(grad_output))
        input_stream = current_stream(get_device(prev_stream))

        with use_stream(prev_stream), use_stream(next_stream):
            for x in reversed(grad_output):
                y = x.to(get_device(prev_stream))
                grad_input.appendleft(y)

                # 'next_stream' is not where 'x' has been allocated.
                record_stream(x, next_stream)
                # 'y' has been allocated on 'prev_stream'.
                # It might be used on the current stream captured as 'input_stream'.
                record_stream(y, input_stream)

        grad_streams: Tuple[Optional[Tensor], ...] = (None, None)
        return grad_streams + tuple(grad_input) 

def save_rng_states(device: torch.device,
                    rng_states: Deque[RNGStates],
                    ) -> None:
    """:meth:`Checkpoint.forward` captures the current PyTorch's random number
    generator states at CPU and GPU to reuse in :meth:`Recompute.backward`.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state = torch.get_rng_state()

    gpu_rng_state: Optional[ByteTensor]
    if device.type == 'cuda':
        gpu_rng_state = torch.cuda.get_rng_state(device)
    else:
        gpu_rng_state = None

    rng_states.append((cpu_rng_state, gpu_rng_state)) 

def restore_rng_states(device: torch.device,
                       rng_states: Deque[RNGStates],
                       ) -> Generator[None, None, None]:
    """:meth:`Recompute.backward` restores the random number generator states
    captured by :func:`save_rng_states` within its context.

    .. seealso:: :ref:`Referential Transparency`

    """
    cpu_rng_state, gpu_rng_state = rng_states.pop()

    gpu_devices: List[torch.device] = []
    if device.type == 'cuda':
        gpu_devices.append(device)

    with torch.random.fork_rng(gpu_devices):
        torch.set_rng_state(cpu_rng_state)
        if gpu_rng_state is not None:
            torch.cuda.set_rng_state(gpu_rng_state, device)
        yield 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> TensorOrTensors:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        save_rng_states(input[0].device, ctx.rng_states)

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        with torch.no_grad(), enable_checkpointing():
            output = function(input[0] if input_atomic else input)

        return output 

def forward(ctx: Context,  # type: ignore
                phony: Tensor,
                recomputed: Deque[Recomputed],
                rng_states: Deque[RNGStates],
                function: Function,
                input_atomic: bool,
                *input: Tensor,
                ) -> Tensor:
        ctx.recomputed = recomputed
        ctx.rng_states = rng_states

        ctx.function = function
        ctx.input_atomic = input_atomic
        ctx.save_for_backward(*input)

        return phony 

def _type_names(
    proto_file: FileDescriptorProto,
    message_type: DescriptorProto,
    parents: Optional[Deque[str]] = None,
) -> Iterator[Tuple[str, str]]:
    if parents is None:
        parents = deque()

    proto_name_parts = ['']
    if proto_file.package:
        proto_name_parts.append(proto_file.package)
    proto_name_parts.extend(parents)
    proto_name_parts.append(message_type.name)

    py_name_parts = [_proto2pb2_module_name(proto_file.name)]
    py_name_parts.extend(parents)
    py_name_parts.append(message_type.name)

    yield '.'.join(proto_name_parts), '.'.join(py_name_parts)

    parents.append(message_type.name)
    for nested in message_type.nested_type:
        yield from _type_names(proto_file, nested, parents=parents)
    parents.pop() 

def reorder_items(items: "Sequence[nodes.Item]") -> "List[nodes.Item]":
    argkeys_cache = {}  # type: Dict[int, Dict[nodes.Item, Dict[_Key, None]]]
    items_by_argkey = {}  # type: Dict[int, Dict[_Key, Deque[nodes.Item]]]
    for scopenum in range(0, scopenum_function):
        d = {}  # type: Dict[nodes.Item, Dict[_Key, None]]
        argkeys_cache[scopenum] = d
        item_d = defaultdict(deque)  # type: Dict[_Key, Deque[nodes.Item]]
        items_by_argkey[scopenum] = item_d
        for item in items:
            # cast is a workaround for https://github.com/python/typeshed/issues/3800.
            keys = cast(
                "Dict[_Key, None]",
                order_preserving_dict.fromkeys(
                    get_parametrized_fixture_keys(item, scopenum), None
                ),
            )
            if keys:
                d[item] = keys
                for key in keys:
                    item_d[key].append(item)
    # cast is a workaround for https://github.com/python/typeshed/issues/3800.
    items_dict = cast(
        "Dict[nodes.Item, None]", order_preserving_dict.fromkeys(items, None)
    )
    return list(reorder_items_atscope(items_dict, argkeys_cache, items_by_argkey, 0)) 

def get_n_step_info_from_demo(
    demo: List, n_step: int, gamma: float
) -> Tuple[List, List]:
    """Return 1 step and n step demos."""
    assert n_step > 1

    demos_1_step = list()
    demos_n_step = list()
    n_step_buffer: Deque = deque(maxlen=n_step)

    for transition in demo:
        n_step_buffer.append(transition)

        if len(n_step_buffer) == n_step:
            # add a single step transition
            demos_1_step.append(n_step_buffer[0])

            # add a multi step transition
            curr_state, action = n_step_buffer[0][:2]
            reward, next_state, done = get_n_step_info(n_step_buffer, gamma)
            transition = (curr_state, action, reward, next_state, done)
            demos_n_step.append(transition)

    return demos_1_step, demos_n_step