11222535141

Committed 07 Oct 2024 07:24PM UTC coverage: 83.924% (+0.009%) from 83.915%

Build # 11222535141

Build Type

push

github

Committed by

web-flow

Commit Message

Feat/emcomposition/add soft max args (#3058)

* • emcomposition.py
  - implement softmax_output_format Parameter
  - docstring edits
  - ALL -> WEIGHTED
  - MAX_INDICATOR -> ARG_MAX
  - more docstring edits
  - add check that ARG_MAX is not used with enable_learning

• test_emcomposition.py
  - add test_softmax_choice

* • project:
  3D -> 3d in docstrings and comments
  Axis -> axis in docstrings and comments

* • readme:
  update of What PsyNeuLink is not

Run Details

11710 of 15064 branches covered (77.73%)

Branch coverage included in aggregate %.

8 of 9 new or added lines in 2 files covered. (88.89%)

27 existing lines in 1 file now uncovered.

32169 of 37220 relevant lines covered (86.43%)

0.86 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

88.16

/psyneulink/library/components/mechanisms/processing/leabramechanism.py

# Princeton University licenses this file to You under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.  You may obtain a copy of the License at:
#     http://www.apache.org/licenses/LICENSE-2.0
# Unless required by applicable law or agreed to in writing, software distributed under the License is distributed
# on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and limitations under the License.

# ********************************************  LeabraMechanism  ******************************************************

"""

Contents
--------

  * `LeabraMechanism_Overview`
  * `LeabraMechanism_Creation`
  * `LeabraMechanism_Structure`
  * `LeabraMechanism_Execution`
  * `LeabraMechanism_Class_Reference`


.. _LeabraMechanism_Overview:

Overview
--------
A LeabraMechanism is a subclass of `ProcessingMechanism` that wraps a leabra network. Leabra is an artificial neural
network algorithm (`O'Reilly, 1996 <ftp://grey.colorado.edu/pub/oreilly/thesis/oreilly_thesis.all.pdf>`_). For more
info about leabra, please see `O'Reilly and Munakata, 2016 <https://grey.colorado.edu/emergent/index.php/Leabra>`_.

.. note::
    The LeabraMechanism uses the leabra Python package, which can be found `here <https://github.com/benureau/leabra>`_
    at Github. While the LeabraMechanism should always match the output of an equivalent network in the leabra package,
    the leabra package itself is still in development, so it is not guaranteed to be correct yet.

.. _LeabraMechanism_Creation:

Creating a LeabraMechanism
--------------------------

A LeabraMechanism can be created in two ways. Users can specify the size of the input layer (**input_size**), size
of the output layer (**output_size**), number of hidden layers (**hidden_layers**), and sizes of the hidden layers
(**hidden_sizes**). In this case, the LeabraMechanism will initialize the connections as uniform random values between
0.55 and 0.95. Alternatively, users can provide a leabra Network object from the leabra package as an argument
(**network**), in which case the **network** will be used as the network wrapped by the LeabraMechanism.
This option requires users to be familiar with the leabra package, but allows more flexibility in specifying parameters.
In the former method of creating a LeabraMechanism, the **training_flag** argument specifies whether the network should
be learning (updating its weights) or not.

.. _LeabraMechanism_Structure:

Structure
---------

The LeabraMechanism has an attribute `training_flag <LeabraMechanism.training_flag>` which can be set to True/False to
determine whether the network is currently learning. The `training_flag <LeabraMechanism.training_flag>` can also be
changed after creation of the LeabraMechanism, causing it to start/stop learning.

.. note::
    If the training_flag is True, the network will learn using the Leabra learning algorithm. Other algorithms may be
    added later.

The LeabraMechanism has two InputPorts: the *MAIN_INPUT* `InputPort` and the *LEARNING_TARGET* InputPort. The
*MAIN_INPUT* InputPort is the input to the leabra network, while the *LEARNING_TARGET* InputPort is the learning
target for the LeabraMechanism. The input to the *MAIN_INPUT* InputPort should have length equal to
`input_size <LeabraMechanism.input_size>` and the input to the *LEARNING_TARGET* InputPort should have length equal to
`output_size <LeabraMechanism.output_size>`.

.. note::
    Currently, there is a bug where LeabraMechanism (and other Mechanisms with multiple input ports) cannot be
    used as `ORIGIN Mechanisms <System_Mechanisms>` for a `System`. If you desire to use a LeabraMechanism as an ORIGIN
    Mechanism, you can work around this bug by creating two `TransferMechanisms <TransferMechanism_Overview>` as ORIGIN
    Mechanisms instead, and have these two TransferMechanisms pass their output to the InputPorts of the
    LeabraMechanism. Here is an example of how to do this. In the example, T2 passes the training_data to the
    *LEARNING_TARGET* InputPort of L (L.input_ports[1])::
        L = LeabraMechanism(input_size=input_size, output_size=output_size)
        T1 = TransferMechanism(name='T1', size=input_size, function=Linear)
        T2 = TransferMechanism(name='T2', size=output_size, function=Linear)
        p1 = Process(pathway=[T1, L])
        proj = MappingProjection(sender=T2, receiver=L.input_ports[1])
        p2 = Process(pathway=[T2, proj, L])
        s = System(processes=[p1, p2])
        s.run(inputs={T1: input_data, T2: training_data})

.. _LeabraMechanism_Execution:

Execution
---------

The LeabraMechanism passes input and training data to the leabra Network it wraps, and the LeabraMechanism passes its
leabra Network's output (after one "trial", default 200 cycles in PsyNeuLink) to its primary `OutputPort`. For details
on Leabra, see `O'Reilly and Munakata, 2016 <https://grey.colorado.edu/emergent/index.php/Leabra>`_ and the
`leabra Python package on Github <https://github.com/benureau/leabra>`_.

.. _LeabraMechanism_Reference:

Class Reference
---------------

"""

import leabra
import numbers

import numpy as np

from psyneulink.core.components.component import ComponentError
from psyneulink.core.components.functions.function import Function_Base
from psyneulink.core.components.mechanisms.processing.processingmechanism import ProcessingMechanism_Base
from psyneulink.core.globals.keywords import LEABRA_FUNCTION, LEABRA_FUNCTION_TYPE, LEABRA_MECHANISM, NETWORK, PREFERENCE_SET_NAME
from psyneulink.core.globals.parameters import FunctionParameter, Parameter, check_user_specified
from psyneulink.core.globals.preferences.basepreferenceset import REPORT_OUTPUT_PREF
from psyneulink.core.globals.preferences.preferenceset import PreferenceEntry, PreferenceLevel
from psyneulink.core.globals.utilities import convert_all_elements_to_np_array
from psyneulink.core.scheduling.time import TimeScale

__all__ = [
    'build_leabra_network', 'convert_to_2d_input', 'input_port_names', 'LeabraError', 'LeabraFunction', 'LeabraMechanism',
    'LEARNING_TARGET', 'MAIN_INPUT', 'MAIN_OUTPUT', 'output_port_name', 'run_leabra_network', 'train_leabra_network',
]

# Used to name input_ports and output_ports:
MAIN_INPUT = 'main_input'
LEARNING_TARGET = 'learning_target'
MAIN_OUTPUT = 'main_output'
input_port_names = [MAIN_INPUT, LEARNING_TARGET]
output_port_name = [MAIN_OUTPUT]


class LeabraError(ComponentError):
    pass


class LeabraFunction(Function_Base):
    """
    LeabraFunction(             \
        default_variable=None,  \
        network=None,           \
        params=None,            \
        owner=None,             \
        prefs=None)

    .. _LeabraFunction:

    LeabraFunction is a custom function that lives inside the LeabraMechanism. As a function, it transforms the
    variable by providing it as input to the leabra network inside the LeabraFunction.

    Arguments
    ---------

    default_variable : number or np.array : default np.zeros() (array of zeros)
        specifies a template for the input to the leabra network.

    network : leabra.Network
        specifies the leabra network to be used.

    params : Dict[param keyword: param value] : default None
        a `parameter dictionary <ParameterPort_Specification>` that specifies the parameters for the
        function.  Values specified for parameters in the dictionary override any assigned to those parameters in
        arguments of the constructor.

    owner : Component
        `component <Component>` to which to assign the Function.

    Attributes
    ----------

    variable : number or np.array
        contains value to be transformed.

    network : leabra.Network
        the leabra network that is being used

    owner : Mechanism
        `component <Component>` to which the Function has been assigned.

    prefs : PreferenceSet or specification dict
        the `PreferenceSet` for the LeabraMechanism; if it is not specified in the **prefs** argument of the
        constructor, a default is assigned using `classPreferences` defined in __init__.py (see `Preferences`
        for details).

    """

    componentType = LEABRA_FUNCTION_TYPE
    componentName = LEABRA_FUNCTION

    classPreferences = {
        PREFERENCE_SET_NAME: 'LeabraFunctionClassPreferences',
        REPORT_OUTPUT_PREF: PreferenceEntry(False, PreferenceLevel.INSTANCE)
    }

    class Parameters(Function_Base.Parameters):
        """
            Attributes
            ----------

                variable
                    see `variable <LeabraFunction.variable>`

                    :default value: numpy.array([[0], [0]])
                    :type: ``numpy.ndarray``
                    :read only: True

                network
                    see `network <LeabraFunction.network>`

                    :default value: None
                    :type:
        """
        variable = Parameter(np.array([[0], [0]]), read_only=True, pnl_internal=True, constructor_argument='default_variable')
        network = None

    @check_user_specified
    def __init__(self,
                 default_variable=None,
                 network=None,
                 params=None,
                 owner=None,
                 prefs=None):

        if network is None:
            raise LeabraError('network was None. Cannot create function for Leabra Mechanism if network is not specified.')

        if default_variable is None:
            input_size = len(network.layers[0].units)
            output_size = len(network.layers[-1].units)
            default_variable = [np.zeros(input_size), np.zeros(output_size)]

        super().__init__(
            default_variable=default_variable,
            network=network,
            params=params,
            owner=owner,
            prefs=prefs,
        )

    def _validate_variable(self, variable, context=None):
        if not isinstance(variable, (list, np.ndarray, numbers.Number)):
            raise LeabraError("Input Error: the input variable ({}) was of type {}, but instead should be a list, "
                              "numpy array, or number.".format(variable, type(variable)))

        input_size = len(self.network.layers[0].units)
        output_size = len(self.network.layers[-1].units)
        if (not hasattr(self, "owner")) or (not hasattr(self.owner, "training_flag")) or self.owner.training_flag is False:
            if len(convert_to_2d_input(variable)[0]) != input_size:
                # convert_to_2d_input(variable[0]) is just in case variable is a 2D array rather than a vector
                raise LeabraError("Input Error: the input was {}, which was of an incompatible length with the "
                                  "input_size, which should be {}.".format(convert_to_2d_input(variable)[0], input_size))
        else:
            if len(convert_to_2d_input(variable)[0]) != input_size or len(convert_to_2d_input(variable)[1]) != output_size:
                raise LeabraError("Input Error: the input variable was {}, which was of an incompatible length with "
                                  "the input_size or output_size, which should be {} and {} respectively.".
                                  format(variable, input_size, output_size))
        return variable

    def _validate_params(self, request_set, target_set=None, context=None):
        if not isinstance(request_set[NETWORK], leabra.Network):
            raise LeabraError("Error: the network given ({}) was of type {}, but instead must be a leabra Network.".
                              format(request_set[NETWORK], type(request_set[NETWORK])))
        super()._validate_params(request_set, target_set, context)

    def _function(self,
                 variable=None,
                 context=None,
                 params=None,
                 ):
        network = self.parameters.network._get(context)
        # HACK: otherwise the INITIALIZING function executions impact the state of the leabra network
        if self.is_initializing:
            output_size = len(network.layers[-1].units)
            return np.zeros(output_size)

        try:
            training_flag = self.owner.parameters.training_flag._get(context)
        except AttributeError:
            training_flag = False

        # None or False
        if not training_flag:
            if isinstance(variable[0], (list, np.ndarray)):
                variable = variable[0]
            return run_leabra_network(network, input_pattern=variable)

        else:
            # variable = convert_to_2d_input(variable)  # FIX: buggy, doesn't handle lists well
            if len(variable) != 2:
                raise LeabraError("Input Error: the input given ({}) for training was not the right format: the input "
                                  "should be a 2D array containing two vectors, corresponding to the input and the "
                                  "training target.".format(variable))
            if len(variable[0]) != len(network.layers[0].units) or len(variable[1]) != len(network.layers[-1].units):
                raise LeabraError("Input Error: the input given ({}) was not the right format: it should be a 2D array "
                                  "containing two vectors, corresponding to the input (which should be length {}) and "
                                  "the training target (which should be length {})".
                                  format(variable, network.layers[0], len(network.layers[-1].units)))
            return train_leabra_network(network, input_pattern=variable[0], output_pattern=variable[1])


def _training_flag_setter(value, self=None, owning_component=None, context=None):
    if value is not self._get(context):
        try:
            set_training(owning_component.parameters.network._get(context), value)
        except AttributeError:
            # do not return None here, or training_flag will always be
            # set to None
            pass

    return value


class LeabraMechanism(ProcessingMechanism_Base):
    """
    LeabraMechanism(                \
    network=None,                   \
    input_size=1,                   \
    output_size=1,                  \
    hidden_layers=0,                \
    hidden_sizes=None,              \
    training_flag=False,            \
    params=None,                    \
    name=None,                      \
    prefs=None)

    Subclass of `ProcessingMechanism` that is a wrapper for a Leabra network in PsyNeuLink.
    See `Mechanism <Mechanism_Class_Reference>` for additional arguments and attributes.

    Arguments
    ---------

    network : Optional[leabra.Network]
        a network object from the leabra package. If specified, the LeabraMechanism's network becomes **network**,
        and the other arguments that specify the network are ignored (**input_size**, **output_size**,
        **hidden_layers**, **hidden_sizes**).

    input_size : int : default 1
        an integer specifying how many units are in (the size of) the first layer (input) of the leabra network.

    output_size : int : default 1
        an integer specifying how many units are in (the size of) the final layer (output) of the leabra network.

    hidden_layers : int : default 0
        an integer specifying how many hidden layers are in the leabra network.

    hidden_sizes : int or List[int] : default input_size
        if specified, this should be a list of integers, specifying the size of each hidden layer. If **hidden_sizes**
        is a list, the number of integers in **hidden_sizes** should be equal to the number of hidden layers. If not
        specified, hidden layers will default to the same size as the input layer. If hidden_sizes is a single integer,
        then all hidden layers are of that size.

    training_flag : boolean : default None
        a boolean specifying whether the leabra network should be learning. If True, the leabra network will adjust
        its weights using the "leabra" algorithm, based on the training pattern (which is read from its second output
        state). The `training_flag` attribute can be changed after initialization, causing the leabra network to
        start/stop learning. If None, `training_flag` will default to False if **network** argument is not provided.
        If **network** argument is provided and `training_flag` is None, then the existing learning rules of the
        **network** will be preserved.

    quarter_size : int : default 50
        an integer specifying how many times the Leabra network cycles each time it is run. Lower values of
        quarter_size result in shorter execution times, though very low values may cause slight fluctuations in output.
        Lower values of quarter_size also effectively reduce the magnitude of learning weight changes during
        a given trial.

    Attributes
    ----------

    function : LeabraFunction
        the function that wraps and executes the leabra mechanism

    input_size : int : default 1
        an integer specifying how many units are in (the size of) the first layer (input) of the leabra network.

    output_size : int : default 1
        an integer specifying how many units are in (the size of) the final layer (output) of the leabra network.

    hidden_layers : int : default 0
        an integer specifying how many hidden layers are in the leabra network.

    hidden_sizes : int or List[int] : default input_size
        an integer or list of integers, specifying the size of each hidden layer.

    training_flag : boolean
        a boolean specifying whether the leabra network should be learning. If True, the leabra network will adjust
        its weights using the "leabra" algorithm, based on the training pattern (which is read from its second output
        state). The `training_flag` attribute can be changed after initialization, causing the leabra network to
        start/stop learning.

    quarter_size : int : default 50
        an integer specifying how many times the Leabra network cycles each time it is run. Lower values of
        quarter_size result in shorter execution times, though very low values may cause slight fluctuations in output.
        Lower values of quarter_size also effectively reduce the magnitude of learning weight changes during
        a given trial.

    network : leabra.Network
        the leabra.Network object which is executed by the LeabraMechanism. For more info about leabra Networks,
        please see the `leabra package <https://github.com/benureau/leabra>` on Github.

    Returns
    -------
    instance of LeabraMechanism : LeabraMechanism
    """

    componentType = LEABRA_MECHANISM

    is_self_learner = True  # CW 11/27/17: a flag; "True" if the mechanism self-learns. Declared in ProcessingMechanism

    classPreferenceLevel = PreferenceLevel.SUBTYPE
    # These will override those specified in TYPE_DEFAULT_PREFERENCES
    classPreferences = {
        PREFERENCE_SET_NAME: 'TransferCustomClassPreferences',
        REPORT_OUTPUT_PREF: PreferenceEntry(False, PreferenceLevel.INSTANCE)
    }

    # LeabraMechanism parameter and control signal assignments):

    class Parameters(ProcessingMechanism_Base.Parameters):
        """
            Attributes
            ----------

                hidden_layers
                    see `hidden_layers <LeabraMechanism.hidden_layers>`

                    :default value: 0
                    :type: ``int``

                hidden_sizes
                    see `hidden_sizes <LeabraMechanism.hidden_sizes>`

                    :default value: None
                    :type:

                input_size
                    see `input_size <LeabraMechanism.input_size>`

                    :default value: 1
                    :type: ``int``

                network
                    see `network <LeabraMechanism.network>`

                    :default value: None
                    :type:

                output_size
                    see `output_size <LeabraMechanism.output_size>`

                    :default value: 1
                    :type: ``int``

                quarter_size
                    see `quarter_size <LeabraMechanism.quarter_size>`

                    :default value: 50
                    :type: ``int``

                training_flag
                    see `training_flag <LeabraMechanism.training_flag>`

                    :default value: None
                    :type:
        """
        input_size = 1
        output_size = 1
        hidden_layers = 0
        hidden_sizes = None
        quarter_size = 50

        function = Parameter(LeabraFunction, stateful=False, loggable=False)

        network = FunctionParameter(None)
        training_flag = Parameter(False, setter=_training_flag_setter, dependencies='network')

    @check_user_specified
    def __init__(self,
                 network=None,
                 input_size=None,
                 output_size=None,
                 hidden_layers=None,
                 hidden_sizes=None,
                 training_flag=None,
                 quarter_size=None,
                 params=None,
                 name=None,
                 prefs=None
    ):

        if network is not None:
            input_size = len(network.layers[0].units)
            output_size = len(network.layers[-1].units)
            hidden_layers = len(network.layers) - 2
            hidden_sizes = list(map(lambda x: len(x.units), network.layers))[1:-2]
            quarter_size = network.spec.quarter_size
            training_flag = infer_training_flag_from_network(network)
        else:
            if hidden_sizes is None:
                hidden_sizes = input_size

            # don't directly assign defaults to their corresponding variable
            # because that may cause their parameter to be incorrectly assigned
            # _user_specified=True
            network = build_leabra_network(
                input_size if input_size is not None else self.class_defaults.input_size,
                output_size if output_size is not None else self.class_defaults.output_size,
                hidden_layers if hidden_layers is not None else self.class_defaults.hidden_layers,
                hidden_sizes if hidden_sizes is not None else self.class_defaults.hidden_sizes,
                training_flag if training_flag is not None else self.class_defaults.training_flag,
                quarter_size if quarter_size is not None else self.class_defaults.quarter_size,
            )

        size = [
            input_size if input_size is not None else self.class_defaults.input_size,
            output_size if output_size is not None else self.class_defaults.output_size
        ]

        super().__init__(
            size=size,
            network=network,
            input_size=input_size,
            output_size=output_size,
            hidden_layers=hidden_layers,
            hidden_sizes=hidden_sizes,
            quarter_size=quarter_size,
            training_flag=training_flag,
            params=params,
            name=name,
            prefs=prefs,
        )

    def _execute(
        self,
        variable=None,
        context=None,
        runtime_params=None,
        time_scale=TimeScale.TRIAL,

    ):

        if runtime_params:
            if "training_flag" in runtime_params.keys():
                self.parameters.training_flag._set(runtime_params["training_flag"], context)
                del runtime_params["training_flag"]

        return super()._execute(
            variable=variable,
            context=context,
            runtime_params=runtime_params,

        )


def convert_to_2d_input(array_like):
    if isinstance(array_like, (np.ndarray, list)):
        if isinstance(array_like[0], (np.ndarray, list)):
            if isinstance(array_like[0][0], (np.ndarray, list)):
                print("array_like ({}) is at least 3d, which may cause conversion errors".format(array_like))
            out = []
            for a in array_like:
                out.append(np.array(a))
            return out
        elif isinstance(array_like[0], numbers.Number):
            return [np.array(array_like)]
    elif isinstance(array_like, numbers.Number):
        return [np.array([array_like])]


def build_leabra_network(n_input, n_output, n_hidden, hidden_sizes=None, training_flag=None, quarter_size=50):

    # specifications
    learning_rule = 'leabra' if training_flag is True else None
    unit_spec = leabra.UnitSpec(adapt_on=True, noisy_act=True)
    layer_spec = leabra.LayerSpec(lay_inhib=True)
    conn_spec = leabra.ConnectionSpec(proj='full', rnd_type='uniform', rnd_mean=0.75, rnd_var=0.2, lrule=learning_rule)

    # input/outputs
    input_layer = leabra.Layer(n_input, spec=layer_spec, unit_spec=unit_spec, name='input_layer')
    output_layer = leabra.Layer(n_output, spec=layer_spec, unit_spec=unit_spec, name='output_layer')

    # creating the required numbers of hidden layers and connections
    layers = [input_layer]
    connections = []
    if isinstance(hidden_sizes, numbers.Number):
        hidden_sizes = [hidden_sizes] * n_hidden
    for i in range(n_hidden):
        if hidden_sizes is not None:
            hidden_size = hidden_sizes[i]
        else:
            hidden_size = n_input
        hidden_layer = leabra.Layer(hidden_size, spec=layer_spec, unit_spec=unit_spec, name='hidden_layer_{}'.format(i))
        hidden_conn = leabra.Connection(layers[-1],  hidden_layer, spec=conn_spec)
        layers.append(hidden_layer)
        connections.append(hidden_conn)

    last_conn = leabra.Connection(layers[-1],  output_layer, spec=conn_spec)
    connections.append(last_conn)
    layers.append(output_layer)

    network_spec = leabra.NetworkSpec(quarter_size=quarter_size)
    network = leabra.Network(layers=layers, connections=connections, spec=network_spec)

    return network


def run_leabra_network(network, input_pattern):
    assert len(network.layers[0].units) == len(input_pattern)

    # check training flag: should be handled earlier, but just in case
    # we temporarily set training false, then turn it on again after we are done
    # TODO: maybe add a warning message here?
    if infer_training_flag_from_network(network):
        set_training(network, False)
        network.set_inputs({'input_layer': input_pattern})
        network.set_outputs({})  # clear network._outputs
        network.trial()
        set_training(network, True)
    else:
        network.set_inputs({'input_layer': input_pattern})
        network.set_outputs({})  # clear network._outputs
        network.trial()
    return convert_all_elements_to_np_array(
        [unit.act_m for unit in network.layers[-1].units]
    )


def train_leabra_network(network, input_pattern, output_pattern):
    assert len(network.layers[0].units) == len(input_pattern)
    assert len(network.layers[-1].units) == len(output_pattern)

    # check training flag: should be handled earlier, but just in case
    # we temporarily set training true, then turn it off again after we are done
    # TODO: maybe add a warning message here?
    if not infer_training_flag_from_network(network):
        set_training(network, True)
        network.set_inputs({'input_layer': input_pattern})
        network.set_outputs({'output_layer': output_pattern})
        network.trial()
        set_training(network, False)
    else:
        network.set_inputs({'input_layer': input_pattern})
        network.set_outputs({'output_layer': output_pattern})
        network.trial()

    return convert_all_elements_to_np_array(
        [unit.act_m for unit in network.layers[-1].units]
    )


# infer whether the network is using the None or 'leabra' training rule
# this currently assumes either all connections or no connections are being trained
def infer_training_flag_from_network(network):
    return False if network.connections[0].spec.lrule is None else True


def set_training(network, val):
    if val is None or val is False:
        for conn in network.connections:
            conn.spec.lrule = None
    else:
        for conn in network.connections:
            conn.spec.lrule = 'leabra'

PrincetonUniversity / PsyNeuLink / 11222535141

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous