API接口: NeurAI 包#
Subpackages#
Submodules#
- neurai.config.set_platform(platform='cpu')#
Sets the status for JAX platform.
- Parameters:
platform (str, optional) – A string specifying the platform. Default is ‘cpu’.
- Returns:
None
- neurai.config.set_simulate_status(enable_simulate_jit=True, multi_thread_connect=True, dt=0.1, save_ir_collector=False, use_brainmodel=True)#
Sets the status for compiler settings.
- Parameters:
enable_simulate_jit (bool, optional) – A boolean flag indicating if Just-In-Time (JIT) compilation is enabled. Default is True.
multi_thread_connect (bool, optional) – A boolean flag indicating if multi-threading is enabled. Default is False. If set to True, due to the creation of multiple threads, different PRNGKey will be generated when creating synapses, which could potentially lead to variations in the results of multiple simulation runs.
save_ir_collector (bool) – A boolean flag indicating if save the simulate network in ir.
- Returns:
None
- class neurai.monitor.CallbackRecorder(callback=None)#
Bases:
Recorder
Record data using callback
- Parameters:
callback (callable, optional) – Data is processed through callback functions, by default None
- class neurai.monitor.FileRecorder(file_path='./', over_write=False)#
Bases:
Recorder
Record data using file
- class neurai.monitor.Monitor(monitors=None, recorder=None)#
Bases:
object
Abstract base class used to build new monitors.
- Parameters:
recorder (Recorder,) – Record data to different locations.
monitors (List[MonitorConf],) – Configure the parameters or variables that need to be monitored.
- class neurai.monitor.MonitorBS(monitors=None, recorder=None, start=0, stop=inf, interval=1, maxlen=10000, mode='latest')#
Bases:
Monitor
Class for monitoring brain simulation models.
- Parameters:
recorder (Recorder,) – Record data to different locations.
monitors (MonitorConf,) – Configure the parameters or variables that need to be monitored.
start (float, optional) – The start time step for monitoring, by default 0.
stop (float, optional) – The stop time step for monitoring, by default np.inf.
interval (int, optional) – The interval between two successive monitoring, by default 1.
maxlen (int , optional) – The maximum number of monitored data,by default 10000.
mode (str , optional) – The data record mode ,by default ‘oldest’.
- simulate_init(param, sim_t, dt)#
Initializes the monitoring process by creating a dictionary for storing the monitored data.
- step_record(param, mon, step, step_logs=None)#
Records the monitored data at the end of each time step.
- class neurai.monitor.MonitorConf(layer_name=None, var_name=None, grad=None, slices=None)#
Bases:
object
Configures the variables that need to be monitored.
- Parameters:
var_name (str) – Name of the variable to monitor.
layer_name (str, optional) – Optional name of the layer the variable belongs to, by default None
grad (str, optional) – Optional gradient associated with the variable, by default None
slices (range, optional) – Monitor the slice information of variables, by default None