NPUBroadcastGlobalVariablesCallback Constructor

Applicability

Product	Supported (√/x)
Atlas 350 Accelerator Card	√
Atlas A3 training product/Atlas A3 inference product	√
Atlas A2 training product/Atlas A2 inference product	√
Atlas 200I/500 A2 inference product	X
Atlas inference product	X
Atlas training product	√

Description

Broadcasts variables in Keras scenarios to ensure that the initial values of variables on each device are the same in distributed scenarios.

Prototype

class NPUBroadcastGlobalVariablesCallback(BroadcastGlobalVariablesCallbackImpl, keras.callbacks.Callback):
    def __init__(self, root_rank)

Parameters

Parameter	Input/Output	Description
root_rank	Input	Identifies the device whose variables are to be broadcast to other devices.

Returns

An object of the NPUBroadcastGlobalVariablesCallback class

Example

Before porting

callbacks = [hvd.callbacks.BroadcastGlobalVariablesCallback(0)]

import numpy as np
data = np.random.random((1000, 100))
labels = np random.randint(2, size=(1000,1))
model.fit(data, labels, epochs=10, batch_size=32, callbacks=callbacks)

After porting

from npu_bridge.npu_init import *
callbacks = [NPUBroadcastGlobalVariablesCallback(0)]

import numpy as np
data = np.random.random((1000, 100))
labels = np random.randint(2, size=(1000,1))
model.fit(data, labels, epochs=10, batch_size=32, callbacks=callbacks)

Parent topic: npu_bridge.estimator.npu.npu_callbacks