White Papers
15 CheXNet β Inference with Nvidia T4 on Dell EMC PowerEdge R7425
Figure 7. Subsequent calls to train(), evaluate(), or predict(). Source [7]
Variable Scope: When building the custom model, itβs important to create it placing the
variables under the same variable scope as the checkpoints; otherwise, the system will
output errors similar to βtensorbatch_normalization/beta is not found in
resnet_v2_imagenet_checkpointβ. Variable scopes allow you to control variable reuse when
calling functions which implicitly create and use variables. They also allow to name the
variables in a hierarchical and understandable way [8].
For evaluation mode:
if mode == tf.estimator.ModeKeys.EVAL:
for i in range(14):
metrics.update({classes[i]: tf.metrics.auc(labels[: i], probs[:, i])})
return tf.estimator.EstimatorSpec(
mode=mode,
loss=loss,
predictions=predictions,
train_op=train_op,
eval_metric_ops=metrics)
For predict mode: