TensorBoard EventAccumulator 未显示标量数据，而 TensorBoard UI 正确显示图形

Question

我使用以下代码读取 TensorFlow 生成的事件文件的内容并显示标量数据。该代码能够正确列出所有标签，但不显示这些标签的内容。

import os
import tensorflow as tf
from tensorboard.backend.event_processing.event_accumulator import EventAccumulator

def process_event_file(event_file):
    print(f"Processing event file: {event_file}")
    
    event_acc = EventAccumulator(event_file)
    event_acc.Reload()
    
    print(event_acc.Tags())
    scalar_tags = event_acc.Tags()['scalars']
    for tag in scalar_tags:
        scalar_events = event_acc.Scalars(tag)
        print(f"\nData for tag '{tag}':")
        for event in scalar_events:
            print(f"Step {event.step}, Wall time {event.wall_time}: {event.value}")

def find_and_process_event_files(directory):
    for root, _, files in os.walk(directory):
        for file in files:
            if file.startswith("events.out.tfevents"):
                event_file_path = os.path.join(root, file)
                process_event_file(event_file_path)
                print("\n" + "=" * 80 + "\n")

find_and_process_event_files("/content/drive/MyDrive/logs")

但是，当我在浏览器中运行 TensorBoard 时，所有图表都正确绘制。为什么即使 TensorBoard 正确显示图形，代码也无法打印标量数据的内容？我缺少什么，或者如何解决这个问题？

更新：打印输出看起来

Processing event file: /content/drive/some-folder/logs/2023-04-13 13:52:36.450081/tensorboard/trainer/UNIQUE_ID_HERE/events.out.tfevents.xxxxxx.localhost.localdomain.xxxx.x.v2

{'images': [], 'audio': [], 'histograms': [], 'scalars': [], 'distributions': [], 'tensors': ........}

Answer 1

我也遇到了同样的问题，对于 TensorFlow 2.x 编写的摘要，标量不再位于“scalars”键下，而是位于“tensors”键下（您可以在 Reload( 之后，使用调试器检查 EventAccumulator 对象来看到这一点) )).

您可以通过 .Tensors() 方法检索数据，然后将其传递给 tf.make_ndarray() 以获取标量值。

请参阅 nfelt 的评论：https://github.com/tensorflow/tensorboard/issues/5395

Answer 2

我也遇到过类似的问题。下面的代码可能有效，你可以尝试一下

import os
from tensorboard.backend.event_processing.event_accumulator import EventAccumulator

def process_event_file(event_file):
    print(f"Processing event file: {event_file}")
    
    event_acc = EventAccumulator(event_file)
    event_acc.Reload()
    
    print(event_acc.tensors.Keys())
    tensor_tags = event_acc.tensors.Keys()
    for tag in tensor_tags:
        events = event_acc.tensors.Items(tag)
        print(f"\nData for tag '{tag}':")
        for event in events:
            print(f"Step {event.step}, Wall time {event.wall_time}: {event.tensor_proto}")
            

def find_and_process_event_files(directory):
    for root, _, files in os.walk(directory):
        for file in files:
            if file.startswith("events.out.tfevents"):
                event_file_path = os.path.join(root, file)
                process_event_file(event_file_path)
                print("\n" + "=" * 80 + "\n")

find_and_process_event_files("/content/drive/MyDrive/logs")