Extract all tensorboard events files to pandas dataframe

tensorboard_events_to_csv.py

#!/usr/bin/env python3

'''
This script exctracts training variables from all logs from 
tensorflow event files ("event*"), writes them to Pandas 
and finally stores in long-format to a CSV-file including
all (readable) runs of the logging directory.

The magic "5" infers there are only the following v.tags:
[lr, loss, acc, val_loss, val_acc]
'''

import tensorflow as tf
import glob
import os
import pandas as pd


# Get all event* runs from logging_dir subdirectories
logging_dir = './logs'
event_paths = glob.glob(os.path.join(logging_dir, "*","event*"))


# Extraction function
def sum_log(path):
    runlog = pd.DataFrame(columns=['metric', 'value'])
    try:
        for e in tf.train.summary_iterator(path):
            for v in e.summary.value:
                r = {'metric': v.tag, 'value':v.simple_value}
                runlog = runlog.append(r, ignore_index=True)
    
    # Dirty catch of DataLossError
    except:
        print('Event file possibly corrupt: {}'.format(path))
        return None

    runlog['epoch'] = [item for sublist in [[i]*5 for i in range(0, len(runlog)//5)] for item in sublist]
    
    return runlog


# Call & append
all_log = pd.DataFrame()
for path in event_paths:
    log = sum_log(path)
    if log is not None:
        if all_log.shape[0] == 0:
            all_log = log
        else:
            all_log = all_log.append(log)


# Inspect
print(all_log.shape)
all_log.head()    
            
# Store
all_log.to_csv('all_training_logs_in_one_file.csv', index=None)

unfortunately I am not involved in TF and PT related projects anymore, so do not expect any updates on this. Sorry guys.

Feel free to modify the code in every way it becomes useful.

thank you @theRealSuperMario !