How to plot accuracy and loss curve?

Every time you run $caffe_root/build/tools/caffe train, the log file will be stored in /tmp/. The file name format is caffe.$ServerName.$UserName.log.INFO.$Date-$Time.$Pid. We can pass the name of the log file we want to plot as the first and the only argument to the python script below. This python script will produce a png file with the same name as the log file.

import numpy as np
import re
import click
import matplotlib
matplotlib.use('Agg')
from matplotlib import pylab as plt
import os

@click.command()
@click.argument('files', nargs=-1, type=click.Path(exists=True))
def main(files):
    plt.style.use('ggplot')
    fig, ax1 = plt.subplots()
    ax2 = ax1.twinx()
    ax1.set_xlabel('iteration')
    ax1.set_ylabel('loss')
    ax2.set_ylabel('accuracy %')
    for i, log_file in enumerate(files):
        loss_iterations, losses, accuracy_iterations, accuracies, accuracies_iteration_checkpoints_ind = parse_log(log_file)
        disp_results(fig, ax1, ax2, loss_iterations, losses, accuracy_iterations, accuracies, accuracies_iteration_checkpoints_ind, color_ind=i)
    #plt.show()
    fileName = os.path.split(files[0])[1]
    plt.savefig(fileName + '.png', bbox_inches='tight')

def parse_log(log_file):
    with open(log_file, 'r') as log_file:
        log = log_file.read()

    loss_pattern = r"Iteration (?P<iter_num>\d+), loss = (?P<loss_val>[+-]?(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?)"
    losses = []
    loss_iterations = []

    for r in re.findall(loss_pattern, log):
        loss_iterations.append(int(r[0]))
        losses.append(float(r[1]))

    loss_iterations = np.array(loss_iterations)
    losses = np.array(losses)

    accuracy_pattern = r"Iteration (?P<iter_num>\d+), Testing net \(#0\)\n.* [aA]ccuracy.* = (?P<accuracy>[+-]?(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?)"
    accuracies = []
    accuracy_iterations = []
    accuracies_iteration_checkpoints_ind = []

    for r in re.findall(accuracy_pattern, log):
        iteration = int(r[0])
        accuracy = float(r[1]) * 100

        if iteration % 10000 == 0 and iteration > 0:
            accuracies_iteration_checkpoints_ind.append(len(accuracy_iterations))

        accuracy_iterations.append(iteration)
        accuracies.append(accuracy)

    accuracy_iterations = np.array(accuracy_iterations)
    accuracies = np.array(accuracies)

    return loss_iterations, losses, accuracy_iterations, accuracies, accuracies_iteration_checkpoints_ind


def disp_results(fig, ax1, ax2, loss_iterations, losses, accuracy_iterations, accuracies, accuracies_iteration_checkpoints_ind, color_ind=0):
    modula = len(plt.rcParams['axes.color_cycle'])
    ax1.plot(loss_iterations, losses, color=plt.rcParams['axes.color_cycle'][(color_ind * 2 + 0) % modula])
    ax2.plot(accuracy_iterations, accuracies, plt.rcParams['axes.color_cycle'][(color_ind * 2 + 1) % modula])
    ax2.plot(accuracy_iterations[accuracies_iteration_checkpoints_ind], accuracies[accuracies_iteration_checkpoints_ind], 'o', color=plt.rcParams['axes.color_cycle'][(color_ind * 2 + 1) % modula])


if __name__ == '__main__':
    main()