archive this plot macro for now

make_loss_dist.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+
+import matplotlib.pyplot as plt
+import pandas as pd
+import numpy as np
+
+df_loss = pd.read_csv('checkpoints/loss-batch.csv', sep=',', header=None)
+loss = df_loss[1].to_numpy()
+print('loss mean = {} median = {}'.format(np.mean(loss), np.median(loss)))
+
+sample_means = []
+for i in range(1000) :
+    sample = np.random.choice(loss, 50, replace=False)
+    sample_means.append(np.mean(sample))
+
+fontsize = 24
+plt.hist(sample_means, bins=100)
+plt.xticks(fontsize=fontsize)
+plt.yticks(fontsize=fontsize)
+plt.xlabel('50 sample mean loss', fontsize=fontsize)
+plt.show()