Machine Learning Loss

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import LinearSegmentedColormap

np.random.seed(5555)
lr = np.random.lognormal(-5, 1, 6000)
loss = 1 / (1 + 1000*lr) + np.abs(lr - 0.01) * 10 + np.random.exponential(0.1, 6000)

fig, ax = plt.subplots(figsize=(10, 8), facecolor='#0f0a1a')
ax.set_facecolor('#0f0a1a')

colors = ['#0f0a1a', '#2a1a4a', '#4a2a7a', '#6a3aaa', '#8a4ada', '#aa6aff', '#cc9aff', '#eeccff']
cmap = LinearSegmentedColormap.from_list('ml', colors, N=256)

hb = ax.hexbin(np.log10(lr), np.log10(loss), gridsize=30, cmap=cmap, mincnt=1, edgecolors='none')

cbar = plt.colorbar(hb, ax=ax, pad=0.02, shrink=0.8)
cbar.ax.set_facecolor('#0f0a1a')
cbar.outline.set_edgecolor('#4a2a7a')
cbar.ax.tick_params(colors='#aa6aff', labelsize=9)
cbar.set_label('Experiments', color='#aa6aff', fontsize=10)

ax.set_xlabel('Learning Rate (log10)', color='#aa6aff', fontsize=11)
ax.set_ylabel('Final Loss (log10)', color='#aa6aff', fontsize=11)
ax.tick_params(colors='#aa6aff', labelsize=10)
for spine in ax.spines.values():
    spine.set_visible(False)

plt.tight_layout()