ION_Archive/COGMOD-HWI
1
0
Fork 0
mirror of https://github.com/ION606/COGMOD-HWI.git synced 2026-05-14 22:16:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

added combined data

Browse Source
This commit is contained in:
ION606
2025-04-24 17:45:54 -04:00
parent 9674db6256
commit f34a4c0c5c
12 changed files with 681 additions and 182 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Project/plot_diagnostics.py
+16 -16
View File
@@ -3,15 +3,16 @@ import json
import pandas as pd
import matplotlib.pyplot as plt
BASE_DIR = "./on_GPU"
# BASE_DIR = "./on_GPU"
# training diagnostics
histories = []
for csv in glob.glob(f"{BASE_DIR}/analytics/history_*.csv"):
df = pd.read_csv(csv) # epochs x metrics
tag = "_".join(csv.split("_")[1:4]) # like sgd_none_42.csv
df['condition'] = tag.replace(".csv", "")
histories.append(df)
for BASE_DIR in ['./on_CPU', './on_GPU']:
for csv in glob.glob(f"{BASE_DIR}/analytics/history_*.csv"):
df = pd.read_csv(csv) # epochs x metrics
tag = "_".join(csv.split("_")[1:4]) # like sgd_none_42.csv
df['condition'] = tag.replace(".csv", "")
histories.append(df)
logs = pd.concat(histories, ignore_index=True)
@@ -20,9 +21,9 @@ mean_log = logs.groupby('epoch').mean(numeric_only=True)
plt.figure()
plt.plot(mean_log.index, mean_log['train_acc'],
color='tab:blue', label='train')
color='#228B22', label='train')
plt.plot(mean_log.index, mean_log['test_acc'],
color='tab:orange', label='validation')
color='#800080', label='validation')
plt.xlabel("epoch")
plt.ylabel("accuracy")
plt.legend()
@@ -31,14 +32,13 @@ ax = plt.gca()
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.tight_layout()
plt.savefig(f"{BASE_DIR}/train_val_accuracy.png")
plt.savefig(f"train_val_accuracy.png")
# Plot train vs validation loss
plt.figure()
plt.plot(mean_log.index, mean_log['train_loss'],
color='tab:blue', label='train')
color='#008080', label='train') # teal
plt.plot(mean_log.index, mean_log['test_loss'],
color='tab:orange', label='validation')
color='#FF00FF', label='validation') # magenta
plt.xlabel("epoch")
plt.ylabel("cross‑entropy loss")
plt.legend()
@@ -47,10 +47,10 @@ ax = plt.gca()
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.tight_layout()
plt.savefig(f"{BASE_DIR}/train_val_loss.png")
plt.savefig(f"train_val_loss.png")
# robustness curves
with open(f"{BASE_DIR}/results.json") as f:
with open(f"./combined_results.json") as f:
res = json.load(f)
df = pd.DataFrame(res)
@@ -68,7 +68,7 @@ for _, row in df.iterrows():
rob_df = pd.DataFrame(records)
pivot = rob_df.groupby(['optimizer', 'sigma']).acc.mean().unstack(0)
ax = pivot.plot(marker='o', linestyle='-', color=['#0072B2', '#D55E00'])
ax = pivot.plot(marker='o', linestyle='-', color=['#FF0000', '#00008B'])
plt.xlabel("Gaussian noise sigma")
plt.ylabel("accuracy")
plt.title("Noise robustness")
@@ -76,5 +76,5 @@ plt.grid(True)
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.tight_layout()
plt.savefig(f"{BASE_DIR}/robustness_curve.png")
plt.savefig(f"./robustness_curve.png")
print("saved robustness_curve.png")