Saving trend figure

.gitignore

@@ -18,4 +18,4 @@ __pycache__
 
 evaluation_results.txt
 model_out
-
+trend.png

eval_model.py

@@ -2,7 +2,7 @@ import torch
 import sys
 from train_model import MLP, PlantsDataset, test
 from torch.utils.data import DataLoader
-
+import matplotlib.pyplot as plt
 
 def load_model():
     model = MLP()
@@ -15,6 +15,15 @@ def load_dev_dataset(batch_size=64):
     return DataLoader(plant_dev, batch_size=batch_size)
 
 
+def make_plot(values):
+    build_nums = list(range(1, len(values) + 1))
+    plt.xlabel('Build number')
+    plt.ylabel('MSE Loss')
+    plt.plot(build_nums, values, label='Model MSE Loss over builds')
+    plt.legend()
+    plt.savefig('trend.png')
+
+
 def main():
     model = load_model()
     dataloader = load_dev_dataset()
@@ -22,8 +31,13 @@ def main():
     loss_fn = torch.nn.MSELoss()
 
     loss = test(dataloader, model, loss_fn)
-    with open('evaluation_results.txt', 'a+') as f:
+    with open('evaluation_results.txt', 'r+') as f:
+        f.read()
         f.write(f'{str(loss)}\n')
+        f.flush()
+        f.seek(0)
+        values = [float(line) for line in f.readlines() if line]
+        make_plot(values)
 
 
 if __name__ == "__main__":

requirements.txt

@@ -2,3 +2,4 @@ kaggle==1.5.12
 pandas==1.4.1
 torch==1.11.0
 numpy~=1.22.3
+matplotlib==3.5.2