Add educational material

Author: Peter Johnson

.DS_Store

@@ -0,0 +1,5 @@
+# Educational Material (for `langModels`)
+
+This folder uses the Evaluation Function to build educational materials to accompany its use. In partilar graphs and plots, and text for lessons using the Function.
+
+This folder is not part of the _Evaluation Function_ (it is not deployed) or part of the docs for the evaluation function.

educational_material/README.md

@@ -0,0 +1,74 @@
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+import torch, argparse
+from pathlib import Path
+from evaluation_function.models.basic_nn import TinyNet, f, train_model, MODEL_PATH
+
+def plot_letter_histogram(show_plots: bool=False, media_dir: Path=None):
+    """Plot a histogram from norvig_letter_single.csv."""
+    csv_path = Path(__file__).parent.parent / "evaluation_function" / "models" / "storage" / "norvig_letter_single.csv"
+    df = pd.read_csv(csv_path)
+
+    df = df.sort_values(by="Percent", ascending=False)
+
+    plt.bar(df["Letter"], df["Percent"], color="skyblue", edgecolor="black")
+    plt.xlabel("Letter")
+    plt.ylabel("Frequency")
+    plt.tight_layout()
+
+    out_path = media_dir / "letter_histogram.png"
+    plt.savefig(out_path, dpi=150, bbox_inches="tight")
+    if show_plots:
+        print(f"Plot saved to {out_path}, displaying plot now.")
+        plt.show()
+    else:
+        print(f"Plot saved to {out_path}.")  
+
+def plot_neural_network_results(show_plots: bool=False, media_dir: Path=None):
+    """Plot the results of a neural network model against the data.
+
+    Args:
+        x (torch.Tensor): Input data.
+        y (torch.Tensor): Target data.
+        model (torch.nn.Module): Trained neural network model.
+    """
+    # Load trained model (or train if needed)
+    model = TinyNet().to(device)
+    model.load_state_dict(torch.load(MODEL_PATH, map_location=device))
+    model.eval()
+
+    # Recreate training data for plotting
+    x = torch.linspace(-2*torch.pi, 2*torch.pi, 200).unsqueeze(1).to(device)
+    y = (f(x) + 0.1*torch.randn_like(x)).to(device)
+
+    with torch.no_grad():
+        # Make domain twice as wide as training range
+        x_plot = torch.linspace(2*x.min().item(), 2*x.max().item(), 800, device=x.device).unsqueeze(1)
+        y_plot = model(x_plot)
+
+        plt.scatter(x.cpu(), y.cpu(), s=10, label="Data")
+        plt.plot(x_plot.cpu(), y_plot.cpu(), color="red", label="Model")
+        plt.legend()
+        out_path = media_dir / "basic_nn_plot.png"
+        plt.savefig(out_path, dpi=150, bbox_inches="tight")  # good web resolution
+        if show_plots:
+            print(f"Plot saved to {out_path}, displaying plot now.")
+            plt.show()
+        else:
+            print(f"Plot saved to {out_path}.")  
+
+
+if __name__ == "__main__":
+    device = torch.device("mps" if torch.backends.mps.is_available() else "cpu")
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--show-plots",
+        action="store_true",
+        help="Display plots interactively instead of just saving them."
+    )
+    args = parser.parse_args()
+    media_dir = Path(__file__).parent / "media"
+    media_dir.mkdir(exist_ok=True)
+    #plot_neural_network_results(show_plots=args.show_plots, media_dir=media_dir)
+    plot_letter_histogram(show_plots=args.show_plots, media_dir=media_dir)

educational_material/__init__.py

@@ -1,7 +1,7 @@
 {
   "basic_nn": {
     "response": "1.0",
-    "answer": "1.0",
+    "answer": "10.0",
     "model": "basic_nn",
     "refresh": false
   },

educational_material/main.py

@@ -19,8 +19,8 @@ priority = "explicit"
 python = ">=3.11,<3.14"
 
 # Uncomment below and comment out cpu if using macOS, the marker feature does not work (at least on my machine) - MM
-#torch = { version = "^2.8.0", markers = "sys_platform != 'linux'" }
-torch = { version = "^2.8.0+cpu", markers = "sys_platform == 'linux'", source = "pytorch-cpu" }
+torch = { version = "^2.8.0", markers = "sys_platform != 'linux'" }
+#torch = { version = "^2.8.0+cpu", markers = "sys_platform == 'linux'", source = "pytorch-cpu" }
 
 typing_extensions = "^4.12.2"
 lf_toolkit = { git = "https://github.com/lambda-feedback/toolkit-python.git", branch = "main", extras = [
@@ -39,6 +39,8 @@ sympy = ">=1.13.3"
 pytest = "^8.2.2"
 flake8 = "^7.1.0"
 nltk = "^3.9.2"
+matplotlib = "^3.8.0"
+pandas = "^2.2.1"
 
 [build-system]
 requires = ["poetry-core"]