diff --git a/.idea/misc.xml b/.idea/misc.xml
index 6d0e4cd..8d1afa2 100644
--- a/.idea/misc.xml
+++ b/.idea/misc.xml
@@ -3,5 +3,5 @@
   <component name="Black">
     <option name="sdkName" value="Python 3.14 (-Comp-551-Assignment-2)" />
   </component>
-  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.14 (-Comp-551-Assignment-2)" project-jdk-type="Python SDK" />
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.14" project-jdk-type="Python SDK" />
 </project>
\ No newline at end of file
diff --git a/effect-of-regularization-on-loss.py b/effect-of-regularization-on-loss.py
deleted file mode 100644
index 426f1b1..0000000
--- a/effect-of-regularization-on-loss.py
+++ /dev/null
@@ -1,72 +0,0 @@
-import numpy as np
-import matplotlib.pyplot as plt
-
-# Generating Synthetic Data
-def generate_linear_data(n):
-    x = np.random.uniform(0, 10, n) # initialize x
-    eps = np.random.normal(0, 1, n) # initialize epsilon
-    y = -3 * x + 8 + 2 * eps # y = −3x + 8 + 2ϵ
-    return x.reshape(-1, 1), y
-
-# Gradient Descent with L1/L2
-def gradient_descent(x, y, lam, reg_type, lr, iters):
-    x_b = np.hstack([np.ones_like(x), x]) # initialize x
-    w = np.zeros(x_b.shape[1]) # initialize weight
-    path = [w.copy()]
-
-    for i in range(iters):
-        pred = x_b @ w # linear regression prediction
-        error = pred - y # error
-        grad = x_b.T @ error / len(y) # gradient formula
-
-        if reg_type == 'l2':
-            grad += lam * w # L2 formula
-        elif reg_type == 'l1':
-            grad += lam * np.sign(w) # L1 formula
-
-        w -= lr * grad # loss calculation
-        path.append(w.copy())
-
-    return w, np.array(path)
-
-# Plotting the loss
-def plot_contour(x, y, reg_type, lam):
-    x_b = np.hstack([np.ones_like(x), x]) # initialize x
-    w0, w1 = np.meshgrid(np.linspace(-10, 10, 100), np.linspace(-10, 10, 100)) # initialize intercept and slope
-    loss = np.zeros_like(w0) # initialize loss
-
-    for i in range(w0.shape[0]):
-        for j in range(w0.shape[1]):
-            w = np.array([w0[i, j], w1[i, j]])
-            error = y - x_b @ w # error
-            mse = np.mean(error ** 2) # mean square error
-            reg = lam * (np.sum(w ** 2) if reg_type == 'l2' else np.sum(np.abs(w))) # regularization
-            loss[i, j] = mse + reg # regularization and mse for the loss
-
-    _, path = gradient_descent(x, y, lam, reg_type, 0.01, 500)
-
-    # plotting the figure
-    plt.figure(figsize=(6, 5))
-    plt.contour(w0, w1, loss, levels=50, cmap='viridis')
-    plt.plot(path[:, 0], path[:, 1], 'ro-', markersize=2, label='Gradient Descent Path')
-    plt.title(f"{reg_type.upper()} Regularization (λ={lam})")
-    plt.xlabel("w0 (intercept)")
-    plt.ylabel("w1 (slope)")
-    plt.grid(True)
-    plt.legend()
-    plt.tight_layout()
-    plt.savefig('results/task4-effect-of-regularization-on-loss-' + reg_type + '-'  + str(lam) + '.png')
-
-if __name__ == "__main__":
-    print("Running Task 4: Effect of L1 and L2 Regularization on Loss Landscape")
-
-    # Generate dataset
-    x, y = generate_linear_data(30)
-
-    # Values of lambda to visualize
-    lambda_values = [0.01, 0.1, 1.0]
-
-    # Plot for both L1 and L2 regularization
-    for reg_type in ['l1', 'l2']:
-        for lam in lambda_values:
-            plot_contour(x, y, reg_type, lam)
\ No newline at end of file
diff --git a/results/task3-bias-decomposition-l1.png b/results/task3-bias-decomposition-l1.png
index 89968bc..b0e229e 100644
Binary files a/results/task3-bias-decomposition-l1.png and b/results/task3-bias-decomposition-l1.png differ
diff --git a/results/task3-bias-decomposition-l2.png b/results/task3-bias-decomposition-l2.png
index dc3eff0..8c55104 100644
Binary files a/results/task3-bias-decomposition-l2.png and b/results/task3-bias-decomposition-l2.png differ
diff --git a/results/task3-train-validation-errors-l1.png b/results/task3-train-validation-errors-l1.png
index b05e7d9..8f9381d 100644
Binary files a/results/task3-train-validation-errors-l1.png and b/results/task3-train-validation-errors-l1.png differ
diff --git a/results/task3-train-validation-errors-l2.png b/results/task3-train-validation-errors-l2.png
index 119b280..afb0d02 100644
Binary files a/results/task3-train-validation-errors-l2.png and b/results/task3-train-validation-errors-l2.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l1-0.01.png b/results/task4-effect-of-regularization-on-loss-l1-0.01.png
index 487af0a..26e2d62 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l1-0.01.png and b/results/task4-effect-of-regularization-on-loss-l1-0.01.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l1-0.1.png b/results/task4-effect-of-regularization-on-loss-l1-0.1.png
index 7481165..095966b 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l1-0.1.png and b/results/task4-effect-of-regularization-on-loss-l1-0.1.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l1-1.0.png b/results/task4-effect-of-regularization-on-loss-l1-1.0.png
index b891e6e..104bee1 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l1-1.0.png and b/results/task4-effect-of-regularization-on-loss-l1-1.0.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l2-0.01.png b/results/task4-effect-of-regularization-on-loss-l2-0.01.png
index 21ac1d5..9fbc76e 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l2-0.01.png and b/results/task4-effect-of-regularization-on-loss-l2-0.01.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l2-0.1.png b/results/task4-effect-of-regularization-on-loss-l2-0.1.png
index 951f90e..f168988 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l2-0.1.png and b/results/task4-effect-of-regularization-on-loss-l2-0.1.png differ
diff --git a/results/task4-effect-of-regularization-on-loss-l2-1.0.png b/results/task4-effect-of-regularization-on-loss-l2-1.0.png
index 1df9703..9fcc59e 100644
Binary files a/results/task4-effect-of-regularization-on-loss-l2-1.0.png and b/results/task4-effect-of-regularization-on-loss-l2-1.0.png differ
diff --git a/linear-regression-w-non-linear-functions.py b/tasks-1-&-2.py
similarity index 100%
rename from linear-regression-w-non-linear-functions.py
rename to tasks-1-&-2.py
diff --git a/regularization-w-cross-validation.py b/tasks-3-&-4.py
similarity index 69%
rename from regularization-w-cross-validation.py
rename to tasks-3-&-4.py
index 8375744..48ab651 100644
--- a/regularization-w-cross-validation.py
+++ b/tasks-3-&-4.py
@@ -137,6 +137,63 @@ def bias_variance_decomp(reg_type, lam_values, num_datasets, N, D):
     plt.grid(True)
     plt.savefig('results/task3-bias-decomposition-' + reg_type + '.png')
 
+# Generating Synthetic Data
+def generate_linear_data(n):
+    x = np.random.uniform(0, 10, n) # initialize x
+    eps = np.random.normal(0, 1, n) # initialize epsilon
+    y = -3 * x + 8 + 2 * eps # y = −3x + 8 + 2ϵ
+    return x.reshape(-1, 1), y
+
+# Gradient Descent with L1/L2
+def gradient_descent(x, y, lam, reg_type, lr, iters):
+    x_b = np.hstack([np.ones_like(x), x]) # initialize x
+    w = np.zeros(x_b.shape[1]) # initialize weight
+    path = [w.copy()]
+
+    for i in range(iters):
+        pred = x_b @ w # linear regression prediction
+        error = pred - y # error
+        grad = x_b.T @ error / len(y) # gradient formula
+
+        if reg_type == 'l2':
+            grad += lam * w # L2 formula
+        elif reg_type == 'l1':
+            grad += lam * np.sign(w) # L1 formula
+
+        w -= lr * grad # loss calculation
+        path.append(w.copy())
+
+    return w, np.array(path)
+
+# Plotting the loss
+def plot_contour(x, y, reg_type, lam):
+    x_b = np.hstack([np.ones_like(x), x]) # initialize x
+    w0, w1 = np.meshgrid(np.linspace(-10, 10, 100), np.linspace(-10, 10, 100)) # initialize intercept and slope
+    loss = np.zeros_like(w0) # initialize loss
+
+    for i in range(w0.shape[0]):
+        for j in range(w0.shape[1]):
+            w = np.array([w0[i, j], w1[i, j]])
+            error = y - x_b @ w # error
+            mse = np.mean(error ** 2) # mean square error
+            reg = lam * (np.sum(w ** 2) if reg_type == 'l2' else np.sum(np.abs(w))) # regularization
+            loss[i, j] = mse + reg # regularization and mse for the loss
+
+    _, path = gradient_descent(x, y, lam, reg_type, 0.01, 500)
+
+    # plotting the figure
+    plt.figure(figsize=(6, 5))
+    plt.contour(w0, w1, loss, levels=50, cmap='viridis')
+    plt.plot(path[:, 0], path[:, 1], 'ro-', markersize=2, label='Gradient Descent Path')
+    plt.title(f"{reg_type.upper()} Regularization (λ={lam})")
+    plt.xlabel("w0 (intercept)")
+    plt.ylabel("w1 (slope)")
+    plt.grid(True)
+    plt.legend()
+    plt.tight_layout()
+    plt.savefig('results/task4-effect-of-regularization-on-loss-' + reg_type + '-'  + str(lam) + '.png')
+
+
 if __name__ == "__main__":
     print("Running Task 3: Regularization with Cross-Validation")
 
@@ -146,4 +203,17 @@ if __name__ == "__main__":
     train_validation_err('l1', lam_values, 50, 20, 45)
 
     bias_variance_decomp('l2', lam_values, 50, 20, 45)
-    bias_variance_decomp('l1', lam_values, 50, 20, 45)
\ No newline at end of file
+    bias_variance_decomp('l1', lam_values, 50, 20, 45)
+
+    print("Running Task 4: Effect of L1 and L2 Regularization on Loss Landscape")
+
+    # Generate dataset
+    x, y = generate_linear_data(30)
+
+    # Values of lambda to visualize
+    lambda_values = [0.01, 0.1, 1.0]
+
+    # Plot for both L1 and L2 regularization
+    for reg_type in ['l1', 'l2']:
+        for lam in lambda_values:
+            plot_contour(x, y, reg_type, lam)
\ No newline at end of file