Random seed

kan/KAN.py

@@ -78,7 +78,7 @@ class KAN(nn.Module):
     '''
 
     def __init__(self, width=None, grid=3, k=3, noise_scale=0.1, scale_base_mu=0.0, scale_base_sigma=1.0, base_fun=torch.nn.SiLU(), symbolic_enabled=True, bias_trainable=False, grid_eps=1.0, grid_range=[-1, 1], sp_trainable=True, sb_trainable=True,
-                 device='cpu', seed=0):
+                 device='cpu', seed=None):
         '''
         initalize a KAN model
 
@@ -123,6 +123,9 @@ def __init__(self, width=None, grid=3, k=3, noise_scale=0.1, scale_base_mu=0.0,
         '''
         super(KAN, self).__init__()
 
+        if seed is None:
+            seed = np.random.randint(0, 1e3)
+
         torch.manual_seed(seed)
         np.random.seed(seed)
         random.seed(seed)

kan/LBFGS.py

@@ -302,8 +302,6 @@ def step(self, closure):
                 and returns the loss.
         """
 
-        torch.manual_seed(0)
-
         assert len(self.param_groups) == 1
 
         # Make sure the closure is always called with grad enabled

kan/MultKAN.py

@@ -23,10 +23,13 @@
 class MultKAN(nn.Module):
 
     # include mult_ops = []
-    def __init__(self, width=None, grid=3, k=3, mult_arity = 2, noise_scale=1.0, scale_base_mu=0.0, scale_base_sigma=1.0, base_fun='silu', symbolic_enabled=True, affine_trainable=False, grid_eps=1.0, grid_range=[-1, 1], sp_trainable=True, sb_trainable=True, device='cpu', seed=0, save_plot_data=True, sparse_init=False, auto_save=True, first_init=True, ckpt_path='./model', state_id=0):
+    def __init__(self, width=None, grid=3, k=3, mult_arity = 2, noise_scale=1.0, scale_base_mu=0.0, scale_base_sigma=1.0, base_fun='silu', symbolic_enabled=True, affine_trainable=False, grid_eps=1.0, grid_range=[-1, 1], sp_trainable=True, sb_trainable=True, device='cpu', seed=None, save_plot_data=True, sparse_init=False, auto_save=True, first_init=True, ckpt_path='./model', state_id=0):
 
         super(MultKAN, self).__init__()
 
+        if seed is None:
+            seed = np.random.randint(0, 1e3)
+
         torch.manual_seed(seed)
         np.random.seed(seed)
         random.seed(seed)

kan/utils.py

@@ -66,7 +66,7 @@ def create_dataset(f,
                    normalize_input=False,
                    normalize_label=False,
                    device='cpu',
-                   seed=0):
+                   seed=None):
     '''
     create dataset
 
@@ -103,6 +103,9 @@ def create_dataset(f,
     torch.Size([100, 2])
     '''
 
+    if seed is None:
+        seed = np.random.randint(0, 1e3)
+
     np.random.seed(seed)
     torch.manual_seed(seed)