Problem with restoring the saved state and retrain from there in a new script

[LeeStr1]

This is how I am saving the trained states:

net = dde.nn.FNN([2] + [50] * 7 + [4], "swish", "Glorot normal")
model = dde.Model(data, net)
model.compile("adam", lr=0.001)
checkpointer = dde.callbacks.ModelCheckpoint(
    "model/prandtle10", verbose=1, save_better_only=True
)

# Train the model
loss_history, train_state = model.train(epochs=20000, callbacks=[checkpointer])

This is how I am trying to restore the previously trained state in a new code and trying to retrain from there (with changed input parameters):

net = dde.nn.FNN([2] + [50] * 7 + [4], "swish", "Glorot normal")
model = dde.Model(data, net)

model.restore("model/prandtle10-3000.ckpt", verbose=1)

model.compile("adam", lr=0.001)

checkpointer = dde.callbacks.ModelCheckpoint(
    "model/prandtle5", verbose=1, save_better_only=True
)

# Train the model
loss_history, train_state = model.train(epochs=20000, callbacks=[checkpointer])

But, I get following errors:

AttributeError Traceback (most recent call last)
Cell In[10], line 93
90 net = dde.nn.FNN([2] + [50] * 7 + [4], "swish", "Glorot normal")
91 model = dde.Model(data, net)
---> 93 model.restore("model/prandtle10-3000.ckpt", verbose=1)
95 model.compile("adam", lr=0.001)
97 checkpointer = dde.callbacks.ModelCheckpoint(
98 "model/prandtle5", verbose=1, save_better_only=True
99 )

File ~\AppData\Local\Programs\Python\Python310\lib\site-packages\deepxde\model.py:1089, in Model.restore(self, save_path, device, verbose)
1087 print("Restoring model from {} ...\n".format(save_path))
1088 if backend_name == "tensorflow.compat.v1":
-> 1089 self.saver.restore(self.sess, save_path)
1090 elif backend_name == "tensorflow":
1091 self.net.load_weights(save_path)

AttributeError: 'NoneType' object has no attribute 'restore'

[lululxvi]

model.compile("adam", lr=0.001)
model.restore("model/prandtle10-3000.ckpt", verbose=1)