.. _save_load:
==================
Saving and Loading
==================
General approach
----------------
Pickling the whole net
^^^^^^^^^^^^^^^^^^^^^^
skorch provides several ways to persist your model. First it is
possible to store the model using Python's :mod:`pickle`
function. This saves the whole model, including hyperparameters. This
is useful when you don't want to initialize your model before loading
its parameters, or when your :class:`.NeuralNet` is part of an sklearn
:class:`~sklearn.pipeline.Pipeline`:
.. code:: python
net = NeuralNet(
module=MyModule,
criterion=torch.nn.NLLLoss,
)
model = Pipeline([
('my-features', get_features()),
('net', net),
])
model.fit(X, y)
# saving
with open('some-file.pkl', 'wb') as f:
pickle.dump(model, f)
# loading
with open('some-file.pkl', 'rb') as f:
model = pickle.load(f)
The disadvantage of pickling is that if your underlying code changes,
unpickling might raise errors. Also, some Python code (e.g. lambda
functions) cannot be pickled.
Pickling specific parts of the net
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
For this reason, we provide a second method for persisting your model.
To use it, call the :func:`~skorch.net.NeuralNet.save_params` and
:func:`~skorch.net.NeuralNet.load_params` method on
:class:`.NeuralNet`. Under the hood, this saves the ``module``\'s
``state_dict``, i.e. only the weights and biases of the ``module``.
This is more robust to changes in the code but requires you to
initialize a :class:`.NeuralNet` to load the parameters again:
.. code:: python
net = NeuralNet(
module=MyModule,
criterion=torch.nn.NLLLoss,
)
model = Pipeline([
('my-features', get_features()),
('net', net),
])
model.fit(X, y)
net.save_params(f_params='some-file.pkl')
new_net = NeuralNet(
module=MyModule,
criterion=torch.nn.NLLLoss,
)
new_net.initialize() # This is important!
new_net.load_params(f_params='some-file.pkl')
In addition to saving the model parameters, the history and optimizer
state can be saved by including the `f_history` and `f_optimizer`
keywords to :func:`~skorch.net.NeuralNet.save_params` and
:func:`~skorch.net.NeuralNet.load_params` on :class:`.NeuralNet`. This
feature can be used to continue training:
.. code:: python
net = NeuralNet(
module=MyModule
criterion=torch.nn.NLLLoss,
)
net.fit(X, y, epochs=2) # Train for 2 epochs
net.save_params(
f_params='model.pkl', f_optimizer='opt.pkl', f_history='history.json')
new_net = NeuralNet(
module=MyModule
criterion=torch.nn.NLLLoss,
)
new_net.initialize() # This is important!
new_net.load_params(
f_params='model.pkl', f_optimizer='opt.pkl', f_history='history.json')
new_net.fit(X, y, epochs=2) # Train for another 2 epochs
.. note:: In order to use this feature, the history
must only contain JSON encodable Python data structures.
Numpy and PyTorch types should not be in the history.
.. note:: :meth:`~skorch.net.NeuralNet.save_params` does not store
learned attributes on the net. E.g.,
:class:`skorch.classifier.NeuralNetClassifier` remembers the
classes it encountered during training in the ``classes_``
attribute. This attribute will be missing after
:meth:`~skorch.net.NeuralNet.load_params`. Therefore, if you need
it, you should :func:`pickle.dump` the whole net.
Using safetensors
^^^^^^^^^^^^^^^^^
skorch also supports storing tensors using the `safetensors
`_ library. There are a few
advantages to using ``safetensors``, which is documented on their website, but
most notably, it is secure as it does not use :mod:`pickle`.
To get started, first, install the library in your virtual environment, if it's
not already installed:
.. code:: bash
python -m pip install safetensors
When using :func:`~skorch.net.NeuralNet.save_params` and
:func:`~skorch.net.NeuralNet.load_params`, under the hood, :func:`torch.save`
and :func:`torch.load` are used, which rely on :mod:`pickle` and are thus
unsafe. If security is a concern (or any of the other advantages of
``safetensors``), you should save parameters using the `use_safetensors=True`
option:
.. code:: python
net = NeuralNet(
module=MyModule
criterion=torch.nn.NLLLoss,
)
net.fit(X, y)
net.save_params(f_params='model.safetensors', use_safetensors=True)
new_net = NeuralNet(
module=MyModule
criterion=torch.nn.NLLLoss,
)
new_net.initialize() # This is important!
new_net.load_params(f_params='model.safetensors', use_safetensors=True)
One disadvantage of using ``safetensors`` is that it can only serialize
``state_dict``\s of torch tensors, nothing else. Therefore, it cannot serialize
components that have other values in the ``state_dict``, e.g. the optimizer. If
you absolutely need to save the optimizer, you need to fall back on pickle, as
described in the previous sections.
Trimming for prediction
-----------------------
If you know that after loading the saved model, it will only be used for
prediction, not for further training, you can get rid of several components of
the net. E.g. since the optimizer, criterion, and callbacks only affect
training, they're not needed anymore and can be removed. skorch provides a
convenience method to achieve this, called :meth:`.trim_for_prediction`.
Using callbacks
---------------
skorch provides :class:`.Checkpoint`, :class:`.TrainEndCheckpoint`,
and :class:`.LoadInitState` callbacks to handle saving and loading
models during training. To demonstrate these features, we generate a
dataset and create a simple module:
.. code:: python
import numpy as np
from sklearn.datasets import make_classification
from torch import nn
X, y = make_classification(1000, 10, n_informative=5, random_state=0)
X = X.astype(np.float32)
y = y.astype(np.int64)
class MyModule(nn.Sequential):
def __init__(self, num_units=10):
super().__init__(
nn.Linear(10, num_units),
nn.ReLU(inplace=True),
nn.Dropout(0.2),
nn.Linear(num_units, 10),
nn.Linear(10, 2),
nn.Softmax(dim=-1)
)
Then we create two different checkpoint callbacks and configure them
to save the model parameters, optimizer, and history into a directory
named ``'exp1'``:
.. code:: python
# First run
from skorch.callbacks import Checkpoint, TrainEndCheckpoint
from skorch import NeuralNetClassifier
cp = Checkpoint(dirname='exp1')
train_end_cp = TrainEndCheckpoint(dirname='exp1')
net = NeuralNetClassifier(
MyModule, lr=0.5, callbacks=[cp, train_end_cp]
)
_ = net.fit(X, y)
# prints
epoch train_loss valid_acc valid_loss cp dur
------- ------------ ----------- ------------ ---- ------
1 0.6200 0.8209 0.4765 + 0.0232
2 0.3644 0.8557 0.3474 + 0.0238
3 0.2875 0.8806 0.3201 + 0.0214
4 0.2514 0.8905 0.3080 + 0.0237
5 0.2333 0.9154 0.2844 + 0.0203
6 0.2177 0.9403 0.2164 + 0.0215
7 0.2194 0.9403 0.2159 + 0.0220
8 0.2027 0.9403 0.2299 0.0202
9 0.1864 0.9254 0.2313 0.0196
10 0.2024 0.9353 0.2333 0.0221
By default, :class:`.Checkpoint` observes ``valid_loss`` metric and
saves the model when the metric improves. This is indicated by the
``+`` mark in the ``cp`` column of the logs.
On our first run, the validation loss did not improve after the 7th
epoch. We can lower the learning rate and continue training from this
checkpoint by using :class:`.LoadInitState`:
.. code:: python
from skorch.callbacks import LoadInitState
cp = Checkpoint(dirname='exp1')
load_state = LoadInitState(cp)
net = NeuralNetClassifier(
MyModule, lr=0.1, callbacks=[cp, load_state]
)
_ = net.fit(X, y)
# prints
epoch train_loss valid_acc valid_loss cp dur
------- ------------ ----------- ------------ ---- ------
8 0.1939 0.9055 0.2626 + 0.0238
9 0.2055 0.9353 0.2031 + 0.0239
10 0.1992 0.9453 0.2101 0.0182
11 0.2033 0.9453 0.1947 + 0.0211
12 0.1825 0.9104 0.2515 0.0185
13 0.2010 0.9453 0.1927 + 0.0187
14 0.1508 0.9453 0.1952 0.0198
15 0.1679 0.9502 0.1905 + 0.0181
16 0.1516 0.9453 0.1864 + 0.0192
17 0.1576 0.9453 0.1804 + 0.0184
The :class:`.LoadInitState` callback is executed once in the beginning
of the training procedure and initializes model, history, and
optimizer parameters from a specified checkpoint (if it exists). In
our case, the previous checkpoint was created at the end of epoch 7,
so the second run resumes from epoch 8. With a lower learning rate,
the validation loss was able to improve!
Notice that in the first run we included a :class:`.TrainEndCheckpoint`
in the list of callbacks. As its name suggests, this callback creates
a checkpoint at the end of training. As before, we can pass it to
:class:`.LoadInitState` to continue training:
.. code:: python
cp_from_final = Checkpoint(dirname='exp1', fn_prefix='from_train_end_')
load_state = LoadInitState(train_end_cp)
net = NeuralNetClassifier(
MyModule, lr=0.1, callbacks=[cp_from_final, load_state]
)
_ = net.fit(X, y)
# prints
epoch train_loss valid_acc valid_loss cp dur
------- ------------ ----------- ------------ ---- ------
11 0.1663 0.9453 0.2166 + 0.0282
12 0.1880 0.9403 0.2237 0.0178
13 0.1813 0.9353 0.1993 + 0.0161
14 0.1744 0.9353 0.1955 + 0.0150
15 0.1538 0.9303 0.2053 0.0077
16 0.1473 0.9403 0.1947 + 0.0078
17 0.1563 0.9254 0.1989 0.0074
18 0.1558 0.9403 0.1877 + 0.0075
19 0.1534 0.9254 0.2318 0.0074
20 0.1779 0.9453 0.1814 + 0.0074
In this run, training started at epoch 11, continuing from the end of
the first run which ended at epoch 10. We created a new
:class:`.Checkpoint` callback with ``fn_prefix`` set to
``'from_train_end_'`` to prefix the saved filenames with
``'from_train_end_'`` to make sure this checkpoint does not override
the checkpoint from the previous run.
Since our ``MyModule`` class allows ``num_units`` to be adjusted, we
can start a new experiment by changing the ``dirname``:
.. code:: python
cp = Checkpoint(dirname='exp2')
load_state = LoadInitState(cp)
net = NeuralNetClassifier(
MyModule, lr=0.5,
callbacks=[cp, load_state],
module__num_units=20,
)
_ = net.fit(X, y)
# prints
epoch train_loss valid_acc valid_loss cp dur
------- ------------ ----------- ------------ ---- ------
1 0.5256 0.8856 0.3624 + 0.0181
2 0.2956 0.8756 0.3416 + 0.0222
3 0.2280 0.9453 0.2299 + 0.0211
4 0.1948 0.9303 0.2136 + 0.0232
5 0.1800 0.9055 0.2696 0.0223
6 0.1605 0.9403 0.1906 + 0.0190
7 0.1594 0.9403 0.2027 0.0184
8 0.1319 0.9303 0.1910 0.0220
9 0.1558 0.9254 0.1923 0.0189
10 0.1432 0.9303 0.2219 0.0192
This stores the model into the ``'exp2'`` directory. Since this is the
first run, the :class:`.LoadInitState` callback does not do anything.
If we were to run the above script again, the :class:`.LoadInitState`
callback will load the model from the checkpoint.
In the run above, the last checkpoint was created at epoch 6, we can
load this checkpoint to predict with it:
.. code:: python
net = NeuralNetClassifier(
MyModule, lr=0.5, module__num_units=20,
)
net.initialize()
net.load_params(checkpoint=cp)
y_pred = net.predict(X)
In this case, it is important to initialize the neural net before
running :meth:`.NeuralNet.load_params`.
In general, all these callbacks also support saving with `safetensors
`_ instead of pickle by passing the
argument ``use_safetensors=True``. The caveat applies that ``safetensors`` can
only serialize torch tensors, hence the optimizer cannot be stored. Therefore,
pass ``f_optimizer=None`` if you want to use ``safetensors``.
Remember that if you want to serialize with ``safetensors``, and you use both
:class:`.Checkpoint` and :class:`.LoadInitState`, both callbacks need to be
initialized with ``use_safetensors=True``.
Saving on Hugging Face Hub
--------------------------
:class:`.Checkpoint` and :class:`.TrainEndCheckpoint` can also be used to store
models on the `Hugging Face Hub `__. For
this to work, instead of indicating a file name for the component to be stored,
use :class:`.skorch.hf.HfHubStorage`.