16502811447

Committed 24 Jul 2025 04:44PM UTC coverage: 97.767% (+0.1%) from 97.651%

Build # 16502811447

Build Type

push

github

Committed by

web-flow

Commit Message

New version 1.5.0

This pull request introduces version 1.5.0 of **nnodely**, featuring several updates:
1. Improved clarity of documentation and examples.
2. Support for managing multi-dataset features is now available.
3. DataFrames can now be used to create datasets.
4. Datasets can now be resampled.
5. Random data training has been fixed for both classic and recurrent training.
6. The `state` variable has been removed.
7. It is now possible to add or remove a connection or a closed loop.
8. Partial models can now be exported.
9. The `train` function and the result analysis have been separated.
10. A new function, `trainAndAnalyse`, is now available.
11. The report now works across all network types.
12. The training function code has been reorganized.

Run Details

2901 of 2967 new or added lines in 53 files covered. (97.78%)

16 existing lines in 6 files now uncovered.

12652 of 12941 relevant lines covered (97.77%)

0.98 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

87.84

/nnodely/basic/optimizer.py

import copy
import torch

from nnodely.support.utils import check

class Optimizer:
    """
    Represents an optimizer for training neural network models.

    Parameters
    ----------
    name : str
        The name of the optimizer.
    optimizer_defaults : dict, optional
        A dictionary of default optimizer settings.
    optimizer_params : list, optional
        A list of parameter groups for the optimizer.

    Attributes
    ----------
    name : str
        The name of the optimizer.
    optimizer_defaults : dict
        A dictionary of default optimizer settings.
    optimizer_params : list
        A list of parameter groups for the optimizer.
    all_params : dict or None
        A dictionary of all parameters in the model.
    params_to_train : list or None
        A list of parameters to be trained.
    """
    def __init__(self, name, optimizer_defaults = {}, optimizer_params = []):
        """
        Initializes the Optimizer object.

        Parameters
        ----------
        name : str
            The name of the optimizer.
        optimizer_defaults : dict, optional
            A dictionary of default optimizer settings. Default is an empty dictionary.
        optimizer_params : list, optional
            A list of parameter groups for the optimizer. Default is an empty list.
        """
        self.name = name
        self.optimizer_defaults = copy.deepcopy(optimizer_defaults)
        self.optimizer_params = self.unfold(copy.deepcopy(optimizer_params))
        self.all_params = None
        self.params_to_train = None

    def set_params_to_train(self, all_params, params_to_train):
        """
        Sets the parameters to be trained by the optimizer.

        Parameters
        ----------
        all_params : dict
            A dictionary of all parameters in the model.
        params_to_train : list
            A list of parameters to be trained.
        """
        self.all_params = all_params
        self.params_to_train = params_to_train
        if self.optimizer_params == []:
            for param_name in self.all_params.keys():
                if param_name in self.params_to_train:
                    self.optimizer_params.append({'params': param_name})
                else:
                    self.optimizer_params.append({'params': param_name, 'lr': 0.0})

    def set_defaults(self, optimizer_defaults):
        """
        Sets the default optimizer settings.

        Parameters
        ----------
        optimizer_defaults : dict
            A dictionary of default optimizer settings.
        """
        self.optimizer_defaults = copy.deepcopy(optimizer_defaults)

    def set_params(self, optimizer_params):
        """
        Sets the parameter groups for the optimizer.

        Parameters
        ----------
        optimizer_params : list
            A list of parameter groups for the optimizer.
        """
        self.optimizer_params = self.unfold(optimizer_params)

    def unfold(self, params):
        """
        Unfolds the parameter groups into a flat list.

        Parameters
        ----------
        params : list
            A list of parameter groups.

        Returns
        -------
        list
            A flat list of parameter groups.

        Raises
        ------
        KeyError
            If the params argument is not a list.
        """
        optimizer_params = []
        check(type(params) is list, KeyError, f'The params {params} must be a list')
        for param in params:
            if type(param['params']) is list:
                par_copy = copy.deepcopy(param)
                del par_copy['params']
                for par in param['params']:
                    optimizer_params.append({'params':par}|par_copy)
            else:
                optimizer_params.append(param)
        return optimizer_params

    def add_defaults(self, option_name, params, overwrite = True):
        """
        Adds default settings to the optimizer.

        Parameters
        ----------
        option_name : str
            The name of the option to add.
        params : any
            The parameters for the option.
        overwrite : bool, optional
            Whether to overwrite existing settings. Default is True.
        """
        if params is not None:
            if overwrite:
                self.optimizer_defaults[option_name] = params
            elif option_name not in self.optimizer_defaults:
                self.optimizer_defaults[option_name] = params

    def add_option_to_params(self, option_name, params, overwrite = True):
        if params is None:
            return
        for key, value in params.items():
            check(self.all_params is not None, RuntimeError, "Call set_params before add_option_to_params")
            old_key = False
            for param in self.optimizer_params:
                if param['params'] == key:
                    old_key = True
                    if overwrite:
                        param[option_name] = value
                    elif option_name not in param:
                        param[option_name] = value
            if old_key == False:
                self.optimizer_params.append({'params': key, option_name: value})

    def replace_key_with_params(self):
        params = copy.deepcopy(self.optimizer_params)
        for param in params:
            if type(param['params']) is list:
                for ind, par in enumerate(param['params']):
                    param['params'][ind] = self.all_params[par]
            else:
                param['params'] = self.all_params[param['params']]
        return params

    def get_torch_optimizer(self):
        raise NotImplemented('The function get_torch_optimizer must be implemented.')

class SGD(Optimizer):
    """
    Stochastic Gradient Descent (SGD) optimizer.

    See also:
            Official PyTorch SGD documentation: 
            `torch.optim.SGD <https://pytorch.org/docs/stable/generated/torch.optim.SGD.html>`_

    Parameters
    ----------
    name : str
        The name of the optimizer.
    optimizer_defaults : dict
        A dictionary of default optimizer settings.
    optimizer_params : list
        A list of parameter groups for the optimizer.

    Attributes
    ----------
    name : str
        The name of the optimizer.
    lr : float, optional
        Learning rate. Default is 0.01.
    momentum : float, optional
        Momentum factor. Default is 0.0.
    dampening : float, optional
        Dampening for momentum. Default is 0.0.
    weight_decay : float, optional
        Weight decay (L2 penalty). Default is 0.0.
    nesterov : bool, optional
        Enables Nesterov momentum. Default is False.
    """
    def __init__(self, optimizer_defaults = {}, optimizer_params = []):
        super(SGD, self).__init__('SGD', optimizer_defaults, optimizer_params)

    def get_torch_optimizer(self):
        return torch.optim.SGD(self.replace_key_with_params(), **self.optimizer_defaults)

class Adam(Optimizer):
    """
    Stochastic Gradient Descent (SGD) optimizer.

    See also:
            Official PyTorch Adam documentation: 
            `torch.optim.Adam <https://pytorch.org/docs/stable/generated/torch.optim.SGD.html>`_

    Parameters
    ----------
    name : str
        The name of the optimizer.
    optimizer_defaults : dict
        A dictionary of default optimizer settings.
    optimizer_params : list
        A list of parameter groups for the optimizer.

    Attributes
    ----------
    name : str
        The name of the optimizer.
    lr : float, optional
        Learning rate. Default is 0.001.
    betas : tuple of (float, float), optional
        Coefficients used for computing running averages of gradient and its square. Default is (0.9, 0.999).
    eps : float, optional
        Term added to the denominator to improve numerical stability. Default is 1e-8.
    weight_decay : float, optional
        Weight decay (L2 penalty). Default is 0.0.
    amsgrad : bool, optional
        Whether to use the AMSGrad variant of this algorithm. Default is False.
    """
    def __init__(self, optimizer_defaults = {}, optimizer_params = []):
        super(Adam, self).__init__('Adam', optimizer_defaults, optimizer_params)

    def get_torch_optimizer(self):
        return torch.optim.Adam(self.replace_key_with_params(), **self.optimizer_defaults)

1	import copy	1✔
2	import torch	1✔
3
4	from nnodely.support.utils import check	1✔
5
6	class Optimizer:	1✔
7	"""
8	Represents an optimizer for training neural network models.
9
10	Parameters
11	----------
12	name : str
13	The name of the optimizer.
14	optimizer_defaults : dict, optional
15	A dictionary of default optimizer settings.
16	optimizer_params : list, optional
17	A list of parameter groups for the optimizer.
18
19	Attributes
20	----------
21	name : str
22	The name of the optimizer.
23	optimizer_defaults : dict
24	A dictionary of default optimizer settings.
25	optimizer_params : list
26	A list of parameter groups for the optimizer.
27	all_params : dict or None
28	A dictionary of all parameters in the model.
29	params_to_train : list or None
30	A list of parameters to be trained.
31	"""
32	def __init__(self, name, optimizer_defaults = {}, optimizer_params = []):	1✔
33	"""
34	Initializes the Optimizer object.
35
36	Parameters
37	----------
38	name : str
39	The name of the optimizer.
40	optimizer_defaults : dict, optional
41	A dictionary of default optimizer settings. Default is an empty dictionary.
42	optimizer_params : list, optional
43	A list of parameter groups for the optimizer. Default is an empty list.
44	"""
45	self.name = name	1✔
46	self.optimizer_defaults = copy.deepcopy(optimizer_defaults)	1✔
47	self.optimizer_params = self.unfold(copy.deepcopy(optimizer_params))	1✔
48	self.all_params = None	1✔
49	self.params_to_train = None	1✔
50
51	def set_params_to_train(self, all_params, params_to_train):	1✔
52	"""
53	Sets the parameters to be trained by the optimizer.
54
55	Parameters
56	----------
57	all_params : dict
58	A dictionary of all parameters in the model.
59	params_to_train : list
60	A list of parameters to be trained.
61	"""
62	self.all_params = all_params	1✔
63	self.params_to_train = params_to_train	1✔
64	if self.optimizer_params == []:	1✔
65	for param_name in self.all_params.keys():	1✔
66	if param_name in self.params_to_train:	1✔
67	self.optimizer_params.append({'params': param_name})	1✔
68	else:
69	self.optimizer_params.append({'params': param_name, 'lr': 0.0})	1✔
70
71	def set_defaults(self, optimizer_defaults):	1✔
72	"""
73	Sets the default optimizer settings.
74
75	Parameters
76	----------
77	optimizer_defaults : dict
78	A dictionary of default optimizer settings.
79	"""
80	self.optimizer_defaults = copy.deepcopy(optimizer_defaults)	1✔
81
82	def set_params(self, optimizer_params):	1✔
83	"""
84	Sets the parameter groups for the optimizer.
85
86	Parameters
87	----------
88	optimizer_params : list
89	A list of parameter groups for the optimizer.
90	"""
91	self.optimizer_params = self.unfold(optimizer_params)	1✔
92
93	def unfold(self, params):	1✔
94	"""
95	Unfolds the parameter groups into a flat list.
96
97	Parameters
98	----------
99	params : list
100	A list of parameter groups.
101
102	Returns
103	-------
104	list
105	A flat list of parameter groups.
106
107	Raises
108	------
109	KeyError
110	If the params argument is not a list.
111	"""
112	optimizer_params = []	1✔
113	check(type(params) is list, KeyError, f'The params {params} must be a list')	1✔
114	for param in params:	1✔
115	if type(param['params']) is list:	1✔
116	par_copy = copy.deepcopy(param)	1✔
117	del par_copy['params']	1✔
118	for par in param['params']:	1✔
119	optimizer_params.append({'params':par}\|par_copy)	1✔
120	else:
121	optimizer_params.append(param)	1✔
122	return optimizer_params	1✔
123
124	def add_defaults(self, option_name, params, overwrite = True):	1✔
125	"""
126	Adds default settings to the optimizer.
127
128	Parameters
129	----------
130	option_name : str
131	The name of the option to add.
132	params : any
133	The parameters for the option.
134	overwrite : bool, optional
135	Whether to overwrite existing settings. Default is True.
136	"""
137	if params is not None:	1✔
138	if overwrite:	1✔
139	self.optimizer_defaults[option_name] = params	1✔
140	elif option_name not in self.optimizer_defaults:	×
141	self.optimizer_defaults[option_name] = params	×
142
143	def add_option_to_params(self, option_name, params, overwrite = True):	1✔
144	if params is None:	1✔
UNCOV 145	return	×
146	for key, value in params.items():	1✔
147	check(self.all_params is not None, RuntimeError, "Call set_params before add_option_to_params")	1✔
148	old_key = False	1✔
149	for param in self.optimizer_params:	1✔
150	if param['params'] == key:	1✔
151	old_key = True	1✔
152	if overwrite:	1✔
153	param[option_name] = value	1✔
154	elif option_name not in param:	×
155	param[option_name] = value	×
156	if old_key == False:	1✔
157	self.optimizer_params.append({'params': key, option_name: value})	×
158
159	def replace_key_with_params(self):	1✔
160	params = copy.deepcopy(self.optimizer_params)	1✔
161	for param in params:	1✔
162	if type(param['params']) is list:	1✔
163	for ind, par in enumerate(param['params']):	×
164	param['params'][ind] = self.all_params[par]	×
165	else:
166	param['params'] = self.all_params[param['params']]	1✔
167	return params	1✔
168
169	def get_torch_optimizer(self):	1✔
170	raise NotImplemented('The function get_torch_optimizer must be implemented.')	×
171
172	class SGD(Optimizer):	1✔
173	"""
174	Stochastic Gradient Descent (SGD) optimizer.
175
176	See also:
177	Official PyTorch SGD documentation:
178	`torch.optim.SGD <https://pytorch.org/docs/stable/generated/torch.optim.SGD.html>`_
179
180	Parameters
181	----------
182	name : str
183	The name of the optimizer.
184	optimizer_defaults : dict
185	A dictionary of default optimizer settings.
186	optimizer_params : list
187	A list of parameter groups for the optimizer.
188
189	Attributes
190	----------
191	name : str
192	The name of the optimizer.
193	lr : float, optional
194	Learning rate. Default is 0.01.
195	momentum : float, optional
196	Momentum factor. Default is 0.0.
197	dampening : float, optional
198	Dampening for momentum. Default is 0.0.
199	weight_decay : float, optional
200	Weight decay (L2 penalty). Default is 0.0.
201	nesterov : bool, optional
202	Enables Nesterov momentum. Default is False.
203	"""
204	def __init__(self, optimizer_defaults = {}, optimizer_params = []):	1✔
205	super(SGD, self).__init__('SGD', optimizer_defaults, optimizer_params)	1✔
206
207	def get_torch_optimizer(self):	1✔
208	return torch.optim.SGD(self.replace_key_with_params(), **self.optimizer_defaults)	1✔
209
210	class Adam(Optimizer):	1✔
211	"""
212	Stochastic Gradient Descent (SGD) optimizer.
213
214	See also:
215	Official PyTorch Adam documentation:
216	`torch.optim.Adam <https://pytorch.org/docs/stable/generated/torch.optim.SGD.html>`_
217
218	Parameters
219	----------
220	name : str
221	The name of the optimizer.
222	optimizer_defaults : dict
223	A dictionary of default optimizer settings.
224	optimizer_params : list
225	A list of parameter groups for the optimizer.
226
227	Attributes
228	----------
229	name : str
230	The name of the optimizer.
231	lr : float, optional
232	Learning rate. Default is 0.001.
233	betas : tuple of (float, float), optional
234	Coefficients used for computing running averages of gradient and its square. Default is (0.9, 0.999).
235	eps : float, optional
236	Term added to the denominator to improve numerical stability. Default is 1e-8.
237	weight_decay : float, optional
238	Weight decay (L2 penalty). Default is 0.0.
239	amsgrad : bool, optional
240	Whether to use the AMSGrad variant of this algorithm. Default is False.
241	"""
242	def __init__(self, optimizer_defaults = {}, optimizer_params = []):	1✔
243	super(Adam, self).__init__('Adam', optimizer_defaults, optimizer_params)	1✔
244
245	def get_torch_optimizer(self):	1✔
246	return torch.optim.Adam(self.replace_key_with_params(), **self.optimizer_defaults)	1✔

tonegas / nnodely / 16502811447

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous