pyerualjetwork 4.0.5__py3-none-any.whl

pyerualjetwork/model_operations_cuda.py

@@ -0,0 +1,414 @@
+import cupy as cp
+from colorama import Fore, Style
+import sys
+from datetime import datetime
+import pickle
+from scipy import io
+import scipy.io as sio
+import pandas as pd
+
+
+def save_model(model_name,
+               W,
+               scaler_params=None,
+               model_type='PLAN',
+               test_acc=None,
+               model_path='',
+               activation_potentiation=['linear'],
+               weights_type='npy',
+               weights_format='raw',
+               show_architecture=None,
+               show_info=True
+               ):
+
+    """
+    Function to save a potentiation learning artificial neural network model.
+
+    Arguments:
+
+    model_name (str): Name of the model.
+
+    model_type (str): Type of the model. default: 'PLAN'
+
+    test_acc (float): Test accuracy of the model. default: None
+
+    weights_type (str): Type of weights to save (options: 'txt', 'pkl', 'npy', 'mat'). default: 'npy'
+
+    WeightFormat (str): Format of the weights (options: 'f', 'raw'). default: 'raw'
+
+    model_path (str): Path where the model will be saved. For example: C:/Users/beydili/Desktop/denemePLAN/ default: ''
+
+    scaler_params (list[num, num]): standard scaler params list: mean,std. If not used standard scaler then be: None.
+
+    W: Weights of the model.
+
+    activation_potentiation (list): For deeper PLAN networks, activation function parameters. For more information please run this code: plan.activations_list() default: ['linear']
+
+    show_architecture (str): It draws model architecture. Takes 2 value='basic' or 'detailed'. Default: None(not drawing)
+
+    show_info (bool): Prints model details into console. default: True
+
+    Returns:
+    No return.
+    """
+    
+    from .visualizations_cuda import draw_model_architecture
+
+    class_count = W.shape[0]
+
+    if test_acc != None:
+        test_acc= float(test_acc)
+
+    if weights_type != 'txt' and weights_type != 'npy' and weights_type != 'mat' and weights_type != 'pkl':
+        print(Fore.RED + "ERROR110: Save Weight type (File Extension) Type must be 'txt' or 'npy' or 'mat' or 'pkl' from: save_model" + Style.RESET_ALL)
+        sys.exit()
+
+    if weights_format != 'd' and weights_format != 'f' and weights_format != 'raw':
+        print(Fore.RED + "ERROR111: Weight Format Type must be 'd' or 'f' or 'raw' from: save_model" + Style.RESET_ALL)
+        sys.exit()
+
+    NeuronCount = 0
+    SynapseCount = 0
+
+
+    try:
+            NeuronCount += cp.shape(W)[0] + cp.shape(W)[1]
+            SynapseCount += cp.shape(W)[0] * cp.shape(W)[1]
+    except:
+
+        print(Fore.RED + "ERROR: Weight matrices has a problem from: save_model" + Style.RESET_ALL)
+        sys.exit()
+
+    if scaler_params != None:
+
+        if len(scaler_params) > len(activation_potentiation):
+
+            activation_potentiation += ['']
+
+        elif len(activation_potentiation) > len(scaler_params):
+
+            for i in range(len(activation_potentiation) - len(scaler_params)):
+
+                scaler_params.append(' ')
+
+    data = {'MODEL NAME': model_name,
+            'MODEL TYPE': model_type,
+            'CLASS COUNT': class_count,
+            'NEURON COUNT': NeuronCount,
+            'SYNAPSE COUNT': SynapseCount,
+            'TEST ACCURACY': test_acc,
+            'SAVE DATE': datetime.now(),
+            'WEIGHTS TYPE': weights_type,
+            'WEIGHTS FORMAT': weights_format,
+            'MODEL PATH': model_path,
+            'STANDARD SCALER': scaler_params,
+            'ACTIVATION POTENTIATION': activation_potentiation
+            }
+
+    df = pd.DataFrame(data)
+    df.to_pickle(model_path + model_name + '.pkl')
+
+
+    try:
+
+        if weights_type == 'txt' and weights_format == 'f':
+
+                cp.savetxt(model_path + model_name + '_weights.txt',  W, fmt='%f')
+
+        if weights_type == 'txt' and weights_format == 'raw':
+
+                cp.savetxt(model_path + model_name + '_weights.txt',  W)
+
+        ###
+
+        
+        if weights_type == 'pkl' and weights_format == 'f':
+
+            with open(model_path + model_name + '_weights.pkl', 'wb') as f:
+                pickle.dump(W.astype(float), f)
+
+        if weights_type == 'pkl' and weights_format =='raw':
+        
+            with open(model_path + model_name + '_weights.pkl', 'wb') as f:
+                pickle.dump(W, f)
+
+        ###
+
+        if weights_type == 'npy' and weights_format == 'f':
+
+                cp.save(model_path + model_name + '_weights.npy',  W, W.astype(float))
+
+        if weights_type == 'npy' and weights_format == 'raw':
+
+                cp.save(model_path + model_name + '_weights.npy',  W)
+
+        ###
+
+        if weights_type == 'mat' and weights_format == 'f':
+
+                w = {'w': W.astype(float)}
+                io.savemat(model_path + model_name + '_weights.mat', w)
+
+        if weights_type == 'mat' and weights_format == 'raw':
+                
+                w = {'w': W}
+                io.savemat(model_path + model_name + '_weights.mat', w)
+
+    except:
+
+        print(Fore.RED + "ERROR: Model Weights not saved. Check the Weight parameters. SaveFilePath expl: 'C:/Users/hasancanbeydili/Desktop/denemePLAN/' from: save_model" + Style.RESET_ALL)
+        sys.exit()
+    
+    if show_info:
+        print(df)
+    
+        message = (
+            Fore.GREEN + "Model Saved Successfully\n" +
+            Fore.MAGENTA + "Don't forget, if you want to load model: model log file and weight files must be in the same directory." +
+            Style.RESET_ALL
+        )
+        
+        print(message)
+
+    if show_architecture is not None:
+        draw_model_architecture(model_name=model_name, model_path=model_path, style=show_architecture)
+
+
+
+def load_model(model_name,
+               model_path,
+               ):
+    """
+   Function to load a potentiation learning model.
+
+   Arguments:
+
+   model_name (str): Name of the model.
+
+   model_path (str): Path where the model is saved.
+
+   Returns:
+   lists: W(list[num]), activation_potentiation, DataFrame of the model
+    """
+    try:
+
+         df = pd.read_pickle(model_path + model_name + '.pkl')
+
+    except:
+
+        print(Fore.RED + "ERROR: Model Path error. acceptable form: 'C:/Users/hasancanbeydili/Desktop/denemePLAN/' from: load_model" + Style.RESET_ALL)
+
+        sys.exit()
+
+    activation_potentiation = list(df['ACTIVATION POTENTIATION'])
+    activation_potentiation = [x for x in activation_potentiation if not (isinstance(x, float) and cp.isnan(x))]
+    activation_potentiation = [item for item in activation_potentiation if item != ''] 
+
+    scaler_params = df['STANDARD SCALER'].tolist()
+    
+    try:
+        if scaler_params[0] == None:
+            scaler_params = scaler_params[0]
+
+    except:
+        scaler_params = [item for item in scaler_params if isinstance(item, cp.ndarray)]
+
+     
+    model_name = str(df['MODEL NAME'].iloc[0])
+    WeightType = str(df['WEIGHTS TYPE'].iloc[0])
+
+    if WeightType == 'txt':
+            W = cp.loadtxt(model_path + model_name + '_weights.txt')
+    elif WeightType == 'npy':
+            W = cp.load(model_path + model_name + '_weights.npy')
+    elif WeightType == 'mat':
+            W = sio.loadmat(model_path + model_name + '_weights.mat')
+    elif WeightType == 'pkl':
+        with open(model_path + model_name + '_weights.pkl', 'rb') as f:
+            W = pickle.load(f)
+    else:
+
+        raise ValueError(
+            Fore.RED + "Incorrect weight type value. Value must be 'txt', 'npy', 'pkl' or 'mat' from: load_model." + Style.RESET_ALL)
+        
+    if WeightType == 'mat':
+        W = W['w']
+
+    return W, None, None, activation_potentiation, scaler_params
+
+
+def predict_model_ssd(Input, model_name, model_path='', dtype=cp.float32):
+
+    """
+    Function to make a prediction using a potentiation learning artificial neural network (PLAN).
+    from storage
+
+    Arguments:
+
+    Input (list or ndarray): Input data for the model (single vector or single matrix).
+
+    model_name (str): Name of the model.
+
+    model_path (str): Path of the model. Default: ''
+
+    dtype (cupy.dtype): Data type for the arrays. np.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+    ndarray: Output from the model.
+    """
+            
+    Input = cp.array(Input, dtype=dtype, copy=False)
+        
+    from .plan_cuda import feed_forward
+    from .data_operations_cuda import standard_scaler
+    
+    model = load_model(model_name, model_path)
+    
+    activation_potentiation = model[get_act_pot()]
+    scaler_params = model[get_scaler()]
+    W = model[get_weights()]
+
+    Input = standard_scaler(None, Input, scaler_params)
+
+    neural_layer = Input
+    neural_layer = cp.array(neural_layer)
+    neural_layer = neural_layer.ravel()
+
+    try:
+        neural_layer = feed_forward(neural_layer, cp.copy(W), is_training=False, Class='?', activation_potentiation=activation_potentiation)
+        return neural_layer
+    except:
+        print(Fore.RED + "ERROR: Unexpected Output or wrong model parameters from: predict_model_ssd." + Style.RESET_ALL)
+        sys.exit()
+    
+
+def reverse_predict_model_ssd(output, model_name, model_path='', dtype=cp.float32):
+
+    """
+    reverse prediction function from storage
+    Arguments:
+
+    output (list or ndarray): output layer for the model (single probability vector, output layer of trained model).
+
+    model_name (str): Name of the model.
+
+    model_path (str): Path of the model. Default: ''
+
+    dtype (cupy.dtype): Data type for the arrays. np.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+    ndarray: Input from the model.
+    """
+    
+    output = cp.array(output, dtype=dtype, copy=False)
+        
+    model = load_model(model_name, model_path)
+    
+    W = model[get_weights()]
+
+    try:
+        Input = cp.dot(output, cp.copy(W))
+        return Input
+    except:
+        print(Fore.RED + "ERROR: Unexpected Output or wrong model parameters from: reverse_predict_model_ssd." + Style.RESET_ALL)
+        sys.exit()
+
+
+def predict_model_ram(Input, W, scaler_params=None, activation_potentiation=['linear'], dtype=cp.float32):
+
+    """
+    Function to make a prediction using a potentiation learning artificial neural network (PLAN).
+    from memory.
+
+    Arguments:
+
+    Input (list or ndarray): Input data for the model (single vector or single matrix).
+
+    W (list of ndarrays): Weights of the model.
+
+    scaler_params (list): standard scaler params list: mean,std. (optional) Default: None.
+    
+    activation_potentiation (list): ac list for deep PLAN. default: [None] ('linear') (optional)
+
+    dtype (cupy.dtype): Data type for the arrays. np.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+    ndarray: Output from the model.
+    """
+
+    from .data_operations_cuda import standard_scaler
+    from .plan_cuda import feed_forward
+
+    Input = cp.array(Input, dtype=dtype, copy=False)
+    
+    Input = standard_scaler(None, Input, scaler_params)
+    
+    try:
+        
+        neural_layer = Input
+        neural_layer = cp.array(neural_layer)
+        neural_layer = neural_layer.ravel()
+
+        neural_layer = feed_forward(neural_layer, cp.copy(W), is_training=False, Class='?', activation_potentiation=activation_potentiation)
+
+        return neural_layer
+    
+    except:
+        print(Fore.RED + "ERROR: Unexpected input or wrong model parameters from: predict_model_ram." + Style.RESET_ALL)
+        sys.exit()
+
+def reverse_predict_model_ram(output, W, dtype=cp.float32):
+
+    """
+    reverse prediction function from memory
+
+    Arguments:
+
+    output (list or ndarray): output layer for the model (single probability vector, output layer of trained model).
+
+    W (list of ndarrays): Weights of the model.
+
+    dtype (cupy.dtype): Data type for the arrays. np.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+    ndarray: Input from the model.
+    """
+
+    output = cp.array(output, dtype=dtype, copy=False)
+
+    try:
+        Input = cp.dot(output, cp.copy(W))
+        return Input
+    
+    except:
+        print(Fore.RED + "ERROR: Unexpected Output or wrong model parameters from: reverse_predict_model_ram." + Style.RESET_ALL)
+        sys.exit()
+
+
+def get_weights():
+
+    return 0
+
+
+def get_preds():
+
+    return 1
+
+
+def get_acc():
+
+    return 2
+
+
+def get_act_pot():
+
+    return 3
+
+
+def get_scaler():
+
+    return 4
+
+def get_preds_softmax():
+
+    return 5