## Número de punto flotante FLOP y método de adquisición del programa de parámetros de entrenamiento de ejecución del modelo de aprendizaje profundo

``````# 浮点运行次数
# FLOPS：注意全大写，是floating point operations per second的缩写，意指每秒浮点运算次数，理解为计算速度。是一个衡量硬件性能的指标。
# FLOPs：注意s小写，是floating point operations的缩写（s表复数），意指浮点运算数，理解为计算量。可以用来衡量算法/模型的复杂度。
# To work your code in TF 2.1.0, i have made all necessary changes that are compliant to TF 2.x

# print(tf.__version__)
import tensorflow as tf
# 必须要下面这行代码
tf.compat.v1.disable_eager_execution()
print(tf.__version__)

# 我自己使用的函数
def get_flops_params():
sess = tf.compat.v1.Session()
graph = sess.graph
flops = tf.compat.v1.profiler.profile(graph, options=tf.compat.v1.profiler.ProfileOptionBuilder.float_operation())
params = tf.compat.v1.profiler.profile(graph, options=tf.compat.v1.profiler.ProfileOptionBuilder.trainable_variables_parameter())
print('FLOPs: {};    Trainable params: {}'.format(flops.total_float_ops, params.total_parameters))

# 网上推荐的
# sess = tf.compat.v1.Session()
# graph = sess.graph
# stats_graph(graph)
def stats_graph(graph):
flops = tf.compat.v1.profiler.profile(graph, options=tf.compat.v1.profiler.ProfileOptionBuilder.float_operation())
# print('FLOPs: {}'.format(flops.total_float_ops))
params = tf.compat.v1.profiler.profile(graph, options=tf.compat.v1.profiler.ProfileOptionBuilder.trainable_variables_parameter())
# print('Trainable params: {}'.format(params.total_parameters))
print('FLOPs: {};    Trainable params: {}'.format(flops.total_float_ops, params.total_parameters))

def get_flops(model):
opts = tf.compat.v1.profiler.ProfileOptionBuilder.float_operation()
# We use the Keras session graph in the call to the profiler.
flops = tf.compat.v1.profiler.profile(graph=tf.compat.v1.keras.backend.get_session().graph, run_meta=run_meta, cmd='op', options=opts)
return flops.total_float_ops  # Prints the "flops" of the model.

# 必须使用tensorflow中的keras才能够获取到FLOPs, 模型中的各个函数都必须使用tensorflow.keras中的函数，和keras混用会报错
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense
from tensorflow.keras.models import Sequential

model = Sequential()
model.add(Conv2D(filters=64, kernel_size=(3, 3), input_shape=(28, 28, 1), activation='relu'))
# 获取模型每一层的参数详情
model.summary()
# 获取模型浮点运算总次数和模型的总参数
get_flops_params()``````
• ## resultado de la operación model.summary ()

Modelo: "secuencial"
_________________________________________________________________
Capa (tipo) Forma de salida Parámetro #
==================================== =============================
conv2d (Conv2D) (Ninguno, 26, 26, 64) 640
_________________________________________________________________
max_pooling2d (MaxPooling2D) (Ninguno , 13, 13, 64) 0
_________________________________________________________________
aplanar (aplanar) (Ninguno, 10816) 0
_________________________________________________________________
denso (Denso) (Ninguno, 100) 1081700
_________________________________________________________________
dense_1 (Denso) (Ninguno, 10) 1010
======================================= ==========================
Parámetros totales: 1.083.350
Parámetros
entrenables : 1.083.350 Parámetros no entrenables: 0

• ## resultado de la operación get_flops_params ()

================== Informe de análisis del modelo ======================
Forma incompleta.
Forma incompleta.

Doc:
scope: Los nodos en el gráfico del modelo están organizados por sus nombres, que es jerárquico como un sistema de archivos.
param: Número de parámetros (en la Variable).

Perfil:
nombre de nodo | # parámetros
_TFProfRoot (- / 1.08m params)
conv2d (- / 640 params)
conv2d / bias (64, 64/64 params)
conv2d / kernel (3x3x1x64, 576/576 params)
denso (- / 1.08m params)
denso / sesgo (100, 100/100 params)
denso / kernel (10816x100, 1.08m / 1.08m params)
dense_1 (- / 1.01k params)
dense_1 / bias (10, 10/10 params)
dense_1 / kernel (100x10, 1.00k / 1.00k parámetros)

====================== Fin del informe ========================= =
FLOP: 2166355; Parámetros entrenables: 1083350

### Supongo que te gusta

Origin blog.csdn.net/deephacking/article/details/107873881