mxnet 尝试

1. 读写NDArrays

   from mxnet import nd
   from mxnet.gluon impor nn
   x = nd.ones(3)
   nd.save('x',x)
   
   x2 = nd.load('x')
   x2

   输出结果为：
   
   
   
    

2. 下载模型并使用
    

import mxnet as mx
import numpy as np


batch_size = 1
num_batch = 5
# Adding 0.1 to each of the values
eval_data = np.array([[7,2],[6,10],[12,2]])
eval_label = np.array([11.1,26.1,16.1]) 
eval_iter = mx.io.NDArrayIter(eval_data, eval_label, batch_size, shuffle=False)
# load model
sym,arg_params,aux_params = mx.model.load_checkpoint('test', 5)

mod = mx.mod.Module(symbol=sym,context=mx.gpu(),data_names=['data'],label_names=['lin_reg_label'])
mod.bind(for_training=False,data_shapes=[('data', (1, 2))])
mod.set_params(arg_params,aux_params)
# use model
predict_stress = mod.predict(eval_iter, num_batch)
print(predict_stress.asnumpy())

• batch_size 批尺寸（批处理大小），batch就是每次送入网络中训练的一部分数据，batch_size就是每个batch中训练样本的数量。随着batch_size增大，处理相同数据量的速度越快，达到相同精度所需要的epoch数量越来越多。
• num_batch