mindspore-softmax进行鸢尾花多分类模型

版本：mindspore1.3.0

代码：

import os
# os.environ['DEVICE_ID'] = '6'
import csv
import numpy as np

import mindspore as ms
from mindspore import nn
from mindspore import context
from mindspore import dataset
from mindspore.train.callback import LossMonitor
from mindspore.common.api import ms_function
from mindspore.ops import operations as P

context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
with open('iris.data') as csv_file:
    data = list(csv.reader(csv_file, delimiter=','))

label_map = {
    'Iris-setosa': 0,
    'Iris-versicolor': 1,
    'Iris-virginica':2,
}

X = np.array([[float(x) for x in s[:-1]] for s in data[:150]], np.float32)
 

Y = np.array([[label_map[s[-1]]] for s in data[:150]], np.float32)
 

train_idx = np.random.choice(150, 120, replace=False)
test_idx = np.array(list(set(range(150)) - set(train_idx)))
X_train, Y_train = X[train_idx], Y[train_idx]
X_test, Y_test = X[test_idx], Y[test_idx]
XY_train = list(zip(X_train, Y_train))
ds_train = dataset.GeneratorDataset(XY_train, ['x', 'y'])

ds_train = ds_train.shuffle(buffer_size=80).batch(32, drop_remainder=True)
XY_test = list(zip(X_test, Y_test))
ds_test = dataset.GeneratorDataset(XY_test, ['x', 'y'])
ds_test = ds_test.batch(30)

net = nn.Dense(4, 3)
loss = nn.loss.SoftmaxCrossEntropyWithLogits(sparse=True, reduction='mean')
opt = nn.optim.Momentum(net.trainable_params(), learning_rate=0.05, momentum=0.9)

model = ms.train.Model(net, loss, opt, metrics={'acc', 'loss'})
model.train(15, ds_train, callbacks=[LossMonitor(per_print_times=ds_train.get_dataset_size())], dataset_sink_mode=False)
metrics = model.eval(ds_test)
print(metrics)

1. 按照报错提示，是因为你的dataset对象给多个model使用了。

2. 但是我们拿了你上面的脚本，先从脚本上看，没有发现ds_train / ds_eval给多个model使用的情况，另：在本地运行后，也没有报你上面的报错。

故：你可以再试下你上条评论里的脚本，或者还有没有其他信息提供？我们再分析下。