昇思训练平台的问题
drizzlezyk/book-review
data_path
datasets/drizzlezyk/book-review/train.csv
test_path
drizzlezyk/book-review/test.csv
glove_path
drizzlezyk/book-review/glove.6B.100d.txt
datasets/drizzlezyk/book-review/glove.6B.100d.txt’
parser.add_argument('--pretrain_url', type=str,default='./result/best50.ckpt',help='the pretrain model path')
训练仓库1111123/ResNet50
pretrain_path
drizzlezyk/ResNet50_model/resnet50_224.ckpt
或者MindSpore/ResNet50_model/resnet50_224.ckpt
data_path
drizzlezyk/bee-img/train 修改为drizzlezyk/bee-img/train/ （！！！！！都是文件夹）
drizzlezyk/cifar10/ 假如这个 在第三个空格千万不要不要不要放上/ 改为drizzlezyk/cifar10 （！！！！！都是文件夹）
##output 默认不填
#output_path
#1111123/ResNet501
问题：要么提示用户名和仓库不对应，要么启动后在30秒左右提示启动失败
data_path后面的train文件夹后面加个/呢，因为你引入的是一个文件夹，后台识别为文件了，所以找不到这个文件
response has status:500 Internal Server Error and body:“{“code”:“system_error”,“msg”:“Bad request with: [POST https://modelarts.cn-north-4.myhuaweicloud.com/v2/0704467c7e8025362f1ec0019cc93707/training-jobs], error message: {\“error_msg\”:\“Invalid obs path: /xihe-production-obs/xihe-repos/drizzlezyk/2460/train, reason: object not fount.\”,\“error_code\”:\“ModelArts.2791\”,\“error_solution\”:\“Check the obs path.\”}\n”,“data”:null}”
########################请注意，假如训练提示找不到仓库，请把当前的项目删除，重新跑就行
datasets/zhongxj/bee-img/val
models/xuzhipeng12/l-bee-model/model.ckpt
train/
output_path 1111123/ResNet50guangangzhengshi/train/resnet/
MindSpore/ResNet50_model resnet50_224.ckpt
data_path
drizzlezyk/cifar10
MindSpore/ResNet50_model/
添加参数
parser.add_argument('--pretrain_path',type=str,default='models/drizzlezyk/resnet50_model/resnet50_224.ckpt',help='the pretrain file')parser.add_argument('--data_path',type=str,default='datasets/drizzlezyk/cifar10/',help='the training data')parser.add_argument('--output_path',default='train/resnet/',type=str,help='the path model saved')parser.add_argument('--pretrain_path',type=str,default='models/drizzlezyk/ResNet50_model/resnet50_224.ckpt',help='the pretrain file')parser.add_argument('--data_path',type=str,default='datasets/drizzlezyk/cifar10/',help='the training data')parser.add_argument('--output_path',default='train/resnet/',type=str,help='the path model saved')
