Azure AutoML依赖关系失败

Question

发布：无法从AutoML运行中获得最佳模型。

代码：

best_run, fitted_model = automl_run.get_output()
print(best_run.properties["run_algorithm"])

错误消息:

ErrorResponse 
[stderr]{
[stderr]    "error": {
[stderr]        "code": "UserError",
[stderr]        "message": "The model you attempted to retrieve requires 'xgboost' to be installed at '==1.3.3'. You have 'xgboost==1.3.3', please reinstall 'xgboost==1.3.3' (e.g. `pip install xgboost==1.3.3`) and rerun the previous command.",
[stderr]        "target": "get_output",
[stderr]        "inner_error": {
[stderr]            "code": "NotSupported",
[stderr]            "inner_error": {
[stderr]                "code": "IncompatibleOrMissingDependency"
[stderr]            }
[stderr]        },
[stderr]        "reference_code": "910310e6-2433-40cd-b597-9ec2950bc1d8"
[stderr]    }

Conda依赖关系

# Conda environment specification. The dependencies defined in this file will
# be automatically provisioned for runs with userManagedDependencies=False.

# Details about the Conda environment file format:
# https://conda.io/docs/user-guide/tasks/manage-environments.html#create-env-file-manually

name: project_environment
dependencies:
  # The python interpreter version.
  # Currently Azure ML only supports 3.5.2 and later.
- python=3.6.12

- pip:
  - azureml-train-automl-runtime==1.38.0
  - azureml-train-automl-client==1.38.0
  - inference-schema
  - azureml-interpret==1.38.0
  - azureml-defaults==1.38.0
- numpy>=1.16.0,<1.19.0
- pandas==0.25.1
- scikit-learn==0.22.1
- py-xgboost<=1.3.3
- fbprophet==0.5
- holidays==0.9.11
- psutil>=5.2.2,<6.0.0
- matplotlib=3.3.2
- seaborn=0.9.0
- joblib=0.13.2
- joblib
channels:
- anaconda
- conda-forge

问题：

• 我的conda依赖项中应该有什么可以修复此错误？
• 我试过制作py-xgboost==1.3.3，但没成功。
• 有什么好办法解决的吗？

Answer 1

错误--“您试图检索的模型要求'xgboost‘安装在’=1.3.3‘处。您有'xgboost==1.3.3'，请重新安装'xgboost==1.3.3’(例如pip install xgboost==1.3.3)并重新运行前面的命令。”

如上述错误消息所示，应该是pip install xgboost==1.3.3而不是py-xgboost<=1.3.3。

如果它不能工作，尝试降级版本的xgboost

pip install xgboost==0.90

请参阅此github 链接