PPStructure中KIE任务中RE子任务使用带蒸馏版本的模型预测报“Backbone” KeyError错但yml配置文件中有Backbone字段

[galaxyknight]

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment：CentOS7+Python3.8+PaddleOCR2.6
• 版本号/Version：Paddle： PaddleOCR： 问题相关组件/Related components：
• 运行指令/Command Code：python ./tools/infer_kie_token_ser_re.py \

-c configs/kie/vi_layoutxlm/re_vi_layoutxlm_sinovatio_zh_udml.yml
-o Architecture.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/re_vi_layoutxlm_sinovatio_zh_udml/best_accuracy
Global.infer_img=/data01/wj/public_data/dataset/sinovatio_data/dev/dev_img/
-c_ser configs/kie/vi_layoutxlm/ser_vi_layoutxlm_sinovatio_zh_udml.yml
-o_ser Architecture.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/ser_vi_layoutxlm_sinovatio_zh_udml/best_accuracy

• 完整报错/Complete Error Message：
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/profiler/profiler_statistic.py:1402: DeprecationWarning: invalid escape sequence (
  kernel_name_pattern = re.compile('(.+?)(<.>)((.))')
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/tensor/random.py:31: DeprecationWarning: invalid escape sequence \e
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/distributed/parallel.py:87: SyntaxWarning: "is" with a literal. Did you mean "=="?
  ) or backend is 'xccl':
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/fluid/contrib/slim/quantization/post_training_quantization.py:438: SyntaxWarning: "is" with a literal. Did you mean "=="?
  if self.algo is 'min_max':
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/incubate/nn/layer/fused_linear.py:20: DeprecationWarning: invalid escape sequence _
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_optimizer_stage2.py:208: DeprecationWarning: invalid escape sequence .
  "The param name passed to the optimizer doesn't follow .+[0-9]+..+ patter, "
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/dataset/common.py:38: DeprecationWarning: invalid escape sequence \T
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/sparse/unary.py:555: DeprecationWarning: invalid escape sequence \p
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/sparse/unary.py:588: DeprecationWarning: invalid escape sequence \p
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/sparse/nn/functional/transformer.py:29: DeprecationWarning: invalid escape sequence \s
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/sparse/nn/functional/activation.py:64: DeprecationWarning: invalid escape sequence \e
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/paddle/sparse/nn/functional/activation.py:149: DeprecationWarning: invalid escape sequence _
  """
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/skimage/util/dtype.py:27: DeprecationWarning: np.bool8 is a deprecated alias for np.bool_. (Deprecated NumPy 1.24)
  np.bool8: (False, True),
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/iterative.py:10: DeprecationWarning: Please use LinearOperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import LinearOperator
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/utils.py:7: DeprecationWarning: Please use asmatrix from the scipy.sparse namespace, the scipy.sparse.sputils namespace is deprecated.
  from scipy.sparse.sputils import asmatrix
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/utils.py:9: DeprecationWarning: Please use aslinearoperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator, LinearOperator,
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/utils.py:9: DeprecationWarning: Please use LinearOperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator, LinearOperator,
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/utils.py:9: DeprecationWarning: Please use IdentityOperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator, LinearOperator,
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/lsqr.py:56: DeprecationWarning: Please use aslinearoperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/isolve/lsmr.py:24: DeprecationWarning: Please use aslinearoperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/dsolve/linsolve.py:7: DeprecationWarning: Please use is_pydata_spmatrix from the scipy.sparse namespace, the scipy.sparse.sputils namespace is deprecated.
  from scipy.sparse.sputils import is_pydata_spmatrix
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/eigen/arpack/arpack.py:47: DeprecationWarning: Please use aslinearoperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator, LinearOperator
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/eigen/arpack/arpack.py:47: DeprecationWarning: Please use LinearOperator from the scipy.sparse.linalg namespace, the scipy.sparse.linalg.interface namespace is deprecated.
  from scipy.sparse.linalg.interface import aslinearoperator, LinearOperator
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/eigen/arpack/arpack.py:50: DeprecationWarning: Please use isdense from the scipy.sparse namespace, the scipy.sparse.sputils namespace is deprecated.
  from scipy.sparse.sputils import isdense, is_pydata_spmatrix
  /home/rhino/anaconda3/envs/zkq_py38_paddleocr_cuda11/lib/python3.8/site-packages/scipy/sparse/linalg/eigen/arpack/arpack.py:50: DeprecationWarning: Please use is_pydata_spmatrix from the scipy.sparse namespace, the scipy.sparse.sputils namespace is deprecated.
  from scipy.sparse.sputils import isdense, is_pydata_spmatrix
  :219: RuntimeWarning: scipy._lib.messagestream.MessageStream size changed, may indicate binary incompatibility. Expected 56 from C header, got 64 from PyObject
  Traceback (most recent call last):
  File "./tools/infer_kie_token_ser_re.py", line 186, in
  config, ser_config, device, logger = preprocess()
  File "./tools/infer_kie_token_ser_re.py", line 162, in preprocess
  config = merge_config(config, FLAGS.opt)
  File "/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/tools/program.py", line 113, in merge_config
  cur = cur[sub_key]
  KeyError: 'Backbone'

[an1018]

使用dygraph的代码试试

[galaxyknight]

dygraph的代码具体在哪啊，我在PaddleOCR-release-2.6这个版本了没找到

[mxzgn]

试一试改为这样子

[zhaoqf-cq]

-o Architecture.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/re_vi_layoutxlm_sinovatio_zh_udml/best_accuracy

把这个替换成

-o Architecture.Models.Student.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/re_vi_layoutxlm_sinovatio_zh_udml/best_accuracy

[Alex37882388]

-o Architecture.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/re_vi_layoutxlm_sinovatio_zh_udml/best_accuracy

把这个替换成

-o Architecture.Models.Student.Backbone.checkpoints=/data01/wj/PythonWorkspace/PaddleOCR-release-2.6/output/re_vi_layoutxlm_sinovatio_zh_udml/best_accuracy

你是对的，这个方法对我有用，确实是配置路径问题，蒸馏配置Backbone不在根节点：

[qet119]

有时候官方的还得看仔细

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.

[zhanglong2100]

你在预测的时候可以不用re_vi_layoutxlm_sinovatio_zh_udml文件，改成re_vi_layoutxlm_sinovatio_zh.yml文件。你用udml文件，即使将命令改成Architecture.Models.Student.Backbone.checkpoints这样解决这个问题，但也会遇到其它问题： 'DistillationModel' object has no attribute 'backbone'、ser_results[0] KeyError: '0'。