Cannot feed value of shape (1, 1, 1, 1) for Tensor 'Placeholder_1:0', which has shape '(7, 1, 1, 1)'

[Hongyuan-Liu]

我在用rknn转换百度的paddle的shitu模型pplcnet，其实就是一个特征提取网络，用mmse量化出现如下问题：
2023-04-21 06:41:42.264360: W tensorflow/compiler/jit/mark_for_compilation_pass.cc:1412] (One-time warning): Not using XLA:CPU for cluster because envvar TF_XLA_FLAGS=--tf_xla_cpu_global_jit was not set. If you want XLA:CPU, either set that envvar, or use experimental_jit_scope to enable XLA:CPU. To confirm that XLA is active, pass --vmodule=xla_compilation_cache=1 (as a proper command-line flag, not via TF_XLA_FLAGS) or set the envvar XLA_FLAGS=--xla_hlo_profile.
W:tensorflow:From /usr/local/lib/python3.6/dist-packages/rknn/api/rknn.py:278: The name tf.nn.avg_pool is deprecated. Please use tf.nn.avg_pool2d instead.

/usr/local/lib/python3.6/dist-packages/scipy/ndimage/interpolation.py:611: UserWarning: From scipy 0.13.0, the output shape of zoom() is calculated with round() instead of int() - for these inputs the size of the returned array has changed.
"the returned array has changed.", UserWarning)
MMSE Quant Step 0: 94%|#######################################3 | 106/113 [01:42<00:07, 1.07s/it]E Run layer Mul_p2o.Mul.12_4 catch exception: Cannot feed value of shape (1, 1, 1, 1) for Tensor 'Placeholder_1:0', which has shape '(7, 1, 1, 1)'
MMSE Quant Step 0: 95%|#######################################7 | 107/113 [01:42<00:05, 1.04it/s]
E Catch exception when building RKNN model!
E Traceback (most recent call last):
E File "rknn/base/RKNNlib/app/medusa/workspace.py", line 498, in rknn.base.RKNNlib.app.medusa.workspace.Workspace.sparse_run
E File "rknn/base/RKNNlib/RKNN_session.py", line 37, in rknn.base.RKNNlib.RKNN_session.RKNNSession.run
E File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py", line 950, in run
E run_metadata_ptr)
E File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py", line 1149, in _run
E str(subfeed_t.get_shape())))
E ValueError: Cannot feed value of shape (1, 1, 1, 1) for Tensor 'Placeholder_1:0', which has shape '(7, 1, 1, 1)'
E During handling of the above exception, another exception occurred:
E Traceback (most recent call last):
E File "rknn/base/RKNNlib/RK_nn.py", line 579, in rknn.base.RKNNlib.RK_nn.RKnn.quantize
E File "rknn/base/RKNNlib/app/medusa/quantization.py", line 153, in rknn.base.RKNNlib.app.medusa.quantization.Quantization.run
E File "rknn/base/RKNNlib/app/medusa/quantization.py", line 62, in rknn.base.RKNNlib.app.medusa.quantization.Quantization._run_quantization
E File "rknn/base/RKNNlib/app/medusa/workspace.py", line 501, in rknn.base.RKNNlib.app.medusa.workspace.Workspace.sparse_run
E File "rknn/api/rknn_log.py", line 323, in rknn.api.rknn_log.RKNNLog.e
E ValueError: Run layer Mul_p2o.Mul.12_4 catch exception: Cannot feed value of shape (1, 1, 1, 1) for Tensor 'Placeholder_1:0', which has shape '(7, 1, 1, 1)'
E Please feedback the detailed log file <log_feedback_to_the_rknn_toolkit_dev_team.log> to the RKNN Toolkit development team.
E You can also check github issues: https://github.com/rockchip-linux/rknn-toolkit/issues
build model failed.
如果不用mmse，采用kl_divergence、normal不会报错，但是精度损失比较严重。

[nicklezheng]

我使用toolkit1.7.5采用mmse量化时遇到了类似的问题

E Run sub net (layer: Concat_Concat_36_62_RKNN_mark_perm_163) catch exception: Cannot feed value of shape (1, 20, 36, 64) for Tensor 'Placeholder:0', which has shape '(100, 20, 36, 64)'
E Catch exception when building RKNN model!
E Traceback (most recent call last):
E   File "rknn\base\RKNNlib\app\medusa\quantization.py", line 153, in rknn.base.RKNNlib.app.medusa.quantization.Quantization.run
E   File "rknn\base\RKNNlib\app\medusa\quantization.py", line 62, in rknn.base.RKNNlib.app.medusa.quantization.Quantization._run_quantization
E   File "rknn\base\RKNNlib\app\medusa\workspace.py", line 987, in rknn.base.RKNNlib.app.medusa.workspace.Workspace.sparse_run
E   File "rknn\base\RKNNlib\app\medusa\workspace.py", line 592, in rknn.base.RKNNlib.app.medusa.workspace.Workspace._mmse_fine_tune_layers
E   File "rknn\base\RKNNlib\quantization\mmse_method.py", line 339, in rknn.base.RKNNlib.quantization.mmse_method.mmse_record_outputs
E TypeError: 'NoneType' object is not subscriptable

请问kl_divergence和normal哪种量化效果会更好些？