使用paddleseg中的slim剪枝Bisenetv2出现AssertionError: The length of current_mask must be equal to the size of dimension to be pruned on. about paddlex

[zhaoyj96]

2023-11-20 03:12:52,715-WARNING: Leaves ['batch_norm_0.tmp_0', 'batch_norm_0.tmp_1', 'batch_norm_0.tmp_3', 'batch_norm_1.tmp_0', 'batch_norm_1.tmp_1', 'batch_norm_1.tmp_3', 'batch_norm_2.tmp_0', 'batch_norm_2.tmp_1', 'batch_norm_2.tmp_3', 'batch_norm_3.tmp_0', 'batch_norm_3.tmp_1', 'batch_norm_3.tmp_3', 'batch_norm_4.tmp_0', 'batch_norm_4.tmp_1', 'batch_norm_4.tmp_3', 'batch_norm_5.tmp_0', 'batch_norm_5.tmp_1', 'batch_norm_5.tmp_3', 'batch_norm_6.tmp_0', 'batch_norm_6.tmp_1', 'batch_norm_6.tmp_3', 'batch_norm_7.tmp_0', 'batch_norm_7.tmp_1', 'batch_norm_7.tmp_3', 'batch_norm_8.tmp_0', 'batch_norm_8.tmp_1', 'batch_norm_8.tmp_3', 'batch_norm_9.tmp_0', 'batch_norm_9.tmp_1', 'batch_norm_9.tmp_3', 'batch_norm_10.tmp_0', 'batch_norm_10.tmp_1', 'batch_norm_10.tmp_3', 'pool2d_0.tmp_1', 'batch_norm_11.tmp_0', 'batch_norm_11.tmp_1', 'batch_norm_11.tmp_3', 'batch_norm_12.tmp_0', 'batch_norm_12.tmp_1', 'batch_norm_12.tmp_3', 'batch_norm_13.tmp_0', 'batch_norm_13.tmp_1', 'batch_norm_13.tmp_3', 'batch_norm_14.tmp_0', 'batch_norm_14.tmp_1', 'batch_norm_14.tmp_3', 'batch_norm_15.tmp_0', 'batch_norm_15.tmp_1', 'batch_norm_15.tmp_3', 'batch_norm_16.tmp_0', 'batch_norm_16.tmp_1', 'batch_norm_16.tmp_3', 'batch_norm_17.tmp_0', 'batch_norm_17.tmp_1', 'batch_norm_17.tmp_3', 'batch_norm_18.tmp_0', 'batch_norm_18.tmp_1', 'batch_norm_18.tmp_3', 'batch_norm_19.tmp_0', 'batch_norm_19.tmp_1', 'batch_norm_19.tmp_3', 'batch_norm_20.tmp_0', 'batch_norm_20.tmp_1', 'batch_norm_20.tmp_3', 'batch_norm_21.tmp_0', 'batch_norm_21.tmp_1', 'batch_norm_21.tmp_3', 'batch_norm_22.tmp_0', 'batch_norm_22.tmp_1', 'batch_norm_22.tmp_3', 'batch_norm_23.tmp_0', 'batch_norm_23.tmp_1', 'batch_norm_23.tmp_3', 'batch_norm_24.tmp_0', 'batch_norm_24.tmp_1', 'batch_norm_24.tmp_3', 'batch_norm_25.tmp_0', 'batch_norm_25.tmp_1', 'batch_norm_25.tmp_3', 'batch_norm_26.tmp_0', 'batch_norm_26.tmp_1', 'batch_norm_26.tmp_3', 'batch_norm_27.tmp_0', 'batch_norm_27.tmp_1', 'batch_norm_27.tmp_3', 'batch_norm_28.tmp_0', 'batch_norm_28.tmp_1', 'batch_norm_28.tmp_3', 'batch_norm_29.tmp_0', 'batch_norm_29.tmp_1', 'batch_norm_29.tmp_3', 'batch_norm_30.tmp_0', 'batch_norm_30.tmp_1', 'batch_norm_30.tmp_3', 'batch_norm_31.tmp_0', 'batch_norm_31.tmp_1', 'batch_norm_31.tmp_3', 'batch_norm_32.tmp_0', 'batch_norm_32.tmp_1', 'batch_norm_32.tmp_3', 'batch_norm_33.tmp_0', 'batch_norm_33.tmp_1', 'batch_norm_33.tmp_3', 'batch_norm_34.tmp_0', 'batch_norm_34.tmp_1', 'batch_norm_34.tmp_3', 'batch_norm_35.tmp_0', 'batch_norm_35.tmp_1', 'batch_norm_35.tmp_3', 'batch_norm_36.tmp_0', 'batch_norm_36.tmp_1', 'batch_norm_36.tmp_3', 'batch_norm_37.tmp_0', 'batch_norm_37.tmp_1', 'batch_norm_37.tmp_3', 'batch_norm_38.tmp_0', 'batch_norm_38.tmp_1', 'batch_norm_38.tmp_3', 'batch_norm_39.tmp_0', 'batch_norm_39.tmp_1', 'batch_norm_39.tmp_3', 'batch_norm_40.tmp_0', 'batch_norm_40.tmp_1', 'batch_norm_40.tmp_3', 'batch_norm_41.tmp_0', 'batch_norm_41.tmp_1', 'batch_norm_41.tmp_3', 'batch_norm_42.tmp_0', 'batch_norm_42.tmp_1', 'batch_norm_42.tmp_3', 'batch_norm_43.tmp_0', 'batch_norm_43.tmp_1', 'batch_norm_43.tmp_3', 'batch_norm_44.tmp_0', 'batch_norm_44.tmp_1', 'batch_norm_44.tmp_3', 'batch_norm_45.tmp_0', 'batch_norm_45.tmp_1', 'batch_norm_45.tmp_3', 'batch_norm_46.tmp_0', 'batch_norm_46.tmp_1', 'batch_norm_46.tmp_3', 'batch_norm_47.tmp_0', 'batch_norm_47.tmp_1', 'batch_norm_47.tmp_3', 'batch_norm_48.tmp_0', 'batch_norm_48.tmp_1', 'batch_norm_48.tmp_3', 'batch_norm_49.tmp_0', 'batch_norm_49.tmp_1', 'batch_norm_49.tmp_3', 'batch_norm_50.tmp_0', 'batch_norm_50.tmp_1', 'batch_norm_50.tmp_3', 'batch_norm_51.tmp_0', 'batch_norm_51.tmp_1', 'batch_norm_51.tmp_3', 'batch_norm_52.tmp_0', 'batch_norm_52.tmp_1', 'batch_norm_52.tmp_3', 'dropout_0.tmp_1', 'batch_norm_53.tmp_0', 'batch_norm_53.tmp_1', 'batch_norm_53.tmp_3', 'dropout_1.tmp_1', 'batch_norm_54.tmp_0', 'batch_norm_54.tmp_1', 'batch_norm_54.tmp_3', 'dropout_2.tmp_1', 'batch_norm_55.tmp_0', 'batch_norm_55.tmp_1', 'batch_norm_55.tmp_3', 'dropout_3.tmp_1', 'batch_norm_56.tmp_0', 'batch_norm_56.tmp_1', 'batch_norm_56.tmp_3', 'dropout_4.tmp_1', 'bilinear_interp_v2_2.tmp_0', 'bilinear_interp_v2_3.tmp_0', 'bilinear_interp_v2_4.tmp_0', 'bilinear_interp_v2_5.tmp_0', 'bilinear_interp_v2_6.tmp_0'] will be skipped when parsing graph.
2023-11-20 03:12:57,790-WARNING: ('Variable bilinear_interp_v2_4.tmp_0 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable bilinear_interp_v2_6.tmp_0 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable bilinear_interp_v2_3.tmp_0 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable bilinear_interp_v2_2.tmp_0 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Unsupported operator named concat',)
2023-11-20 03:12:57,790-WARNING: ('Variable dropout_2.tmp_1 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable dropout_3.tmp_1 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable dropout_4.tmp_1 was skipped.',)
2023-11-20 03:12:57,790-WARNING: ('Variable dropout_0.tmp_1 was skipped.',)
2023-11-20 03:12:57,791-WARNING: ('Variable dropout_1.tmp_1 was skipped.',)
2023-11-20 03:12:57,791-WARNING: ('Variable bilinear_interp_v2_5.tmp_0 was skipped.',)
2023-11-20 03:12:57,791-INFO: Found 24 collections.
2023-11-20 03:12:57,823-INFO: Load status from F:\1_DLW\PaddleSeg-release-2.8\output\Bisenetv2-prune0.2\sen.pickle
2023-11-20 03:13:02,746-INFO: Pruning variable [conv2d_12.w_0] and its relatives ['conv2d_12.w_0', 'conv2d_12.b_0', 'batch_norm2d_12.w_0', 'batch_norm2d_12.b_0', 'batch_norm2d_12.w_1', 'batch_norm2d_12.w_2', 'conv2d_13.w_0', 'conv2d_13.b_0', 'batch_norm2d_13.w_0', 'batch_norm2d_13.b_0', 'batch_norm2d_13.w_1', 'batch_norm2d_13.w_2', 'conv2d_14.w_0', 'conv2d_14.b_0', 'batch_norm2d_14.w_0', 'batch_norm2d_14.b_0', 'batch_norm2d_14.w_1', 'batch_norm2d_14.w_2', 'conv2d_15.w_0']
Traceback (most recent call last):
File "F:/1_DLW/PaddleSeg-release-2.8/deploy/slim/prune/prune1.py", line 213, in
main(args)
File "F:/1_DLW/PaddleSeg-release-2.8/deploy/slim/prune/prune1.py", line 164, in main
pruner.sensitive(
File "C:\Users\dell\anaconda3\envs\paddle2.4\lib\site-packages\paddleslim\dygraph\prune\filter_pruner.py", line 124, in sensitive
self._cal_sensitive(
File "C:\Users\dell\anaconda3\envs\paddle2.4\lib\site-packages\paddleslim\dygraph\prune\filter_pruner.py", line 266, in _cal_sensitive
plan = self.prune_var(var_name, dims, ratio)
File "C:\Users\dell\anaconda3\envs\paddle2.4\lib\site-packages\paddleslim\dygraph\prune\filter_pruner.py", line 346, in prune_var
assert len(current_mask) == var_shape[
AssertionError: The length of current_mask must be equal to the size of dimension to be pruned on. But get: len(current_mask): 16; var_shape: (96,); axis: 0; var name: conv2d_13.b_0; len(mask): 16