IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[MO, TF] Support Custom Wide and Deep CTR model by MO (#8505)

Browse Source 
* [MO, TF] Support Custom Wide and Deep CTR model by MO

It implements implicit support of EmbeddingSegmentsMean operation through decomposition.
Also, this extends the current transformation to fuse TensorFlow sub-graph (for Wide and Deep model family)
containing SparseSegmentSum and SparseSegmentMean operations into EmbeddingSegmentsSum or EmbeddingSegmentsMean.

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Fix unit-tests after modifications of SparseToDense and EmbeddingSegmentsOperationFusing

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Document SparseSegmentMean support

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>

* Add computation scheme for normalization coeffs and correct documentation

Signed-off-by: Roman Kazantsev <roman.kazantsev@intel.com>
This commit is contained in:
Roman Kazantsev
2021-11-17 11:44:04 +03:00
committed by GitHub
parent c307f206dc
commit 8e327bd2ff
10 changed files with 329 additions and 128 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
docs/MO_DG/prepare_model/Supported_Frameworks_Layers.md
View File

@@ -304,6 +304,7 @@ Some TensorFlow\* operations do not match to any Inference Engine layer, but are
| SparseFillEmptyRows | Supported only when it is part of a sub-graph of the special form |
| SparseReshape | Supported only when it is part of a sub-graph of the special form |
| SparseSegmentSum | Supported only when it is part of a sub-graph of the special form |
| SparseSegmentMean | Supported only when it is part of a sub-graph of the special form |
| SparseToDense | CPU only |
| Split | |
| SplitV | |

48
docs/MO_DG/prepare_model/convert_model/tf_specific/Convert_WideAndDeep_Family_Models.md
View File

@@ -99,30 +99,30 @@ python mo.py
IteratorGetNext:2[2],
IteratorGetNext:4[2],
IteratorGetNext:7[2],
linear/linear_model/linear_model/linear_model/education/to_sparse_input/indices:0[10 2]{i32},
linear/linear_model/linear_model/linear_model/education/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
linear/linear_model/linear_model/linear_model/education/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
linear/linear_model/linear_model/linear_model/marital_status/to_sparse_input/indices:0[10 2]{i32},
linear/linear_model/linear_model/linear_model/marital_status/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
linear/linear_model/linear_model/linear_model/marital_status/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
linear/l inear_model/linear_model/linear_model/relationship/to_sparse_input/indices:0[10 2]{i32},
linear/linear_model/linear_model/linear_model/relationship/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
linear/linear_model/linear_model/linear_model/relationship/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
linear/linear_model/linear_model/linear_model/workclass/to_sparse_input/indices:0[10 2]{i32},
linear/linear_model/linear_model/linear_model/workclass/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
linear/linear_model/linear_model/linear_model/workclass/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
dnn/input_from_feature_columns/input_layer/education_indicator/to_sparse_input/indices:0[10 2]{i32},
dnn/input_from_feature_columns/input_layer/education_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
dnn/input_from_feature_columns/input_layer/education_indicator/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
dnn/input_from_feature_columns/input_layer/marital_status_indicator/to_sparse_input/indices:0[10 2]{i32},
dnn/input_from_feature_columns/input_layer/marital_status_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
dnn/input_from_feature_columns/input_layer/marital_status_indicator/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
dnn/input_from_feature_columns/input_layer/relationship_indicator/to_sparse_input/indices:0[10 2]{i32},
dnn/input_from_feature_columns/input_layer/relationship_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
dnn/input_from_feature_columns/input_layer/relationship_indicator/to_sparse_input/dense_shape:0[2]{i32}->[2 50],
dnn/input_from_feature_columns/input_layer/workclass_indicator/to_sparse_input/indices:0[10 2]{i32},
dnn/input_from_feature_columns/input_layer/workclass_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i32},
dnn/input_from_feature_columns/input_layer/workclass_indicator/to_sparse_input/dense_shape:0[2]{i32}->[2 50]"
linear/linear_model/linear_model/linear_model/education/to_sparse_input/indices:0[10 2]{i64},
linear/linear_model/linear_model/linear_model/education/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
linear/linear_model/linear_model/linear_model/education/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
linear/linear_model/linear_model/linear_model/marital_status/to_sparse_input/indices:0[10 2]{i64},
linear/linear_model/linear_model/linear_model/marital_status/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
linear/linear_model/linear_model/linear_model/marital_status/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
linear/l inear_model/linear_model/linear_model/relationship/to_sparse_input/indices:0[10 2]{i64},
linear/linear_model/linear_model/linear_model/relationship/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
linear/linear_model/linear_model/linear_model/relationship/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
linear/linear_model/linear_model/linear_model/workclass/to_sparse_input/indices:0[10 2]{i64},
linear/linear_model/linear_model/linear_model/workclass/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
linear/linear_model/linear_model/linear_model/workclass/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
dnn/input_from_feature_columns/input_layer/education_indicator/to_sparse_input/indices:0[10 2]{i64},
dnn/input_from_feature_columns/input_layer/education_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
dnn/input_from_feature_columns/input_layer/education_indicator/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
dnn/input_from_feature_columns/input_layer/marital_status_indicator/to_sparse_input/indices:0[10 2]{i64},
dnn/input_from_feature_columns/input_layer/marital_status_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
dnn/input_from_feature_columns/input_layer/marital_status_indicator/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
dnn/input_from_feature_columns/input_layer/relationship_indicator/to_sparse_input/indices:0[10 2]{i64},
dnn/input_from_feature_columns/input_layer/relationship_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
dnn/input_from_feature_columns/input_layer/relationship_indicator/to_sparse_input/dense_shape:0[2]{i64}->[2 50],
dnn/input_from_feature_columns/input_layer/workclass_indicator/to_sparse_input/indices:0[10 2]{i64},
dnn/input_from_feature_columns/input_layer/workclass_indicator/hash_table_Lookup/LookupTableFindV2:0[10]{i64},
dnn/input_from_feature_columns/input_layer/workclass_indicator/to_sparse_input/dense_shape:0[2]{i64}->[2 50]"
--output head/predictions/probabilities
```