【Hackathon 5th No.97】add paddle tanh_shrink op

src/frontends/paddle/src/op/tanh_shrink.cpp

@@ -0,0 +1,21 @@
+// Copyright (C) 2018-2023 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+//
+
+#include "default_opset.hpp"
+#include "openvino/frontend/paddle/node_context.hpp"
+
+namespace ov {
+namespace frontend {
+namespace paddle {
+namespace op {
+NamedOutputs tanh_shrink(const NodeContext& node) {
+    const auto x = node.get_input("X");
+    const auto tanh = std::make_shared<default_opset::Tanh>(x);
+    return node.default_single_output_mapping({std::make_shared<default_opset::Subtract>(x, tanh)}, {"Out"});
+}
+
+}  // namespace op
+}  // namespace paddle
+}  // namespace frontend
+}  // namespace ov

src/frontends/paddle/src/op_table.cpp

@@ -114,6 +114,7 @@ OP_CONVERTER(strided_slice);
 OP_CONVERTER(sum);
 OP_CONVERTER(swish);
 OP_CONVERTER(tanh);
+OP_CONVERTER(tanh_shrink);
 OP_CONVERTER(tensor_array_to_tensor);
 OP_CONVERTER(tile);
 OP_CONVERTER(top_k_v2);
@@ -244,6 +245,7 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
             {"swish", op::swish},
             {"sync_batch_norm", op::batch_norm},
             {"tanh", op::tanh},
+            {"tanh_shrink", op::tanh_shrink},
             {"tensor_array_to_tensor", op::tensor_array_to_tensor},
             {"tile", op::tile},
             {"top_k_v2", op::top_k_v2},

src/frontends/paddle/tests/op_fuzzy.cpp

@@ -534,6 +534,8 @@ static const std::vector<std::string> models{
     std::string("swish_default_params"),
     std::string("swish_beta"),
     std::string("tanh"),
+    std::string("tanh_shrink_1"),
+    std::string("tanh_shrink_2"),
     std::string("tile_repeat_times_tensor"),
     std::string("tile_list_float32"),
     std::string("tile_list_int32"),

src/frontends/paddle/tests/test_models/gen_scripts/generate_tanh_shrink.py

@@ -0,0 +1,51 @@
+# Copyright (C) 2018-2023 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+#
+# tanh_shrink paddle model generator
+#
+import numpy as np
+import sys
+from save_model import saveModel
+
+
+def tanh_shrink(name: str, x):
+    import paddle
+
+    paddle.enable_static()
+
+    with paddle.static.program_guard(paddle.static.Program(), paddle.static.Program()):
+        node_x = paddle.static.data(name='x', shape=x.shape, dtype=x.dtype)
+        out = paddle.nn.functional.tanhshrink(node_x)
+
+        cpu = paddle.static.cpu_places(1)
+        exe = paddle.static.Executor(cpu[0])
+        # startup program will call initializer to initialize the parameters.
+        exe.run(paddle.static.default_startup_program())
+
+        outs = exe.run(feed={'x': x}, fetch_list=[out])
+
+        saveModel(
+            name,
+            exe,
+            feedkeys=['x'],
+            fetchlist=[out],
+            inputs=[x],
+            outputs=[outs[0]],
+            target_dir=sys.argv[1],
+        )
+
+    return outs[0]
+
+
+def main():
+    data = np.random.uniform(10, 20, [2, 3, 4]).astype(np.float32)
+    tanh_shrink("tanh_shrink_1", data)
+
+    data = np.random.uniform(-10, 20, [4, 3, 2]).astype(np.float32)
+    tanh_shrink("tanh_shrink_2", data)
+
+if __name__ == "__main__":
+    main()
+
+