diff --git a/optimum/exporters/openvino/convert.py b/optimum/exporters/openvino/convert.py
index df2885fd09..4ed66b6c1b 100644
--- a/optimum/exporters/openvino/convert.py
+++ b/optimum/exporters/openvino/convert.py
@@ -389,7 +389,10 @@ def ts_patched_forward(*args, **kwargs):
                 if patch_16bit_model:
                     from openvino.frontend.pytorch.patch_model import __make_16bit_traceable
 
+                    # frontend may riases confusing warnings about modules already patched if model splitted on several parts
+                    logging.disable(logging.WARNING)
                     __make_16bit_traceable(model)
+                    logging.disable(logging.NOTSET)
                 check_dummy_inputs_are_allowed(model, dummy_inputs)
                 input_info = _get_input_info(model, config, dummy_inputs)
                 ov_model = convert_model(
diff --git a/optimum/exporters/openvino/model_configs.py b/optimum/exporters/openvino/model_configs.py
index 108deed57d..823d9fa928 100644
--- a/optimum/exporters/openvino/model_configs.py
+++ b/optimum/exporters/openvino/model_configs.py
@@ -69,6 +69,7 @@
     GptNeoxJapaneseModelPatcher,
     GptNeoxModelPatcher,
     IBertModelPatcher,
+    InputEmbeddingPatcher,
     InternLM2Patcher,
     InternLMModelPatcher,
     InternVLChatImageEmbeddingModelPatcher,
@@ -1263,6 +1264,11 @@ def rename_ambiguous_inputs(self, inputs):
         model_inputs["input"] = inputs["input_ids"]
         return model_inputs
 
+    def patch_model_for_export(
+        self, model: Union["PreTrainedModel", "TFPreTrainedModel"], model_kwargs: Optional[Dict[str, Any]] = None
+    ) -> "ModelPatcher":
+        return InputEmbeddingPatcher(self, model, model_kwargs)
+
 
 class LlavaConfigBehavior(str, enum.Enum):
     LANGUAGE = "language"
diff --git a/optimum/exporters/openvino/model_patcher.py b/optimum/exporters/openvino/model_patcher.py
index b1aa7eaa94..af370eef58 100644
--- a/optimum/exporters/openvino/model_patcher.py
+++ b/optimum/exporters/openvino/model_patcher.py
@@ -2765,6 +2765,7 @@ def __exit__(self, exc_type, exc_value, traceback):
             self._model.pos_embed.forward = self._model.pos_embed._orig_forward
 
 
+<<<<<<< HEAD
 def _minicpmv_resampler_forward(self, image_feature, pos_embed, key_padding_mask):
     bs = image_feature.shape[0]
     image_feature = self.kv_proj(image_feature)  # B * L * D
@@ -2973,3 +2974,26 @@ def __exit__(self, exc_type, exc_value, traceback):
         if is_torch_version(">=", "2.0.0"):
             for layer in self._model.encoder.layers:
                 layer.self_attn.forward = layer.self_attn._orig_forward
+
+
+class InputEmbeddingPatcher(ModelPatcher):
+    def __init__(
+        self,
+        config: "OnnxConfig",
+        model: Union["PreTrainedModel", "TFPreTrainedModel"],
+        model_kwargs: Dict[str, Any],
+    ):
+        # making 16bit tracable overrides embeedings input signature these changes required to prevent this issue
+        model.__orig_forward = model.forward
+
+        def forward(self, input):
+            return self.__orig_forward(input)
+
+        model.forward = types.MethodType(forward, model)
+
+        super().__init__(config, model, model_kwargs)
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        super().__exit__(exc_type, exc_value, traceback)
+        self._model.forward = self._model.__orig_forward
+