diff --git a/ngraph_bridge/ngraph_builder.cc b/ngraph_bridge/ngraph_builder.cc
index 7c7f7c8fe..d015139a5 100644
--- a/ngraph_bridge/ngraph_builder.cc
+++ b/ngraph_bridge/ngraph_builder.cc
@@ -247,6 +247,9 @@ static Status TensorDataToVector(const Tensor& tensor, std::vector<T>* vector) {
   // Else we have to convert.
   else {
     switch (dt) {
+      case DT_HALF:
+        ConvertTensorDataToVector<Eigen::half, T>(tensor, vector);
+        break;
       case DT_FLOAT:
         ConvertTensorDataToVector<float, T>(tensor, vector);
         break;
@@ -350,7 +353,9 @@ Builder::TF_NGRAPH_CONST_MAP() {
           {DataType::DT_UINT16,
            make_pair(MakeConstOp<uint16>, ng::element::u16)},
           {DataType::DT_BOOL,
-           make_pair(MakeConstOp<bool, char>, ng::element::boolean)}};
+           make_pair(MakeConstOp<bool, char>, ng::element::boolean)},
+          {DataType::DT_HALF,
+           make_pair(MakeConstOp<float, ng::float16>, ng::element::f16)}};
   return the_map;
 }
 
diff --git a/ngraph_bridge/ngraph_mark_for_clustering.cc b/ngraph_bridge/ngraph_mark_for_clustering.cc
index fb1cf2bb6..8ab779822 100644
--- a/ngraph_bridge/ngraph_mark_for_clustering.cc
+++ b/ngraph_bridge/ngraph_mark_for_clustering.cc
@@ -77,9 +77,7 @@ static Status TypeConstraintOk(Node* node,
     for (const auto& name_and_set : itr->second) {
       auto& type_attr_name = name_and_set.first;
       auto& allowed_types = name_and_set.second;
-
       DataType dt;
-
       if (GetNodeAttr(node->attrs(), type_attr_name, &dt) != Status::OK() ||
           std::find(allowed_types.begin(), allowed_types.end(), dt) ==
               allowed_types.end()) {
@@ -566,6 +564,7 @@ const TypeConstraintMap& GetTypeConstraintMap() {
     type_constraint_map["NonMaxSuppressionV4"]["T"] = {
         DT_FLOAT};  // TF allows half too
     type_constraint_map["OneHot"]["T"] = NGraphDTypes();
+    type_constraint_map["OneHot"]["TI"] = NGraphIndexDTypes();
     type_constraint_map["Pack"]["T"] = NGraphDTypes();
     type_constraint_map["Pad"]["T"] = NGraphDTypes();
     type_constraint_map["Pad"]["Tpaddings"] = NGraphIndexDTypes();
diff --git a/ngraph_bridge/ngraph_utils.cc b/ngraph_bridge/ngraph_utils.cc
index d1aeb3975..1accaa7ac 100644
--- a/ngraph_bridge/ngraph_utils.cc
+++ b/ngraph_bridge/ngraph_utils.cc
@@ -293,23 +293,23 @@ void print_node_histogram(const std::unordered_map<string, int>& histogram,
 
 const gtl::ArraySlice<DataType>& NGraphDTypes() {
   static gtl::ArraySlice<DataType> result{
-      DT_FLOAT, DT_DOUBLE, DT_INT8,   DT_INT16,   DT_INT32,
-      DT_INT64, DT_UINT8,  DT_UINT16, DT_UINT32,  DT_UINT64,
-      DT_BOOL,  DT_QINT8,  DT_QUINT8, DT_BFLOAT16};
+      DT_FLOAT, DT_DOUBLE, DT_INT8,   DT_INT16,    DT_INT32,
+      DT_INT64, DT_UINT8,  DT_UINT16, DT_UINT32,   DT_UINT64,
+      DT_BOOL,  DT_QINT8,  DT_QUINT8, DT_BFLOAT16, DT_HALF};
   return result;
 }
 
 const gtl::ArraySlice<DataType>& NGraphNumericDTypes() {
   static gtl::ArraySlice<DataType> result{
-      DT_FLOAT, DT_DOUBLE, DT_INT8,   DT_INT16,  DT_INT32,   DT_INT64,
-      DT_UINT8, DT_UINT16, DT_UINT32, DT_UINT64, DT_BFLOAT16};
+      DT_FLOAT, DT_DOUBLE, DT_INT8,   DT_INT16,  DT_INT32,    DT_INT64,
+      DT_UINT8, DT_UINT16, DT_UINT32, DT_UINT64, DT_BFLOAT16, DT_HALF};
   return result;
 }
 
 const gtl::ArraySlice<DataType>& NGraphNumericAndQuantizedDTypes() {
   static gtl::ArraySlice<DataType> result{
-      DT_FLOAT, DT_DOUBLE, DT_INT8,   DT_INT16,  DT_INT32, DT_INT64,
-      DT_UINT8, DT_UINT16, DT_UINT32, DT_UINT64, DT_QINT8, DT_QUINT8};
+      DT_FLOAT,  DT_DOUBLE, DT_INT8,   DT_INT16, DT_INT32,  DT_INT64, DT_UINT8,
+      DT_UINT16, DT_UINT32, DT_UINT64, DT_QINT8, DT_QUINT8, DT_HALF};
   return result;
 }
 
@@ -330,7 +330,8 @@ const gtl::ArraySlice<DataType>& NGraphSupportedQuantizedDTypes() {
 }
 
 const gtl::ArraySlice<DataType>& NGraphRealDTypes() {
-  static gtl::ArraySlice<DataType> result{DT_FLOAT, DT_DOUBLE, DT_BFLOAT16};
+  static gtl::ArraySlice<DataType> result{DT_FLOAT, DT_DOUBLE, DT_BFLOAT16,
+                                          DT_HALF};
   return result;
 }
 
diff --git a/ngraph_bridge/ngraph_utils.h b/ngraph_bridge/ngraph_utils.h
index c43102d91..d0facaab1 100644
--- a/ngraph_bridge/ngraph_utils.h
+++ b/ngraph_bridge/ngraph_utils.h
@@ -97,7 +97,9 @@ Status ValuesFromConstNode(const NodeDef& node,
     return errors::InvalidArgument("Node not a Const");
   }
 
-  if (node.attr().at("dtype").type() != DataTypeToEnum<T>::value) {
+  auto dt = node.attr().at("dtype").type();
+  if (node.attr().count("dtype") == 0 || node.attr().count("value") == 0 ||
+      (dt != DT_HALF && dt != DataTypeToEnum<T>::value)) {
     std::stringstream ss;
     ss << "Invalid data type defined for Const. Defined: "
        << node.attr().at("dtype").type();
@@ -151,25 +153,29 @@ Status ValuesFromConstNode(const NodeDef& node,
       switch (dt) {
         // TODO(amprocte/NGRAPH-2502): there are more element types to support
         // here
+        case DT_HALF:
+          val_size = tensor.half_val_size();
+          if (val_size > 0) val_i = static_cast<T>(tensor.half_val()[i]);
+          break;
         case DT_INT32:
           val_size = tensor.int_val_size();
-          if (val_size > 0) val_i = tensor.int_val()[i];
+          if (val_size > 0) val_i = static_cast<T>(tensor.int_val()[i]);
           break;
         case DT_INT64:
           val_size = tensor.int64_val_size();
-          if (val_size > 0) val_i = tensor.int64_val()[i];
+          if (val_size > 0) val_i = static_cast<T>(tensor.int64_val()[i]);
           break;
         case DT_FLOAT:
           val_size = tensor.float_val_size();
-          if (val_size > 0) val_i = tensor.float_val()[i];
+          if (val_size > 0) val_i = static_cast<T>(tensor.float_val()[i]);
           break;
         case DT_BOOL:
           val_size = tensor.bool_val_size();
-          if (val_size > 0) val_i = tensor.bool_val()[i];
+          if (val_size > 0) val_i = static_cast<T>(tensor.bool_val()[i]);
           break;
         case DT_DOUBLE:
           val_size = tensor.double_val_size();
-          if (val_size > 0) val_i = tensor.double_val()[i];
+          if (val_size > 0) val_i = static_cast<T>(tensor.double_val()[i]);
           break;
         default:
           NGRAPH_VLOG(0)