diff --git a/onnxruntime/core/framework/kernel_registry.cc b/onnxruntime/core/framework/kernel_registry.cc
index cb6e8bc83f..2b73bd6d19 100644
--- a/onnxruntime/core/framework/kernel_registry.cc
+++ b/onnxruntime/core/framework/kernel_registry.cc
@@ -255,7 +255,7 @@ const KernelCreateInfo* KernelRegistry::TryFindKernel(const onnxruntime::Node& n
   const auto& expected_provider = (node_provider.empty() ? exec_provider : node_provider);
 
   auto range = kernel_creator_fn_map_.equal_range(GetMapKey(node.OpType(), node.Domain(), expected_provider));
-  std::vector<std::string> error_strs;
+  std::vector<std::string> verify_kernel_def_error_strs;
   for (auto i = range.first; i != range.second; ++i) {
     if (!i->second.status.IsOK()) {
       LOGS_DEFAULT(ERROR) << "Failed to create kernel for op: " << node.OpType()
@@ -266,10 +266,13 @@ const KernelCreateInfo* KernelRegistry::TryFindKernel(const onnxruntime::Node& n
     if (VerifyKernelDef(node, *i->second.kernel_def, error_str)) {
       return &i->second;
     }
-    error_strs.push_back(error_str);
+    verify_kernel_def_error_strs.push_back(error_str);
+  }
+
+  if (!verify_kernel_def_error_strs.empty()) {
+    LOGS_DEFAULT(INFO) << node.OpType() << " kernel is not supported in " << expected_provider
+                       << " Encountered following errors: (" << ToString(verify_kernel_def_error_strs) << ")";
   }
-  LOGS_DEFAULT(INFO) << node.OpType() << " kernel is not supported in " << expected_provider
-                     << " Encountered following errors: " << ToString(error_strs);
   return nullptr;
 }
 
diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index b22304fd1f..f4edd3b787 100644
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -76,10 +76,15 @@ CUDAExecutionProvider::CUDAExecutionProvider(const CUDAExecutionProviderInfo& in
 
   // TODO: this is actually used for the cuda kernels which explicitly ask for inputs from CPU.
   // This will be refactored/removed when allocator and execution provider are decoupled.
-  DeviceAllocatorRegistrationInfo cpu_memory_info({
-      OrtMemTypeCPUInput,
-      [](int device_id) { return onnxruntime::make_unique<CPUAllocator>(onnxruntime::make_unique<OrtMemoryInfo>("CUDA_CPU", OrtAllocatorType::OrtDeviceAllocator, OrtDevice(), device_id, OrtMemTypeCPUInput)); },
-      std::numeric_limits<size_t>::max()});
+  DeviceAllocatorRegistrationInfo cpu_memory_info({OrtMemTypeCPUInput,
+                                                   [](int device_id) { return onnxruntime::make_unique<CPUAllocator>(
+                                                                           onnxruntime::make_unique<OrtMemoryInfo>(
+                                                                               "CUDA_CPU",
+                                                                               OrtAllocatorType::OrtDeviceAllocator,
+                                                                               OrtDevice(),
+                                                                               device_id,
+                                                                               OrtMemTypeCPUInput)); },
+                                                   std::numeric_limits<size_t>::max()});
   InsertAllocator(CreateAllocator(cpu_memory_info, CPU_ALLOCATOR_DEVICE_ID));
 }
 
@@ -499,8 +504,10 @@ class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain,
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, MLFloat16, LSTM);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, int32_t, Slice);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, int64_t, Slice);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, float, Slice);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, int32_t, Slice);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, int64_t, Slice);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, Slice);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, Compress);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, Flatten);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, float, Upsample);
@@ -850,8 +857,10 @@ static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, MLFloat16, LSTM)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, int32_t, Slice)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, int64_t, Slice)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 9, float, Slice)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, int32_t, Slice)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, int64_t, Slice)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, Slice)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, Compress)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, Flatten)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, float, Upsample)>,
diff --git a/onnxruntime/core/providers/cuda/tensor/slice.cc b/onnxruntime/core/providers/cuda/tensor/slice.cc
index 30a4152405..d7a8f632c2 100644
--- a/onnxruntime/core/providers/cuda/tensor/slice.cc
+++ b/onnxruntime/core/providers/cuda/tensor/slice.cc
@@ -20,6 +20,7 @@ namespace cuda {
 
 REGISTER_VERSIONED_TYPED_SLICE(int32_t) 
 REGISTER_VERSIONED_TYPED_SLICE(int64_t)
+REGISTER_VERSIONED_TYPED_SLICE(float)
 
 #define REGISTER_V10_TYPED_SLICE(TIND)                                                    \
   ONNX_OPERATOR_TYPED_KERNEL_EX(                                                          \
@@ -38,6 +39,7 @@ REGISTER_VERSIONED_TYPED_SLICE(int64_t)
 
 REGISTER_V10_TYPED_SLICE(int32_t) 
 REGISTER_V10_TYPED_SLICE(int64_t)
+REGISTER_V10_TYPED_SLICE(float)
 
 template<typename Tind, bool dynamic>
 Status Slice<Tind, dynamic>::ComputeInternal(OpKernelContext* ctx) const {