Update on "[Ez] Enable Vulkan 4-bit weight only quantization in export_llama"

As title.

Differential Revision: [D64406456](https://our.internmc.facebook.com/intern/diff/D64406456/)

[ghstack-poisoned]

Author: SS-JIA

backends/vulkan/test/op_tests/linear_weight_int4_test.cpp

@@ -202,15 +202,15 @@ void test_vulkan_linear_int4(
   ASSERT_TRUE(at::allclose(vk_out, out_ref, 1e-4, 1e-4));
 }
 
-TEST(VulkanSDPATest, test_reference_impl) {
+TEST(VulkanInt4LinearTest, test_reference_impl) {
   test_reference_linear_int4(
       /*B = */ 1,
       /*M = */ 4,
       /*K = */ 128,
       /*N = */ 32);
 }
 
-TEST(VulkanSDPATest, test_vulkan_impl) {
+TEST(VulkanInt4LinearTest, test_vulkan_impl) {
   if (!vkcompute::api::context()
            ->adapter_ptr()
            ->has_full_int8_buffers_support()) {