class ModelWithDatetimeField(BaseModel):
        dt_field: datetime

        @field_serializer("dt_field")
        def serialize_datetime(self, dt_field: datetime):
            return dt_field.replace(microsecond=0, tzinfo=timezone.utc).isoformat()

    app = FastAPI()
    model = ModelWithDatetimeField(dt_field=datetime(2019, 1, 1, 8))

    @app.get("/model", response_model=ModelWithDatetimeField)

    class ModelWithCustomEncoder(BaseModel):
        dt_field: datetime

        @field_serializer("dt_field")
        def serialize_dt_field(self, dt):
            return dt.replace(microsecond=0, tzinfo=timezone.utc).isoformat()

    class ModelWithCustomEncoderSubclass(ModelWithCustomEncoder):
        pass

    model = ModelWithCustomEncoder(dt_field=datetime(2019, 1, 1, 8))

 *
 * <h2>Collection types</h2>
 *
 * <dl>
 *   <dt>{@link Multimap}
 *   <dd>A new type, which is similar to {@link java.util.Map}, but may contain multiple entries
 *       with the same key. Some behaviors of {@link Multimap} are left unspecified and are provided
 *       only by the subtypes mentioned below.
 *   <dt>{@link ListMultimap}

		return
	}
	visited[dtype] = true
	switch dt := dtype.(type) {
	case *dwarf.TypedefType:
		if strings.HasPrefix(dt.Name, "__builtin") {
			// Don't look inside builtin types. There be dragons.
			return
		}
		if !p.typedefs[dt.Name] {
			p.typedefs[dt.Name] = true
			p.typedefList = append(p.typedefList, typedefInfo{dt.Name, pos})
			p.recordTypedefs1(dt.Type, pos, visited)
		}
	case *dwarf.PtrType:

  EXPECT_EQ(TF_DOUBLE, static_cast<TF_DataType>(tensorflow::DT_DOUBLE));
  EXPECT_EQ(TF_INT32, static_cast<TF_DataType>(tensorflow::DT_INT32));
  EXPECT_EQ(TF_UINT8, static_cast<TF_DataType>(tensorflow::DT_UINT8));
  EXPECT_EQ(TF_INT16, static_cast<TF_DataType>(tensorflow::DT_INT16));
  EXPECT_EQ(TF_INT8, static_cast<TF_DataType>(tensorflow::DT_INT8));

}

TEST(TensorPjRtBufferUtilTest, GetPjRtCBufferFromTensorNoBuffer) {
  auto allocator = std::make_unique<AsyncValueAllocator>();
  tensorflow::Tensor tensor(allocator.get(), DT_FLOAT, {1});

  EXPECT_THAT(
      GetPjRtCBufferFromTensor(&tensor),
      StatusIs(error::INTERNAL, HasSubstr(absl::StrCat(
                                    "Input tensor does not have PjRtBuffer"))));
}

	}
	consumed = int(dirent.Reclen)
	if direntInode(dirent) == 0 { // File absent in directory.
		return
	}
	switch dirent.Type {
	case syscall.DT_REG:
		typ = 0
	case syscall.DT_DIR:
		typ = os.ModeDir
	case syscall.DT_LNK:
		typ = os.ModeSymlink
	default:
		// Skip all other file types. Revisit if/when this code needs
		// to handle such files, MinIO is only interested in

inline bool IsDtypeTrainable(DataType dtype) {
  switch (dtype) {
    case DT_HALF:
    case DT_BFLOAT16:
    case DT_FLOAT:
    case DT_DOUBLE:
    case DT_COMPLEX64:
    case DT_COMPLEX128:
    case DT_RESOURCE:
    case DT_VARIANT:
      return true;
    case DT_QINT8:
    case DT_QINT16:
    case DT_QINT32:
    case DT_QUINT8:
    case DT_QUINT16:
      return tensorflow::flags::Global()

      "    signature {"
      "      name: 'AddVariablesFunction'"
      "      input_arg {"
      "        name: 'var'"
      "        type: DT_RESOURCE"
      "      }"
      "      output_arg {"
      "        name: 'sum'"
      "        type: DT_FLOAT"
      "      }"
      "    }"
      "    node_def {"
      "      name: 'read0'"
      "      op: 'ReadVariableOp'"
      "      input: 'var'"

  AbstractTensorHandlePtr x;
  {
    tracing::TracingTensorHandle* x_raw = nullptr;
    PartialTensorShape shape;
    Status s = dyn_cast<tracing::TracingContext>(ctx.get())->AddParameter(
        DT_FLOAT, shape, &x_raw);
    ASSERT_EQ(errors::OK, s.code()) << s.message();
    x.reset(x_raw);
  }

  PartialTensorShape shape;
  Status s = x->Shape(&shape);
  ASSERT_EQ(errors::OK, s.code()) << s.message();