dlm_tensor->dl_tensor.device = tf_dlm_context;
  int ndim = tensor->dims();
  dlm_tensor->dl_tensor.ndim = ndim;
  dlm_tensor->dl_tensor.data = tf_dlm_data;
  dlm_tensor->dl_tensor.dtype = tf_dlm_type;

  std::vector<int64_t>* shape_arr = &tf_dlm_tensor_ctx->shape;
  std::vector<int64_t>* stride_arr = &tf_dlm_tensor_ctx->strides;
  shape_arr->resize(ndim);
  stride_arr->resize(ndim, 1);
  for (int i = 0; i < ndim; i++) {

Ayrıca, API'ınızla iletişim kuracak önyüz, mobil veya IoT uygulamaları gibi istemciler için otomatik olarak kod oluşturabilirsiniz.

## Adım Adım Özetleyelim

### Adım 1: `FastAPI`yı Projemize Dahil Edelim

```Python hl_lines="1"
{!../../../docs_src/first_steps/tutorial001.py!}
```

`FastAPI`, API'niz için tüm işlevselliği sağlayan bir Python sınıfıdır.

	_, ok = ps[policy2]
	if !ok {
		c.Fatalf("policy was missing!")
	}

	err = s.adm.RemoveUser(ctx, accessKey)
	if err != nil {
		c.Fatalf("user could not be deleted: %v", err)
	}

	err = s.adm.RemoveCannedPolicy(ctx, policy1)
	if err != nil {
		c.Fatalf("policy del err: %v", err)
	}

	err = s.adm.RemoveCannedPolicy(ctx, policy2)
	if err != nil {
		c.Fatalf("policy del err: %v", err)
	}
}

Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint
occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

## Decision

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna

Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint
occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

## Decision

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna

        "Tensor expected to have rank 2 found rank: ", shape.dims());
  }
  int64_t dim_sizes[] = {2, 4};
  for (int i = 0; i < shape.dims(); i++) {
    if (shape.dim_size(i) != dim_sizes[i]) {
      return errors::InvalidArgument("Dim ", i, " expected to be of size ",
                                     dim_sizes[i],
                                     " found: ", shape.dim_size(i));
    }
  }
  return absl::OkStatus();
}

  auto num_dims_analytical = TF_NumDims(analytical_tensor);
  ASSERT_EQ(dims.size(), num_dims_analytical);
  for (int j = 0; j < num_dims_analytical; j++) {
    auto dim_analytical = TF_Dim(analytical_tensor, j);
    ASSERT_EQ(dims[j], dim_analytical);
    num_elem_analytical *= dim_analytical;
  }

  float* danalytical = new float[num_elem_analytical]{0};
  memcpy(&danalytical[0], TF_TensorData(analytical_tensor),

   "Resource": [
    "arn:aws:s3:::${aws:username}-*"
   ]
  }
 ]
}`)
	err = s.adm.AddCannedPolicy(ctx, policy, policyBytes)
	if err != nil {
		c.Fatalf("policy add error: %v", err)
	}

	if err = s.adm.AddUser(ctx, "dillon", "dillon-123"); err != nil {
		c.Fatalf("policy add error: %v", err)
	}

	err = s.adm.SetPolicy(ctx, policy, "dillon", false)
	if err != nil {
		c.Fatalf("Unable to set policy: %v", err)

For now we just have this global repository of ADRs.
If we see fit, we can break these out to per-platform ones, or keep a hybrid approach to having global and platform-specific ADSs.

Our aim is to keep the process lightweight and approachable.

  }
  auto imm_ctx = dyn_cast<ImmediateExecutionContext>(ctx);
  AbstractTensorPtr minus_1(imm_ctx->CreateInt32Scalar(-1));
  ImmediateTensorHandlePtr dim(imm_ctx->CreateLocalHandle(minus_1.get()));
  AbstractTensorHandle* expand_dims_outputs;
  TF_RETURN_IF_ERROR(
      ops::ExpandDims(ctx, vec, dim.get(), &expand_dims_outputs, "ExpandDims"));
  TF_RETURN_IF_ERROR(
      ops::Mul(ctx, expand_dims_outputs, mat, &outputs[0], "Mul"));