/* Want to test simple MatMul example:
    [[0,0],    *   [[0,0],    =   [[0,0],
     [0,0]]         [0,0]]         [0,0]]
  */

  // Build an abstract input tensor.
  int64_t dims[] = {2, 2};  // Matrices will be 2 x 2
  int num_dims = sizeof(dims) / sizeof(dims[0]);

  float vals[] = {0.0f, 0.0f, 0.0f, 0.0f};
  TFE_Context* eager_ctx = TF_ExecutionContextGetTFEContext(ctx, status.get());
  TFE_TensorHandle* t =

A queue is used to give a few properties:
* Ability to serially process updates received from a variety of different sources. This avoids need for other synchronization mechanisms like mutexes.
* Correctness at startup; with the sequencing above, items are only processed once all informers are synced. This means queries will not return stale data at startup.
* Deduping of identical events

* **Machine Learning**: Normalmente exige muita multiplicação de matrizes e vetores. Pense numa grande folha de papel com números e multiplicando todos eles juntos e ao mesmo tempo.

  TFE_DeleteContextOptions(opts);

  TFE_ContextSetServerDef(ctx, 0, serialized.data(), serialized.size(), status);
  EXPECT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  // Use large matrices so that RPCs don't return before we get a chance
  // to call TFE_DeleteContext.
  TFE_TensorHandle* h0_task0 = TestMatrixTensorHandle100x100(ctx);
  TFE_TensorHandle* h1_task0 = TestMatrixTensorHandle100x100(ctx);

* L'apprentissage automatique (ou **Machine Learning**) : cela nécessite de nombreuses multiplications de matrices et vecteurs. Imaginez une énorme feuille de calcul remplie de nombres que vous multiplierez entre eux tous au même moment.

* **Machine Learning**: normalmente requiere muchas multiplicaciones de "matrices" y "vectores". Imagina en una enorme hoja de cálculo con números y tener que multiplicarlos todos al mismo tiempo.

    matrices or batches of matrices (CPU only).
*   Added gradients for eigenvalues and eigenvectors computed using
    `self_adjoint_eig` or `self_adjoint_eigvals`.
*   Eliminated `batch_*` methods for most linear algebra and FFT ops and
    promoted the non-batch version of the ops to handle batches of matrices.