failed(AddFloatAttr(func, attrs, "nms_iou_threshold", &fbb)) ||
      failed(AddFloatAttr(func, attrs, "y_scale", &fbb)) ||
      failed(AddFloatAttr(func, attrs, "x_scale", &fbb)) ||
      failed(AddFloatAttr(func, attrs, "h_scale", &fbb)) ||
      failed(AddFloatAttr(func, attrs, "w_scale", &fbb)))
    return failure();
  auto use_regular_nms =
      mlir::dyn_cast_or_null<BoolAttr>(attrs.get("use_regular_nms"));

  std::vector<int64_t> in_shape{input_type.getShape().vec()};
  std::vector<int64_t> out_shape{output_type.getShape().vec()};

  // If the reshape changes the number of dimensions so it cannot be interpreted
  // as a transpose.
  if (in_shape.size() != out_shape.size()) {
    return false;
  }

  in_shape.erase(std::remove(in_shape.begin(), in_shape.end(), 1),
                 in_shape.end());

  %0 = "tf.Const"() {value = dense<0.0> : tensor<f32>} : () -> tensor<f32>

Usá-lo em seu editor é o que realmente te mostra os benefícios do FastAPI, ver quão pouco código você tem que escrever, todas as conferências de tipo, auto completações etc.

---

## Instale o FastAPI

O primeiro passo é instalar o FastAPI.

Para o tutorial, você deve querer instalá-lo com todas as dependências e recursos opicionais.

<div class="termy">

```console
$ pip install "fastapi[all]"

    // perfectly with the specified min/max. Nudge them into the right range.
    auto min_scaled =
        rewriter.create<DivOp>(op.getLoc(), float_min, quant_to_float);
    auto min_scaled_sub =
        rewriter.create<SubOp>(op.getLoc(), quant_min, min_scaled);

    auto mid_rounded =
        rewriter.create<RoundOp>(op.getLoc(), scalar_ty, min_scaled_sub);

    ```
  }];

  let arguments = (ins
      F32:$out_scale,
      TFR_TensorListType:$in_scales);

  let results = (outs TFR_TensorType:$scale_factor);

  let assemblyFormat = [{
    `(` $out_scale `,` $in_scales `)` attr-dict `:` functional-type(operands, results)
  }];

  let hasCanonicalizer = 1;
}

	if _, ok := amount.SetString(value); !ok {
		return Quantity{}, ErrNumeric
	}

	// So that no one but us has to think about suffixes, remove it.
	if base == 10 {
		amount.SetScale(amount.Scale() + Scale(exponent).infScale())
	} else if base == 2 {
		// numericSuffix = 2 ** exponent
		numericSuffix := big.NewInt(1).Lsh(bigOne, uint(exponent))
		ub := amount.UnscaledBig()
		amount.SetUnscaledBig(ub.Mul(ub, numericSuffix))
	}

  ArrayRef<int64_t> in_shape = ranked_type.getShape();
  if (in_shape.empty() || in_shape[0] < 0) {
    return context_op->emitOpError()
           << "A map_outside_compilation op's input and output shapes must "
              "have rank at least one and the first dimension must be known.";
  }
  int64_t split_size = in_shape[0] / num_cores_per_replica;
  if (in_shape[0] % num_cores_per_replica != 0) {

## Crie um `main.py`
Copie o exemplo em um arquivo `main.py`:

```Python
{!../../../docs_src/security/tutorial001.py!}
```

## Execute-o

!!! info "informação"
	Primeiro, instale <a href="https://github.com/Kludex/python-multipart" class="external-link" target="_blank">`python-multipart`</a>.

	Ex: `pip install python-multipart`.

    "'h_scale' type: 'float'} attr : { name: 'max_classes_per_detection' "
    "type: 'int'} attr : { name: 'max_detections' type: 'int'} attr : { "
    "name: 'nms_iou_threshold' type: 'float'} attr : { name: "
    "'nms_score_threshold' type: 'float'} attr : { name: 'num_classes' type: "
    "'int'} attr : { name: 'w_scale' type: 'float'} attr : { name: 'x_scale' "