minReg := 14 // _savegpr0_{n}, _savegpr1_{n}, _savefpr_{n}, 14 <= n <= 31
	offMul := 4  // 1 instruction per register op.
	switch s[1] {
	case "savegpr0", "savegpr1", "savefpr":
	case "restgpr0", "restgpr1", "restfpr":
	case "savevr", "restvr":
		minReg = 20 // _savevr_{n} or _restvr_{n}, 20 <= n <= 31
		offMul = 8  // 2 instructions per register op.
	default:
		return 0, false // Not an abi func
	}
	n, e := strconv.Atoi(s[2])

	mc := o.newMetacache()
	meta := metaCacheRPC{meta: &mc, cancel: cancel, rpc: globalNotificationSys.restClientFromHash(pathJoin(o.Bucket, o.Prefix)), o: *o}

	// Save listing...
	go func() {
		if err := saver.saveMetaCacheStream(listCtx, &meta, saveCh); err != nil {
			meta.setErr(err.Error())
		}
		cancel()
	}()

	// Do listing...
	go func(o listPathOptions) {
		err := z.listMerged(listCtx, o, inCh)

    init_op_name: Name of the node for initialization.
    saver_def: `saver_pb2.SaverDef` to create a `saver.Saver` from. The created
      saver will be used to save and load variables. This may be `None` if no
      variables exist in the graph.
    checkpoint_dir: Path to checkpoint file where variable values are saved.
    function_aliases: Function name -> function alias mapping.

    return "Inserts a new function that wraps a SaveV2 op. The SaveV2 op saves "
           "the values of the VarHandleOps that are found in the initializer "
           "function of 'restore_op' type.";
  }

  void runOnOperation() override;
};

// Finds `tf.AssignVariableOp(tf.VarHandleOp, tf.Const)` patterns and removes
// `tf.AssignVariableOp`s and `tf.Const`s. Collects and returns the

	MOVV	R11, 88(R3)
	MOVV	R12, 96(R3)
	// R13 already saved
	MOVV	R14, 104(R3)
	MOVV	R15, 112(R3)
	MOVV	R16, 120(R3)
	MOVV	R17, 128(R3)
	MOVV	R18, 136(R3)
	// R19 already saved
	MOVV	R20, 144(R3)
	MOVV	R21, 152(R3)
	// R22 is g.
	MOVV	R23, 160(R3)
	MOVV	R24, 168(R3)
	MOVV	R25, 176(R3)
	MOVV	R26, 184(R3)
	// R27 already saved
	// R28 already saved.
	MOVV	R29, 192(R3)
	// R30 is tmp register.

class RepresentativeDatasetSaver:
  """Representative dataset saver.

  Exposes a single method `save` that saves the provided representative dataset
  into files.

  This is useful when you would like to keep a snapshot of your representative
  dataset at a file system or when you need to pass the representative dataset
  as files.
  """

  def save(
      self, representative_dataset: RepresentativeDatasetMapping

        "test_graph_tfadd_with_ckpt.ckpt",
        "test_graph_tfadd_with_ckpt.pb",
        "test_graph_tfadd_with_ckpt_saver.ckpt",
        "test_graph_tfadd_with_ckpt_saver.pb",
        "test_graph_tfadd_with_ckpt_saver.saver",
        "test_graph_tfassert_eq.pb",
        "test_graph_tfcond.pb",
        "test_graph_tffunction.pb",
        "test_graph_tfgather.pb",
        "test_graph_tfmatmul.pb",
        "test_graph_tfmatmulandadd.pb",

// RUN: stablehlo-quant-opt %s -split-input-file -stablehlo-insert-calibration-statistics-saver='aggregator-ops-to-ignore=skipping_id' | FileCheck %s

func.func @serving_default(%arg0: tensor<1x3x4x3xf32>) -> (tensor<1x2x2x2xf32>) attributes {tf.entry_function = {control_outputs = "", inputs = "serving_default_input_tensor:0", outputs = "PartitionedCall:0"}} {

## Documents OpenAPI et API

Si vous renvoyez directement des codes HTTP et des réponses supplémentaires, ils ne seront pas inclus dans le schéma OpenAPI (la documentation de l'API), car FastAPI n'a aucun moyen de savoir à l'avance ce que vous allez renvoyer.

// Creates a pass that creates a new function that wraps the newly created
// SaveV2 op. The new function's name is "tf_quant__save". The function accepts
// a single string tensor as argument, which specifies the path to the
// checkpoint to which the variable's tensor values are saved. It finds
// `tf.AssignVariableOp(tf.VarHandleOp, tf.Const)` pattern in the initializer