const int64_t id = target_tensor_ids[i];
    if (output_gradients.empty() || output_gradients[i] == nullptr) {
      auto tensor_it = tensor_tape.find(id);
      if (tensor_it != tensor_tape.end() && tensor_it->second != -1) {
        auto op_it = op_tape.find(tensor_it->second);
        if (op_it == op_tape.end()) {
          return errors::Internal(
              "Internal state of the gradient tape is invalid: "

# A representative sample is a map of: input_key -> input_value.
# Ex.: {'dense_input': tf.constant([1, 2, 3])}
# Ex.: {'x1': np.ndarray([4, 5, 6]}
RepresentativeSample = Mapping[str, core.TensorLike]

# A representative dataset is an iterable of representative samples.
RepresentativeDataset = Iterable[RepresentativeSample]

# A type representing a map from: signature key -> representative dataset.

      output << " input_index=" << xla_output_description.input_index;
    if (xla_output_description.is_constant) output << " constant";
    if (xla_output_description.is_tensor_list) output << " tensor_list";
    output << '\n';
  }

  for (const auto& resource_update : compilation_result.resource_updates) {
    output << "// ResourceUpdate input_index=" << resource_update.input_index

					mg.g.Require(m, append(reqs, vendorMod))
				}
				mg.g.Require(vendorMod, vendorList)

			} else {
				mainModule := MainModules.mustGetSingleMainModule()
				mg.g.Require(mainModule, append(rs.rootModules, vendorMod))
				mg.g.Require(vendorMod, vendorList)
			}
		}

		rs.graph.Store(&cachedGraph{mg, nil})
	})
}

		// return the full list of modules from modules.txt.
		// We don't know what versions the vendored module actually relies on,
		// so assume that it requires everything.
		summary.require = vendorList
		return summary, nil
	}

	actual := resolveReplacement(m)
	if mustHaveSums() && actual.Version != "" {
		key := module.Version{Path: actual.Path, Version: actual.Version + "/go.mod"}

*   Others
    *   Retain parent namescope for ops added inside
        `tf.while_loop`/`tf.cond`/`tf.switch_case`.
    *   Update `tf.vectorized_map` to support vectorizing `tf.while_loop` and
        TensorList operations.
    *   `tf.custom_gradient` can now be applied to functions that accept nested
        structures of `tensors` as inputs (instead of just a list of tensors).

			modFiles = append(modFiles, MainModules.ModFile(m))
			modRoots = append(modRoots, MainModules.ModRoot(m))
		}
		checkVendorConsistency(indexes, modFiles, modRoots)
		rs.initVendor(vendorList)
	}

	if inWorkspaceMode() {
		// We don't need to update the mod file so return early.
		requirements = rs
		return rs, nil
	}

	mainModule := MainModules.mustGetSingleMainModule()