-> (tensor<?x256x56x56xf32>, tensor<256xf32>, tensor<256xf32>, tensor<256xf32>, tensor<256xf32>, tensor<*xf32>)

  // CHECK: %[[BATCH_NORM1:[_a-z0-9]*]], {{.*}} = "tf.FusedBatchNormV3"
  // CHECK-SAME: %[[CONV1]]
  // CHECK-SAME: data_format = "NHWC"

  // ------------------------------------------------------------------------ //
  // Convolution layer #2.

func (bto *BootstrapTokenOptions) AddTokenFlag(fs *pflag.FlagSet) {
	fs.StringVar(
		&bto.TokenStr, TokenStr, "",
		"The token to use for establishing bidirectional trust between nodes and control-plane nodes. The format is [a-z0-9]{6}\\.[a-z0-9]{16} - e.g. abcdef.0123456789abcdef",
	)
}

// AddTTLFlag adds the --token-ttl flag to the given flagset
func (bto *BootstrapTokenOptions) AddTTLFlag(fs *pflag.FlagSet) {
	bto.AddTTLFlagWithName(fs, TokenTTL)
}

  }
}
versions {
  producer: 29
  min_consumer: 12
}

# Find PartitionedCall ops in main and match the callee name.
# CHECK: func @main
# CHECK: "tf.PartitionedCall"
# CHECK-SAME: f = @[[FUNCTION_FOO:[a-zA-Z0-9_]*]]

# Find callee and verify it has the stateful attribute set.
# CHECK: func private @[[FUNCTION_FOO]]
# CHECK-SAME: attributes

// CHECK: [[CALL_RESULT_REG:%.*]], [[CALL_RESULT_REG_control:%.*]] = tf_executor.island wraps "tf.StatefulPartitionedCall"([[IDENTITY_REG]])
// CHECK-SAME: f = @[[FUNCTION:[a-zA-Z0-9_]*]]

// Match the inserted Identity op for call output.
// CHECK: "tf.Identity"([[CALL_RESULT_REG]])

// Match the function name

		0x2b0198237e3db5d5,
	}
	// Use the exponent to extract the 3 appropriate uint64 digits from mPi,
	// B ~ (z0, z1, z2), such that the product leading digit has the exponent -64.
	// Note, exp >= 50 since x >= reduceThreshold and exp < 971 for maximum float64.
	digit, bitshift := uint(exp+64)/64, uint(exp+64)%64
	z0 := (mPi[digit] << bitshift) | (mPi[digit+1] >> (64 - bitshift))

# shellcheck disable=SC1001
LATEST_DISTROLESS_SHA256=$(crane digest cgr.dev/chainguard/static | awk -F\: '{print $2}')

        <module name="MemberName"/>
        <module name="MethodName"/>
        <module name="MethodTypeParameterName"/>
        <module name="PackageName">
            <property name="format" value="^[a-z]+(\.[a-z][a-z0-9]*)*$"/>
        </module>
        <module name="ParameterName"/>
        <module name="StaticVariableName"/>
        <module name="TypeName"/>
    </module>

        <module name="MemberName"/>
        <module name="MethodName"/>
        <module name="MethodTypeParameterName"/>
        <module name="PackageName">
            <property name="format" value="^[a-z]+(\.[a-z][a-z0-9]*)*$"/>
        </module>
        <module name="ParameterName"/>
        <module name="StaticVariableName"/>
        <module name="TypeName"/>
    </module>

  func.return
}

// CHECK-NOT: tf_device.launch


// Test empty launch with some results is folded away.
// CHECK-LABEL: func @empty_launch
// CHECK-SAME: (%[[ARG_0:[a-z0-9]*]]: tensor<i1>, %[[ARG_1:[a-z0-9]*]]: tensor<i32>)
func.func @empty_launch(%arg0 : tensor<i1>, %arg1 : tensor<i32>) -> (tensor<i32>, tensor<i1>) {
  %result:2 = "tf_device.launch"() ({
    tf_device.return %arg0, %arg1 : tensor<i1>, tensor<i32>

     */
    default String getId() {
        return convertToId(getDisplayName());
    }

    static String convertToId(String displayName) {
        return displayName.replaceAll("[^a-zA-Z0-9]", "-");
    }

    /**
     * A human consumable display name for this performance test.
     */
    default String getDisplayName() {
        return getExperiment().getDisplayName();
    }