cmplw := ir.Node(ir.NewIndexExpr(base.Pos, cmpl, ir.NewInt(base.Pos, i)))
				cmplw = typecheck.Conv(cmplw, elemType) // convert to unsigned
				cmplw = typecheck.Conv(cmplw, convType) // widen
				cmprw := ir.Node(ir.NewIndexExpr(base.Pos, cmpr, ir.NewInt(base.Pos, i)))
				cmprw = typecheck.Conv(cmprw, elemType)
				cmprw = typecheck.Conv(cmprw, convType)
				// For code like this:  uint32(s[0]) | uint32(s[1])<<8 | uint32(s[2])<<16 ...

		case fmtTypeID, fmtTypeIDName:
			if verb == 'S' {
				tconv2(b, t.Elem(), 'S', mode, visited)
				return
			}
		}
		tconv2(b, t.Elem(), 'v', mode, visited)

	case TARRAY:
		b.WriteByte('[')
		b.WriteString(strconv.FormatInt(t.NumElem(), 10))
		b.WriteByte(']')
		tconv2(b, t.Elem(), 0, mode, visited)

	case TSLICE:
		b.WriteString("[]")
		tconv2(b, t.Elem(), 0, mode, visited)

	case TCHAN:

  // CHECK: %[[CONV:.+]] = mhlo.convolution(%[[INPUT]], %[[NEW_FILTER]]) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {stride = [1, 1], pad = {{\[\[}}0, 0], [0, 0]], rhs_dilate = [1, 1]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<?x256x256x3xf32>, tensor<1x1x3x2xf32>) -> tensor<?x256x256x2xf32>
  // CHECK: %[[SHAPE:.+]] = shape.shape_of %[[CONV]] : tensor<?x256x256x2xf32> -> tensor<4xindex>

// CHECK-DAG: %[[CONV:.*]] = stablehlo.convolution(%[[ARG]], %[[CONST_1]]) {{.*}} : (tensor<1x3x4x3xf32>, tensor<2x3x3x2xf32>) -> tensor<1x3x2x2xf32>
// CHECK-DAG: %[[BROADCAST:.*]] = stablehlo.broadcast_in_dim %[[CONST_0]], dims = [3] : (tensor<2xf32>) -> tensor<1x3x2x2xf32>
// CHECK-DAG: %[[ADD:.*]] = stablehlo.add %[[CONV]], %[[BROADCAST]] : tensor<1x3x2x2xf32>

        given:
        repositories {
            maven("""content {
                onlyForConfigurations("conf2")
            }""")
        }
        buildFile << """
            configurations {
                conf2
            }
            dependencies {
                conf "org:foo:1.0"
                conf2 "org:foo:1.0"
            }
            tasks.register("verify") {
                doFirst {

  [(HasRankOf<1> $add_rhs_value),
   (HasEqualElementSize<[-1], [0]> $conv_out, $add_rhs)], [], (addBenefit -1)>;

// Convert conv+sub+mul pattern to conv+mul+add.
// (conv - sub) * mul -> conv * mul + (-sub) * mul
//
// This is needed to support Conv+BatchNorm pattern from Jax models converted
// using jax2tf w/o native serialization. Note that Jax2tf patterns always

  using OpRewritePattern<TFL::SplitVOp>::OpRewritePattern;

  LogicalResult matchAndRewrite(TFL::SplitVOp splitv_op,
                                PatternRewriter& rewriter) const override;
};

// Ensure bias for conv2d op.
struct EnsureBiasForConv2d : public OpRewritePattern<TFL::Conv2DOp> {
  using OpRewritePattern<TFL::Conv2DOp>::OpRewritePattern;

  LogicalResult matchAndRewrite(TFL::Conv2DOp conv_op,

// CHECK: %[[DEQUANTIZE:.*]] = "quantfork.dcast"(%[[QUANTIZE]])
// CHECK: %[[CONV:.*]] = "tf.Conv2D"(%arg0, %[[DEQUANTIZE]])
// CHECK: return %[[CONV]]
}

// CHECK-LABEL: perChannelFakeQuantWithConv2D
func.func @perChannelFakeQuantWithConv2D(tensor<256x32x32x3xf32>) -> (tensor<256x8x7x16xf32>) {
^bb0(%arg: tensor<256x32x32x3xf32>) :

    }
  }
  attr {
    key: "_class"
    value {
      list {
        s: "loc:@MobilenetV1/Conv2d_0/weights"
      }
    }
  }
}
node {
  name: "MobilenetV1/MobilenetV1/Conv2d_0/Conv2D"
  op: "Conv2D"
  input: "input"
  input: "MobilenetV1/Conv2d_0/weights/read"
  attr {
    key: "T"
    value {
      type: DT_FLOAT
    }
  }
  attr {
    key: "padding"
    value {

  <wb-module deploy-name="name">
    <property name="context-root" value="name"/>
    <wb-resource deploy-path="WEB-INF/lib" source-path="conf1"/>
    <wb-resource deploy-path="WEB-INF/lib" source-path="conf2/child"/>
    <property name="java-output-path" value="bin"/>
    <dependent-module deploy-path="/WEB-INF/lib" handle="module:/classpath/lib//lib/a.jar">
      <dependency-type>uses</dependency-type>
    </dependent-module>