input: "BoxPredictor_4/ClassPredictor/act_quant/FakeQuantWithMinMaxVars"
  input: "BoxPredictor_4/Reshape_1/shape"
  attr {
    key: "T"
    value {
      type: DT_FLOAT
    }
  }
  attr {
    key: "Tshape"
    value {
      type: DT_INT32
    }
  }
}

# MLIR-LABEL: func @main

  %r1 = "tf.Reshape"(%in, %s1) {T = f32, Tshape = i32, device = ""} : (tensor<3x3x48x1xf32>, tensor<3xi32>) -> tensor<3x3x48xf32>
  %fq = "tf.FakeQuantWithMinMaxVarsPerChannel"(%r1, %mini, %maxi) {num_bits = 3, narrow_range = false} : (tensor<3x3x48xf32>, tensor<48xf32>, tensor<48xf32>) -> tensor<3x3x48xf32>
  %r2 = "tf.Reshape"(%fq, %s2) {T = f32, Tshape = i32, device = ""} : (tensor<3x3x48xf32>, tensor<4xi32>) -> tensor<3x3x48x1xf32>

  input: "BoxPredictor_4/ClassPredictor/act_quant/FakeQuantWithMinMaxVars"
  input: "BoxPredictor_4/Reshape_1/shape"
  attr {
    key: "T"
    value {
      type: DT_FLOAT
    }
  }
  attr {
    key: "Tshape"
    value {
      type: DT_INT32
    }
  }
}

# MLIR-LABEL: func @main

  input: "BoxPredictor_4/ClassPredictor/act_quant/FakeQuantWithMinMaxVars"
  input: "BoxPredictor_4/Reshape_1/shape"
  attr {
    key: "T"
    value {
      type: DT_FLOAT
    }
  }
  attr {
    key: "Tshape"
    value {
      type: DT_INT32
    }
  }
}

# MLIR-LABEL: func @main

  );

  let results = (outs
    TF_FloatTensor:$output
  );

  TF_DerivedOperandTypeAttr T = TF_DerivedOperandTypeAttr<1>;
  TF_DerivedOperandTypeAttr Tshape = TF_DerivedOperandTypeAttr<0>;
}

def TF_CosOp : TF_Op<"Cos", [Pure, TF_SameOperandsAndResultTypeResolveRef]> {
  let summary = "Computes cos of x element-wise.";

  let description = [{

    %19 = "tf.Pack"(%18, %cst_5) {N = 2 : i64, T = i32, axis = 0 : i64, device = ""} : (tensor<i32>, tensor<i32>) -> tensor<2xi32>
    // CHECK: -> tensor<1x512xf32>
    %40 = "tf.Reshape"(%39, %19) {T = f32, Tshape = i32, device = ""} : (tensor<1x4x4x32xf32>, tensor<2xi32>) -> tensor<?x?xf32>
   func.return
  }

  // CHECK-LABEL: func @shape_partial_eval