[PIR] add an api to get param map when translating

[kangguangli]

PR types

Others

PR changes

Others

Description

For program below,

# tests.py
import paddle
from paddle import pir
paddle.enable_static()

x = paddle.randn([4, 4])
main_program, start_program = (
    paddle.static.Program(),
    paddle.static.Program(),
)
with paddle.static.program_guard(main_program, start_program):
    x_s = paddle.static.data('x', [4, 4], x.dtype)
    x_s.stop_gradient = False
    y_s = paddle.matmul(x_s, x_s)
    z_s = paddle.add(y_s, y_s)
    k_s = paddle.tanh(z_s)
newir_program, mappings = pir.translate_to_new_ir_with_param_map(main_program.desc)

print(newir_program)
print(mappings)

Run FLAGS_enable_new_ir_in_executor=1 python test.py, we get

{
 (%0) = "pd_op.data" () {dtype:(pd_op.DataType)float32,is_persisable:[false],name:"x",place:(pd_op.Place)Place(undefined:0),shape:(pd_op.IntArray)[4,4],stop_gradient:[false]} : () -> pd_op.tensor<4x4xf32>
 (%1) = "pd_op.matmul" (%0, %0) {is_persisable:[false],stop_gradient:[false],transpose_x:false,transpose_y:false} : (pd_op.tensor<4x4xf32>, pd_op.tensor<4x4xf32>) -> pd_op.tensor<4x4xf32>
 (%2) = "pd_op.add" (%1, %1) {is_persisable:[false],stop_gradient:[false]} : (pd_op.tensor<4x4xf32>, pd_op.tensor<4x4xf32>) -> pd_op.tensor<4x4xf32>
 (%3) = "pd_op.tanh" (%2) {is_persisable:[false],stop_gradient:[false]} : (pd_op.tensor<4x4xf32>) -> pd_op.tensor<4x4xf32>
}

{'x': [Value(define_op_name=pd_op.data, index=0, dtype=pd_op.tensor<4x4xf32>)], 'matmul_v2_0.tmp_0': [Value(define_op_name=pd_op.matmul, index=0, dtype=pd_op.tensor<4x4xf32>)], 'elementwise_add_0': [Value(define_op_name=pd_op.add, index=0, dtype=pd_op.tensor<4x4xf32>)], 'tanh_0.tmp_0': [Value(define_op_name=pd_op.tanh, index=0, dtype=pd_op.tensor<4x4xf32>)]}

Others

Pcard-67164

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.