diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index e58400d0f..4fc39b23e 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -8,11 +8,6 @@ repos:
         files: \.(md|yml)$
     -   id: remove-tabs
         files: \.(md|yml)$
--   repo: https://github.com/PaddlePaddle/mirrors-yapf.git
-    rev: v0.16.2
-    hooks:
-    -   id: yapf
-        files: (.*\.(py|bzl)|BUILD|.*\.BUILD|WORKSPACE)$
 -   repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.4.0
     hooks:
diff --git a/ce_tests/dygraph/quant/src/eval.py b/ce_tests/dygraph/quant/src/eval.py
index b6afdc1dd..e8880ba14 100644
--- a/ce_tests/dygraph/quant/src/eval.py
+++ b/ce_tests/dygraph/quant/src/eval.py
@@ -17,9 +17,13 @@
 
 
 def eval(args):
-    model_file = os.path.join(args.model_path, args.model_filename)
-    params_file = os.path.join(args.model_path, args.params_filename)
-    config = paddle_infer.Config(model_file, params_file)
+    if '2' in paddle.__version__.split('.')[0]:
+        config = paddle.inference.Config(
+            os.path.join(args.model_path, args.model_filename),
+            os.path.join(args.model_path, args.params_filename))
+    else:
+        model_prefix = args.model_filename.split('.')[0]
+        config = paddle.inference.Config(os.path.join(args.model_path, model_prefix))
     config.enable_mkldnn()
     config.switch_ir_optim(False)
 
diff --git a/ce_tests/dygraph/quant/src/test.py b/ce_tests/dygraph/quant/src/test.py
index 1dbd85d75..6ad395ed4 100644
--- a/ce_tests/dygraph/quant/src/test.py
+++ b/ce_tests/dygraph/quant/src/test.py
@@ -18,9 +18,13 @@
 
 def eval(args):
     # create predictor
-    model_file = os.path.join(args.model_path, args.model_filename)
-    params_file = os.path.join(args.model_path, args.params_filename)
-    config = paddle_infer.Config(model_file, params_file)
+    if '2' in paddle.__version__.split('.')[0]:
+        config = paddle.inference.Config(
+            os.path.join(args.model_path, args.model_filename),
+            os.path.join(args.model_path, args.params_filename))
+    else:
+        model_prefix = args.model_filename.split('.')[0]
+        config = paddle.inference.Config(os.path.join(args.model_path, model_prefix))
     if args.use_gpu:
         config.enable_use_gpu(1000, 0)
     if not args.ir_optim:
diff --git a/demo/dygraph/post_quant/eval.py b/demo/dygraph/post_quant/eval.py
index f9af9e39f..e9dd8341a 100644
--- a/demo/dygraph/post_quant/eval.py
+++ b/demo/dygraph/post_quant/eval.py
@@ -30,9 +30,14 @@
 
 def eval():
     # create predictor
-    model_file = os.path.join(FLAGS.model_path, FLAGS.model_filename)
-    params_file = os.path.join(FLAGS.model_path, FLAGS.params_filename)
-    config = paddle_infer.Config(model_file, params_file)
+    if '2' in paddle.__version__.split('.')[0]:
+        config = paddle.inference.Config(
+            os.path.join(FLAGS.model_path, FLAGS.model_filename),
+            os.path.join(FLAGS.model_path, FLAGS.params_filename))
+    else:
+        model_prefix = FLAGS.model_filename.split('.')[0]
+        config = paddle.inference.Config(os.path.join(FLAGS.model_path, model_prefix))
+
     if FLAGS.use_gpu:
         config.enable_use_gpu(1000, 0)
     if not FLAGS.ir_optim:
diff --git a/example/auto_compression/detection/paddle_inference_eval.py b/example/auto_compression/detection/paddle_inference_eval.py
index 838d37e13..6d7cd26f7 100644
--- a/example/auto_compression/detection/paddle_inference_eval.py
+++ b/example/auto_compression/detection/paddle_inference_eval.py
@@ -243,9 +243,13 @@ def load_predictor(
         raise ValueError(
             "Predict by TensorRT mode: {}, expect device=='GPU', but device == {}".
             format(precision, device))
-    config = Config(
+    # support paddle 2.x
+    if '2' in paddle.__version__.split('.')[0]:
+        config = Config(
         os.path.join(model_dir, "model.pdmodel"),
         os.path.join(model_dir, "model.pdiparams"))
+    else:
+        config = Config(os.path.join(model_dir, "model"))
 
     config.enable_memory_optim()
     if device == "GPU":
diff --git a/example/auto_compression/image_classification/paddle_inference_eval.py b/example/auto_compression/image_classification/paddle_inference_eval.py
index d36073875..4e6214795 100644
--- a/example/auto_compression/image_classification/paddle_inference_eval.py
+++ b/example/auto_compression/image_classification/paddle_inference_eval.py
@@ -113,9 +113,14 @@ def __init__(self):
 
     def _create_paddle_predictor(self):
         inference_model_dir = args.model_path
-        model_file = os.path.join(inference_model_dir, args.model_filename)
-        params_file = os.path.join(inference_model_dir, args.params_filename)
-        config = paddle.inference.Config(model_file, params_file)
+        if '2' in paddle.__version__.split('.')[0]:
+            config = paddle.inference.Config(
+            os.path.join(inference_model_dir, args.model_filename),
+            os.path.join(inference_model_dir, args.params_filename))
+        else:
+            model_prefix = args.model_filename.split('.')[0]
+            config = paddle.inference.Config(os.path.join(inference_model_dir, model_prefix))
+        
         precision = paddle.inference.Config.Precision.Float32
         if args.use_int8:
             precision = paddle.inference.Config.Precision.Int8
diff --git a/example/auto_compression/nlp/paddle_inference_eval.py b/example/auto_compression/nlp/paddle_inference_eval.py
index 073f032e5..0a6584e4e 100644
--- a/example/auto_compression/nlp/paddle_inference_eval.py
+++ b/example/auto_compression/nlp/paddle_inference_eval.py
@@ -207,9 +207,13 @@ def create_predictor(cls, args):
         create_predictor func
         """
         cls.rerun_flag = False
-        config = paddle.inference.Config(
-            os.path.join(args.model_path, args.model_filename),
-            os.path.join(args.model_path, args.params_filename))
+        if '2' in paddle.__version__.split('.')[0]:
+            config = paddle.inference.Config(
+                os.path.join(args.model_path, args.model_filename),
+                os.path.join(args.model_path, args.params_filename))
+        else:
+            model_prefix=args.model_filename.split('.')[0]
+            config = paddle.inference.Config(os.path.join(args.model_path, model_prefix))
         config.switch_ir_debug(True)
         # 适用于ERNIE 3.0-Medium模型
         # config.exp_disable_tensorrt_ops(["elementwise_add"])
diff --git a/example/auto_compression/nlp/paddle_inference_eval_uie.py b/example/auto_compression/nlp/paddle_inference_eval_uie.py
index 2f378ef02..2a3319170 100644
--- a/example/auto_compression/nlp/paddle_inference_eval_uie.py
+++ b/example/auto_compression/nlp/paddle_inference_eval_uie.py
@@ -162,10 +162,14 @@ def create_predictor(cls, args):
         """
         create_predictor func
         """
-        cls.rerun_flag = False
-        config = paddle.inference.Config(
-            os.path.join(args.model_path, args.model_filename),
-            os.path.join(args.model_path, args.params_filename))
+        cls.rerun_flag = False    
+        if '2' in paddle.__version__.split('.')[0]:
+            config = paddle.inference.Config(
+                os.path.join(args.model_path, args.model_filename),
+                os.path.join(args.model_path, args.params_filename))
+        else:
+            model_prefix=args.model_filename.split('.')[0]
+            config = paddle.inference.Config(os.path.join(args.model_path, model_prefix))
         if args.device == "gpu":
             # set GPU configs accordingly
             config.enable_use_gpu(100, 0)
diff --git a/example/auto_compression/pytorch_huggingface/paddle_inference_eval.py b/example/auto_compression/pytorch_huggingface/paddle_inference_eval.py
index 338825a66..6ce2b1453 100644
--- a/example/auto_compression/pytorch_huggingface/paddle_inference_eval.py
+++ b/example/auto_compression/pytorch_huggingface/paddle_inference_eval.py
@@ -188,9 +188,14 @@ def create_predictor(cls, args):
         create_predictor func
         """
         cls.rerun_flag = False
-        config = paddle.inference.Config(
-            os.path.join(args.model_path, args.model_filename),
-            os.path.join(args.model_path, args.params_filename))
+
+        if '2' in paddle.__version__.split('.')[0]:
+            config = paddle.inference.Config(
+                os.path.join(args.model_path, args.model_filename),
+                os.path.join(args.model_path, args.params_filename))
+        else:
+            model_prefix = args.model_filename.split('.')[0]
+            config = paddle.inference.Config(os.path.join(args.model_path, model_prefix))
 
         if args.device == "gpu":
             # set GPU configs accordingly
diff --git a/example/auto_compression/pytorch_yolo_series/paddle_inference_eval.py b/example/auto_compression/pytorch_yolo_series/paddle_inference_eval.py
index a1df31b78..77c04d76a 100644
--- a/example/auto_compression/pytorch_yolo_series/paddle_inference_eval.py
+++ b/example/auto_compression/pytorch_yolo_series/paddle_inference_eval.py
@@ -253,9 +253,13 @@ def load_predictor(
         raise ValueError(
             "Predict by TensorRT mode: {}, expect device=='GPU', but device == {}".
             format(precision, device))
-    config = Config(
+    if '2' in paddle.__version__.split('.')[0]:
+        config = Config(
         os.path.join(model_dir, "model.pdmodel"),
         os.path.join(model_dir, "model.pdiparams"))
+    else:
+        config = Config(os.path.join(model_dir, "model"))
+        
     if device == "GPU":
         # initial GPU memory(M), device ID
         config.enable_use_gpu(200, 0)
diff --git a/example/auto_compression/pytorch_yolo_series/paddle_trt_infer.py b/example/auto_compression/pytorch_yolo_series/paddle_trt_infer.py
index 5b471690c..40c155e38 100644
--- a/example/auto_compression/pytorch_yolo_series/paddle_trt_infer.py
+++ b/example/auto_compression/pytorch_yolo_series/paddle_trt_infer.py
@@ -244,9 +244,12 @@ def load_predictor(model_dir,
         raise ValueError(
             "Predict by TensorRT mode: {}, expect device=='GPU', but device == {}"
             .format(run_mode, device))
-    config = Config(
-        os.path.join(model_dir, 'model.pdmodel'),
-        os.path.join(model_dir, 'model.pdiparams'))
+    if '2' in paddle.__version__.split('.')[0]:
+        config = Config(
+        os.path.join(model_dir, "model.pdmodel"),
+        os.path.join(model_dir, "model.pdiparams"))
+    else:
+        config = Config(os.path.join(model_dir, "model"))
     if device == 'GPU':
         # initial GPU memory(M), device ID
         config.enable_use_gpu(200, 0)
diff --git a/example/auto_compression/semantic_segmentation/paddle_inference_eval.py b/example/auto_compression/semantic_segmentation/paddle_inference_eval.py
index f9066389b..b71426e07 100644
--- a/example/auto_compression/semantic_segmentation/paddle_inference_eval.py
+++ b/example/auto_compression/semantic_segmentation/paddle_inference_eval.py
@@ -44,9 +44,13 @@ def load_predictor(args):
     load predictor func
     """
     rerun_flag = False
-    model_file = os.path.join(args.model_path, args.model_filename)
-    params_file = os.path.join(args.model_path, args.params_filename)
-    pred_cfg = PredictConfig(model_file, params_file)
+    if '2' in paddle.__version__.split('.')[0]:
+        pred_cfg = PredictConfig(
+        os.path.join(args.model_path, args.model_filename),
+        os.path.join(args.model_path, args.params_filename))
+    else:
+        model_prefix = args.model_filename.split(".")[0]
+        pred_cfg = PredictConfig(os.path.join(args.model_path, model_prefix))
     pred_cfg.enable_memory_optim()
     pred_cfg.switch_ir_optim(True)
     if args.device == "GPU":
diff --git a/example/quantization/ptq/classification/eval.py b/example/quantization/ptq/classification/eval.py
index ef7dc749e..80c202194 100644
--- a/example/quantization/ptq/classification/eval.py
+++ b/example/quantization/ptq/classification/eval.py
@@ -30,9 +30,13 @@
 
 def eval():
     # create predictor
-    model_file = os.path.join(FLAGS.model_path, FLAGS.model_filename)
-    params_file = os.path.join(FLAGS.model_path, FLAGS.params_filename)
-    config = paddle_infer.Config(model_file, params_file)
+    if '2' in paddle.__version__.split('.')[0]:
+        config = paddle.inference.Config(
+            os.path.join(FLAGS.model_path, FLAGS.model_filename),
+            os.path.join(FLAGS.model_path, FLAGS.params_filename))
+    else:
+        model_prefix = FLAGS.model_filename.split('.')[0]
+        config = paddle.inference.Config(os.path.join(FLAGS.model_path, model_prefix))
     if FLAGS.use_gpu:
         config.enable_use_gpu(1000, 0)
     if not FLAGS.ir_optim: