diff --git a/demo/slimfacenet/slim_eval.sh b/demo/slimfacenet/slim_eval.sh
index b26fa3637a867e919c19b2b957f6a107912f1cd4..922adaf6f403bef6dbbdea5cfe18790819aefac7 100644
--- a/demo/slimfacenet/slim_eval.sh
+++ b/demo/slimfacenet/slim_eval.sh
@@ -15,5 +15,5 @@
 #!/bin/bash
 export CUDA_VISIBLE_DEVICES=0
 python train_eval.py --action test \
-    --train_data_dir=/PATH_TO_CASIA_Dataset \
-    --test_data_dir=/PATH_TO_lfw \
+    --train_data_dir=./CASIA/ \
+    --test_data_dir=./lfw/ \
diff --git a/demo/slimfacenet/slim_quant.sh b/demo/slimfacenet/slim_quant.sh
index 8df4b12606637be4c4fd8f5ee186ecafefe9d577..754565715baac92582779a009b0fe965757d3930 100644
--- a/demo/slimfacenet/slim_quant.sh
+++ b/demo/slimfacenet/slim_quant.sh
@@ -15,5 +15,6 @@
 #!/bin/bash
 export CUDA_VISIBLE_DEVICES=0
 python train_eval.py --action quant \
-    --train_data_dir=/PATH_TO_CASIA_Dataset \
-    --test_data_dir=/PATH_TO_lfw \
+    --train_data_dir=./CASIA/ \
+    --test_data_dir=./lfw/ \
+    --seed=1
diff --git a/demo/slimfacenet/slim_train.sh b/demo/slimfacenet/slim_train.sh
index 55ad2250d65a0fae8a88cb8016e86de8257f5f59..62cbd192fd5a1fff994a564a3a28f56867455ad4 100644
--- a/demo/slimfacenet/slim_train.sh
+++ b/demo/slimfacenet/slim_train.sh
@@ -15,7 +15,7 @@
 #!/bin/bash
 export CUDA_VISIBLE_DEVICES=0
 python -u train_eval.py \
-    --train_data_dir=/PATH_TO_CASIA_Dataset \
-    --test_data_dir=/PATH_TO_LFW \
+    --train_data_dir=./CASIA/ \
+    --test_data_dir=./lfw/ \
     --action train \
     --model=SlimFaceNet_B_x0_75
diff --git a/demo/slimfacenet/train_eval.py b/demo/slimfacenet/train_eval.py
index 125dba368a19be5b7c8bb9f79533b3ab1a44b2ea..421496f0463ce99a48b19f483f970f44ceb022dd 100644
--- a/demo/slimfacenet/train_eval.py
+++ b/demo/slimfacenet/train_eval.py
@@ -120,6 +120,7 @@ def train(exe, train_program, train_out, test_program, test_out, args):
     compiled_prog = compiler.CompiledProgram(
         train_program, build_strategy=build_strategy).with_data_parallel(
             loss_name=loss.name, build_strategy=build_strategy)
+    best_ave = 0
     for epoch_id in range(args.start_epoch, args.total_epoch):
         for batch_id, data in enumerate(train_reader()):
             loss, acc, global_lr = exe.run(compiled_prog,
@@ -135,14 +136,17 @@ def train(exe, train_program, train_out, test_program, test_out, args):
             model_path = os.path.join(args.save_ckpt, str(epoch_id))
             fluid.io.save_persistables(
                 executor=exe, dirname=model_path, main_program=train_program)
-            test(exe, test_program, test_out, args)
-    out_feature, test_reader, flods, flags = test_out
-    fluid.io.save_inference_model(
-        executor=exe,
-        dirname='./out_inference',
-        feeded_var_names=['image_test'],
-        target_vars=[out_feature],
-        main_program=test_program)
+            temp_ave = test(exe, test_program, test_out, args)
+            if temp_ave > best_ave:
+                best_ave = temp_ave
+                print('Best AVE: {}'.format(best_ave))
+                out_feature, test_reader, flods, flags = test_out
+                fluid.io.save_inference_model(
+                    executor=exe,
+                    dirname='./out_inference',
+                    feeded_var_names=['image_test'],
+                    target_vars=[out_feature],
+                    main_program=test_program)
 
 
 def build_program(program, startup, args, is_train=True):
@@ -229,7 +233,9 @@ def quant_val_reader_batch():
     test_dataset = LFW(nl, nr)
     test_reader = paddle.batch(
         test_dataset.reader, batch_size=1, drop_last=False)
-    shuffle_reader = fluid.io.shuffle(test_reader, 1)
+    shuffle_index = args.seed if args.seed else np.random.randint(1000)
+    print('shuffle_index: {}'.format(shuffle_index))
+    shuffle_reader = fluid.io.shuffle(test_reader, shuffle_index)
 
     def _reader():
         while True:
@@ -283,6 +289,7 @@ def main():
         '--start_epoch', default=0, type=int, help='start_epoch')
     parser.add_argument(
         '--total_epoch', default=80, type=int, help='total_epoch')
+    parser.add_argument('--seed', default=None, type=int, help='shuffle seed')
     parser.add_argument(
         '--save_frequency', default=1, type=int, help='save_frequency')
     parser.add_argument(
diff --git a/paddleslim/models/slimfacenet.py b/paddleslim/models/slimfacenet.py
index 6a1c0a7dcdb176a779b66658cdfe9c3be1eba80d..95b445fcd65b6fdfd0a7a4e991f3a18496b1807a 100644
--- a/paddleslim/models/slimfacenet.py
+++ b/paddleslim/models/slimfacenet.py
@@ -362,7 +362,7 @@ def SlimFaceNet_B_x0_75(class_dim=None, scale=0.6, arch=None):
 
 def SlimFaceNet_C_x0_75(class_dim=None, scale=0.6, arch=None):
     scale = 0.75
-    arch = [1, 1, 2, 1, 0, 2, 1, 0, 1, 0, 1, 1, 2, 2, 3]
+    arch = [1, 3, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1, 5, 5, 5]
     return SlimFaceNet(class_dim=class_dim, scale=scale, arch=arch)