diff --git a/.gitignore b/.gitignore
index eedc441..c5d7b1c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -74,3 +74,5 @@ dataset/scannetv2/scannetv2-labels.combined.tsv
 
 dataset/s3dis/preprocess
 dataset/s3dis/val_gt
+dataset/s3dis/preprocess_sample
+dataset/s3dis/Stanford3dDataset_v1.2
diff --git a/data/s3dis_inst.py b/data/s3dis_inst.py
index 0e4410f..2c8b177 100644
--- a/data/s3dis_inst.py
+++ b/data/s3dis_inst.py
@@ -39,14 +39,6 @@ class Dataset:
 
 
     def trainLoader(self):
-        # if self.train_split == 'trainval':
-        #     train_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'train', '*' + self.filename_suffix))
-        #         + glob.glob(os.path.join(self.data_root, self.dataset, 'val', '*' + self.filename_suffix))
-        #     )
-        # elif self.train_split == 'train':
-        #     train_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'train', '*' + self.filename_suffix)))
-        # else:
-        #     raise Exception
         train_file_names = []
         for area in self.train_areas:
             train_file_names += glob.glob(os.path.join(self.data_root, self.dataset, 'preprocess', area + '*' + self.filename_suffix))
@@ -83,7 +75,7 @@ class Dataset:
                                     drop_last=False, pin_memory=True)
 
     def valLoader(self):
-        val_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'preprocess/_sample', self.test_area + '*' + self.filename_suffix)))
+        val_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'preprocess_sample', self.test_area + '*' + self.filename_suffix)))
         self.val_files = val_file_names
 
         logger.info('Validation samples: {}'.format(len(self.val_files)))
@@ -94,7 +86,7 @@ class Dataset:
 
 
     def testLoader(self):
-        self.test_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'preprocess/', self.test_area + '*' + self.filename_suffix)))
+        self.test_file_names = sorted(glob.glob(os.path.join(self.data_root, self.dataset, 'preprocess', self.test_area + '*' + self.filename_suffix)))
         self.test_files = self.test_file_names
 
         logger.info('Testing samples ({}): {}'.format(self.test_split, len(self.test_files)))
diff --git a/dataset/s3dis/downsample.py b/dataset/s3dis/downsample.py
index 6c7a164..90e06f3 100644
--- a/dataset/s3dis/downsample.py
+++ b/dataset/s3dis/downsample.py
@@ -75,7 +75,7 @@ def get_parser():
         description="downsample s3dis by voxelization")
     parser.add_argument("--data-dir",
                         type=str,
-                        default="./inputs",
+                        default="./preprocess",
                         help="directory save processed data")
     parser.add_argument("--ratio",
                         type=float,
diff --git a/dataset/s3dis/prepare_data.sh b/dataset/s3dis/prepare_data.sh
new file mode 100644
index 0000000..91d56fd
--- /dev/null
+++ b/dataset/s3dis/prepare_data.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+python prepare_data_inst.py
+python downsample.py
+python prepare_data_inst_gttxt.py
diff --git a/dataset/s3dis/prepare_data_inst.py b/dataset/s3dis/prepare_data_inst.py
index 3893a4f..e284e17 100644
--- a/dataset/s3dis/prepare_data_inst.py
+++ b/dataset/s3dis/prepare_data_inst.py
@@ -105,11 +105,11 @@ def get_parser():
     parser = argparse.ArgumentParser(description="s3dis data prepare")
     parser.add_argument("--data-root",
                         type=str,
-                        default="./data",
+                        default="./Stanford3dDataset_v1.2",
                         help="root dir save data")
     parser.add_argument("--save-dir",
                         type=str,
-                        default="./inputs",
+                        default="./preprocess",
                         help="directory save processed data")
     parser.add_argument(
         "--patch",
diff --git a/requirements.txt b/requirements.txt
index ee1df80..08d9f8b 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -5,3 +5,5 @@ tensorboardX
 pyyaml==5.4.1
 scipy
 six
+pandas
+scikit-learn
diff --git a/test_s3dis.py b/test_s3dis.py
index da37a63..7ded63f 100644
--- a/test_s3dis.py
+++ b/test_s3dis.py
@@ -172,12 +172,12 @@ def test(model, model_fn, data_name, epoch):
                 # import pdb; pdb.set_trace()
 
                 nclusters = clusters.shape[0]
-                if nclusters > cfg.max_clusters:
-                    nclusters = cfg.max_clusters
-                    _, topk_inds = cluster_scores.topk(cfg.max_clusters)
-                    clusters = clusters[topk_inds]
-                    cluster_scores = cluster_scores[topk_inds]
-                    cluster_semantic_id = cluster_semantic_id[topk_inds]
+                # if nclusters > cfg.max_clusters:
+                #     nclusters = cfg.max_clusters
+                #     _, topk_inds = cluster_scores.topk(cfg.max_clusters)
+                #     clusters = clusters[topk_inds]
+                #     cluster_scores = cluster_scores[topk_inds]
+                #     cluster_semantic_id = cluster_semantic_id[topk_inds]
 
 
                 # prepare for evaluation