diff --git a/README.md b/README.md
index 9597f187726c4d00f8aa8af72b68bdc248a11e64..8675efe4c3bff20414e09aa5f10ec01b3883184e 100644
--- a/README.md
+++ b/README.md
@@ -28,11 +28,11 @@ Link the datasets in the following folders with symlink.
 The pretrained models are available on [onedrive](https://mailstsinghuaeducn-my.sharepoint.com/:f:/g/personal/liuxy21_mails_tsinghua_edu_cn/EgT-saAOPtROonp9o_nmPDQBeIO2x7RjvhMSXhMMTMgrlg?e=L2bX2f) (password: groupji).
 
 ## Training
-We train a model separately for each of the BOP datasets. To train the TUDL refinement model, run
+We train a model separately for each of the BOP datasets. To train the TUD-L refinement model, run
 ```
-python train.py --dataset ycbv --batch_size 12 --num_inner_loops 10 --num_solver_steps 3 [--pbr_only]
+python train.py --dataset tudl --batch_size 12 --num_inner_loops 10 --num_solver_steps 3 [--pbr_only]
 ```
-To train a refinement model on other BOP datasets, replace `ycbv` with one of the following: `tless`, `lmo`, `hb`, `tudl`, `icbin`, `itodd`
+To train a refinement model on other BOP datasets, replace `tudl` with one of the following: `tless`, `lmo`, `hb`, `ycbv`, `icbin`, `itodd`
 
 
 ## Testing
diff --git a/test.sh b/test.sh
deleted file mode 100644
index 3c896810806fef643e6303590233b17bc492c9c9..0000000000000000000000000000000000000000
--- a/test.sh
+++ /dev/null
@@ -1,13 +0,0 @@
-#!/usr/bin/env bash
-
-DATASET=$1
-export CUDA_VISIBLE_DEVICES=$3
-
-suffix=""
-if [ "$#" -ge 4 ] # ok
-then
-    suffix=$4
-fi
-
-python test_from_init_pose_pickle.py --save_dir output/my_evaluation_${DATASET}${suffix} --dataset $DATASET --load_weights model_weights/refiner/${DATASET}_rgbd.pth \
---init_pose_file $2 --num_inner_loops 20 --num_solver_steps 5
\ No newline at end of file