diff --git a/README.md b/README.md
index ec62a5a6..94458c80 100644
--- a/README.md
+++ b/README.md
@@ -198,6 +198,8 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --do_train \
     --dataset comparison_gpt4_en \
     --finetuning_type lora \
+    --resume_lora_training False \
+    --checkpoint_dir path_to_sft_checkpoint \
     --output_dir path_to_rm_checkpoint \
     --per_device_train_batch_size 4 \
     --gradient_accumulation_steps 4 \
@@ -219,6 +221,7 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --do_train \
     --dataset alpaca_gpt4_en \
     --finetuning_type lora \
+    --resume_lora_training False \
     --checkpoint_dir path_to_sft_checkpoint \
     --reward_model path_to_rm_checkpoint \
     --output_dir path_to_ppo_checkpoint \
@@ -229,7 +232,6 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --save_steps 1000 \
     --learning_rate 1e-5 \
     --num_train_epochs 1.0 \
-    --resume_lora_training False \
     --plot_loss
 ```
 
diff --git a/README_zh.md b/README_zh.md
index abf674bc..4d882b62 100644
--- a/README_zh.md
+++ b/README_zh.md
@@ -198,6 +198,8 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --do_train \
     --dataset comparison_gpt4_en \
     --finetuning_type lora \
+    --resume_lora_training False \
+    --checkpoint_dir path_to_sft_checkpoint \
     --output_dir path_to_rm_checkpoint \
     --per_device_train_batch_size 4 \
     --gradient_accumulation_steps 4 \
@@ -219,6 +221,7 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --do_train \
     --dataset alpaca_gpt4_en \
     --finetuning_type lora \
+    --resume_lora_training False \
     --checkpoint_dir path_to_sft_checkpoint \
     --reward_model path_to_rm_checkpoint \
     --output_dir path_to_ppo_checkpoint \
@@ -229,7 +232,6 @@ CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
     --save_steps 1000 \
     --learning_rate 1e-5 \
     --num_train_epochs 1.0 \
-    --resume_lora_training False \
     --plot_loss
 ```