fix

Author: Liujie0926

.github/workflows/unittest-gpu.yml

@@ -135,6 +135,7 @@ jobs:
             docker exec -t $container_name /bin/bash -c '
             cd /workspace/PaddleNLP
             rm -rf tests/utils/test_aistudio_download.py
+            '
           fi
       
       - name: Test

tests/llm/test_grpo.py

@@ -92,7 +92,7 @@ def test_grpo(self):
             # 运行主逻辑
             cmd = 'python -u -m paddle.distributed.launch \
                     --devices "$CUDA_VISIBLE_DEVICES" run_rl.py \
-                    ../../config/qwen/reinforce_plus_plus_argument.yaml \
+                    ../../config/qwen/grpo_argument.yaml \
                     --actor_model_name_or_path "Qwen/Qwen2-1.5B" \
                     --max_dec_len 128 \
                     --max_steps 3 \

tests/llm/test_reinforce_plus_plus.py

@@ -92,7 +92,7 @@ def test_reinforce_plus_plus(self):
             # 运行主逻辑
             cmd = 'python -u -m paddle.distributed.launch \
                     --devices "$CUDA_VISIBLE_DEVICES" run_rl.py \
-                    ../../config/qwen/reinforce_plus_plus_argument.yaml \
+                    ../../config/qwen/grpo_argument.yaml \
                     --rl_algorithm "reinforce_plus_plus" \
                     --actor_model_name_or_path "Qwen/Qwen2-1.5B" \
                     --max_dec_len 128 \