diff --git a/.gitee/ISSUE_TEMPLATE/6-application-case.yml b/.gitee/ISSUE_TEMPLATE/6-application-case.yml
index dfbf8e154b1fa9ff77b6e154895f911627c8ce97..6ea117ca91cf112235559b7e17e151e9210c1562 100644
--- a/.gitee/ISSUE_TEMPLATE/6-application-case.yml
+++ b/.gitee/ISSUE_TEMPLATE/6-application-case.yml
@@ -1,5 +1,5 @@
-name: 🤗 Support request for a new application case of science
-description: Submit a proposal/request for a new application case of science
+name: 🤗 Support request for a new application case of AIForScience
+description: Submit a proposal/request for a new application case of AIForScience
 title: "[Application Case]: "
 labels: ["application-case"]
 
diff --git a/MindSPONGE/applications/model_cards/MEGAProtein.md b/MindSPONGE/applications/model_cards/MEGAProtein.md
index dd00ed33eb0f287bda059f5fede62a55abdff962..1d9e74e7ca62714f3ffa374548f5ff315eb01bab 100644
--- a/MindSPONGE/applications/model_cards/MEGAProtein.md
+++ b/MindSPONGE/applications/model_cards/MEGAProtein.md
@@ -102,8 +102,8 @@ msa_feature['decoy_aatype'] = np.pad(aatype, (0, 256 - aatype.shape[0]))
 msa_feature['decoy_atom_positions'] = np.pad(final_atom_positions, ((0, 256 - final_atom_positions.shape[0]), (0, 0), (0, 0)))
 msa_feature['decoy_atom_mask'] = np.pad(final_atom_mask, ((0, 256 - final_atom_mask.shape[0]), (0, 0)))
 
-res = protein_assessment.predict(msa_feature)
-print("score is:", np.mean(res))
+res = protein_assessment.model.predict(msa_feature)
+print("score is:", np.mean(res[:msa_feature['num_residues']]))
 ```
 
 ### 使用场景
diff --git a/docs/api_python/mindflow/cell/mindflow.cell.MultiHeadAttention.rst b/docs/api_python/mindflow/cell/mindflow.cell.MultiHeadAttention.rst
index 4198935102cd0257fd8bef12d30421533156c1ef..5d245889df1929ca2b91d1fb11f5459be19d130b 100644
--- a/docs/api_python/mindflow/cell/mindflow.cell.MultiHeadAttention.rst
+++ b/docs/api_python/mindflow/cell/mindflow.cell.MultiHeadAttention.rst
@@ -10,7 +10,7 @@ mindflow.cell.MultiHeadAttention
         - **num_heads** (int) - 输出的输出特征维度。
         - **enable_flash_attn** (bool) - 是否使能FlashAttention。FlashAttention只支持 `Ascend` 后端。具体细节参见 `FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness <https://arxiv.org/abs/2205.14135>`_ 。
           默认值： ``False`` 。
-        - **fa_dtype** (mindspore.dtype): FlashAttention计算类型。支持以下类型： `mstype.bfloat16`、 `mstype.float16`。默认值： ``mstype.bfloat16`` ，表示 ``mindspore.bfloat16`` 。
+        - **fa_dtype** (mindspore.dtype) - FlashAttention计算类型。支持以下类型： `mstype.bfloat16`、 `mstype.float16`。默认值： ``mstype.bfloat16`` ，表示 ``mindspore.bfloat16`` 。
         - **drop_mode** (str) - dropout方式。默认值： ``dropout`` 。支持以下类型： ``dropout`` 和 ``droppath`` 。
         - **dropout_rate** (float) - dropout层丢弃的比率。取值在 `[0, 1]` 。默认值： ``0.0`` 。
         - **compute_dtype** (mindspore.dtype) - 网络层的数据类型。默认值： ``mstype.float32`` ，表示 ``mindspore.float32`` 。
diff --git a/tests/st/mindsponge/test_megaprotein/test_megaprotein.py b/tests/st/mindsponge/test_megaprotein/test_megaprotein.py
new file mode 100644
index 0000000000000000000000000000000000000000..b47f661c5db46e801e013a4ec8dd2b2cfc88917a
--- /dev/null
+++ b/tests/st/mindsponge/test_megaprotein/test_megaprotein.py
@@ -0,0 +1,32 @@
+import numpy as np
+import mindspore as ms
+from mindsponge import PipeLine
+
+ms.set_context(mode=ms.GRAPH_MODE)
+
+# MEGA-EvoGen������ȡ����������MSA�������
+fasta = "GYDKDLCEWSMTADQTEVETQIEADIMNIVKRDRPEMKAEVQKQLKSGGVMQYNYVLYCDKNFNNKNIIAEVVGE"
+msa_generator = PipeLine(name="MEGAEvoGen")
+msa_generator.set_device_id(0)
+msa_generator.initialize(key="evogen_predict_256")
+msa_generator.model.from_pretrained()
+msa_feature = msa_generator.predict(fasta)
+
+# MEGA-Fold������ȡ�����ʽṹ��Ϣ
+fold_prediction = PipeLine(name="MEGAFold")
+fold_prediction.set_device_id(0)
+fold_prediction.initialize(key="predict_256")
+fold_prediction.model.from_pretrained()
+final_atom_positions, final_atom_mask, aatype, _, _ = fold_prediction.model.predict(msa_feature)
+
+# MEGA-Assessment�Ե����ʽṹ��������
+protein_assessment = PipeLine(name = "MEGAAssessment")
+protein_assessment.set_device_id(0)
+protein_assessment.initialize("predict_256")
+protein_assessment.model.from_pretrained()
+msa_feature['decoy_aatype'] = np.pad(aatype, (0, 256 - aatype.shape[0]))
+msa_feature['decoy_atom_positions'] = np.pad(final_atom_positions, ((0, 256 - final_atom_positions.shape[0]), (0, 0), (0, 0)))
+msa_feature['decoy_atom_mask'] = np.pad(final_atom_mask, ((0, 256 - final_atom_mask.shape[0]), (0, 0)))
+
+res = protein_assessment.model.predict(msa_feature)
+print("score is:", np.mean(res[:msa_feature['num_residues']]))
\ No newline at end of file