diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/DataLoader.py b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/DataLoader.py
new file mode 100644
index 0000000000000000000000000000000000000000..c79b9ce0db3efdb5629e676c6dcc0cc5a12ca4bb
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/DataLoader.py
@@ -0,0 +1,118 @@
+#
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+#   list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+#   this list of conditions and the following disclaimer in the documentation
+#   and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+#   contributors may be used to endorse or promote products derived from
+#   this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+#
+import torch
+import torch.npu
+import os
+NPU_CALCULATE_DEVICE = 0
+if os.getenv('NPU_CALCULATE_DEVICE') and str.isdigit(os.getenv('NPU_CALCULATE_DEVICE')):
+    NPU_CALCULATE_DEVICE = int(os.getenv('NPU_CALCULATE_DEVICE'))
+if torch.npu.current_device() != NPU_CALCULATE_DEVICE:
+    torch.npu.set_device(f'npu:{NPU_CALCULATE_DEVICE}')
+
+class MyDataSet( torch.utils.data.Dataset):
+    def __init__( self, data_path, user_map, material_map, category_map, max_length):
+
+        user = []; material = []; category = []
+        material_historical = []; category_historical = []
+        material_historical_neg = []; category_historical_nge = []
+        mask = []; sequential_length = []
+        target = []
+
+        with open( data_path, 'r') as fin:
+
+            for line in fin:
+                item = line.strip('\n').split('\t')
+                if not item: continue
+
+                user.append( user_map.get( item[1], 0 ) ) 
+                material.append( material_map.get( item[2], 0 ) )
+                category.append( category_map.get( item[3], 0 ) )
+
+                material_historical_item = [0] * max_length
+                temp = item[4].split("")
+                if( len( temp) >= max_length): temp = temp[ -max_length:]
+                for i, m in enumerate( temp):
+                    material_historical_item[i] =  material_map.get( m, 0 ) 
+                material_historical.append( material_historical_item)
+                
+                category_historical_item = [0] * max_length
+                temp = item[5].split("")
+                if( len( temp) >= max_length): temp = temp[ -max_length:]
+                for i, c in enumerate( temp):
+                    category_historical_item[i] =  category_map.get( c, 0 ) 
+                category_historical.append( category_historical_item)
+
+                temp = min( len(temp), max_length)
+                mask_item = [1] * temp + [0] * ( max_length - temp)
+
+                mask.append( mask_item)
+                sequential_length.append( temp)
+
+                target.append( int( item[0]))
+        
+        self.user = torch.tensor( user)
+
+        self.material = torch.tensor( material)
+        self.catetory = torch.tensor( category)
+
+        self.material_historical = torch.tensor( material_historical)
+        self.catetory_historical = torch.tensor( category_historical)
+
+        self.mask = torch.tensor( mask).type( torch.FloatTensor)
+        self.sequential_length = torch.tensor( sequential_length).type( torch.FloatTensor)
+
+        self.target = torch.tensor( target).type( torch.FloatTensor)
+
+
+    def __len__( self):
+        return len( self.user)
+
+    def __getitem__(self, index):
+        if torch.is_tensor( index):
+            index = index.tolist()
+
+        user = self.user[ index]
+
+        material_historical = self.material_historical[ index, :]
+        category_historical = self.catetory_historical[ index, :]
+        mask = self.mask[ index, :]
+        sequential_length = self.sequential_length[ index]
+
+        material = self.material[ index]
+        category = self.catetory[ index]
+
+        target = self.target[ index]
+
+        return user, material_historical, category_historical, mask, sequential_length , \
+            material, category, 0, 0, target
\ No newline at end of file
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/LICENSE b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/LICENSE
new file mode 100644
index 0000000000000000000000000000000000000000..09d493bf1fc257505c1336f3f87425568ab9da3c
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/LICENSE
@@ -0,0 +1,29 @@
+BSD 3-Clause License
+
+Copyright (c) 2017, 
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+* Neither the name of the copyright holder nor the names of its
+  contributors may be used to endorse or promote products derived from
+  this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/README.md b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..09d4601fb61104bf96425847b391e307719e17e4
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/README.md
@@ -0,0 +1,45 @@
+# DIN/DIEN 
+
+Implementation based on pytorch for DIN recommendation algorithm
+
+
+## Attention
+
+1. For convenience, referring to authors tensorflow implementation, feature-embedding dimension is identical.
+2. Without any L1/L2 normalization or dropout strategy, it's supposed to choose suitable model according to the evaluation stage manually.
+
+## File description 
+|file name|description|
+|--|----|
+|main.ipynb|Session for training and evaluation|
+|model.py|Defination of target models|
+|DataLoader.py|Self-defined data loader|
+|environment.yml|Conda envrionment yaml|
+
+## Original paper
+[Deep Interest Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1706.06978.pdf)
+
+[Deep Interest Evolution Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1809.03672.pdf)
+
+## Source data
+[meta_Books.json.gz](http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/meta_Books.json.gz)
+
+[reviews_Books.json.gz](http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Books.json.gz)
+
+Preprocessed data wrapped within `data.tar.gz` came from [mouna99/dien](https://github.com/mouna99/dien)
+
+## Reference
+
+[mouna99/dien](https://github.com/mouna99/dien)
+
+[alibaba/x-deeplearning](https://github.com/alibaba/x-deeplearning)
+
+[shenweichen/DeepCTR-Torch](https://github.com/shenweichen/DeepCTR-Torch)
+
+
+## To do list
+
+- [x] DIN
+- [x] AUGRU 
+- [ ] DICE activation layer 
+- [ ] Auxialary loss with neg_sample  
\ No newline at end of file
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/environment.yml b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/environment.yml
new file mode 100644
index 0000000000000000000000000000000000000000..b21a399bff2773027ba05bc77a296671fa7cd331
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/environment.yml
@@ -0,0 +1,183 @@
+name: my_env
+channels:
+  - pytorch
+  - https://conda.anaconda.org/anaconda
+  - conda-forge
+  - defaults
+dependencies:
+  - _libgcc_mutex=0.1=main
+  - astroid=2.4.1=py38_0
+  - attrs=19.3.0=py_0
+  - backcall=0.1.0=py38_0
+  - blas=1.0=mkl
+  - bleach=3.1.4=py_0
+  - bzip2=1.0.8=h516909a_2
+  - ca-certificates=2020.6.24=0
+  - cairo=1.16.0=hcf35c78_1003
+  - certifi=2020.6.20=py38_0
+  - cloudpickle=1.4.1=py_0
+  - cudatoolkit=10.2.89=hfd86e86_1
+  - cycler=0.10.0=py_2
+  - cytoolz=0.10.1=py38h7b6447c_0
+  - dask-core=2.17.2=py_0
+  - dbus=1.13.14=hb2f20db_0
+  - decorator=4.4.2=py_0
+  - defusedxml=0.6.0=py_0
+  - entrypoints=0.3=py38_0
+  - expat=2.2.9=he1b5a44_2
+  - ffmpeg=4.2=h167e202_0
+  - fontconfig=2.13.1=h86ecdb6_1001
+  - freetype=2.9.1=h8a8886c_1
+  - gettext=0.19.8.1=hc5be6a0_1002
+  - giflib=5.2.1=h516909a_2
+  - glib=2.64.3=h6f030ca_0
+  - gmp=6.1.2=h6c8ec71_1
+  - gnutls=3.6.13=h79a8f9a_0
+  - graphite2=1.3.13=he1b5a44_1001
+  - gst-plugins-base=1.14.5=h0935bb2_2
+  - gstreamer=1.14.5=h36ae1b5_2
+  - harfbuzz=2.4.0=h9f30f68_3
+  - hdf5=1.10.6=nompi_h3c11f04_100
+  - icu=64.2=he1b5a44_1
+  - imageio=2.8.0=py_0
+  - importlib_metadata=1.5.0=py38_0
+  - intel-openmp=2020.0=166
+  - ipykernel=5.1.4=py38h39e3cac_0
+  - ipython=7.13.0=py38h5ca1d4c_0
+  - ipython_genutils=0.2.0=py38_0
+  - isort=4.3.21=py38_0
+  - jasper=1.900.1=h07fcdf6_1006
+  - jedi=0.17.0=py38_0
+  - jinja2=2.11.2=py_0
+  - joblib=0.16.0=py_0
+  - jpeg=9d=h516909a_0
+  - jsonschema=3.2.0=py38_0
+  - jupyter_client=6.1.3=py_0
+  - jupyter_core=4.6.3=py38_0
+  - kiwisolver=1.2.0=py38hbf85e49_0
+  - lame=3.100=h14c3975_1001
+  - lazy-object-proxy=1.4.3=py38h7b6447c_0
+  - ld_impl_linux-64=2.34=h53a641e_4
+  - libblas=3.8.0=15_mkl
+  - libcblas=3.8.0=15_mkl
+  - libclang=9.0.1=default_hde54327_0
+  - libedit=3.1.20181209=hc058e9b_0
+  - libffi=3.2.1=he1b5a44_1007
+  - libgcc-ng=9.1.0=hdf63c60_0
+  - libgfortran-ng=7.3.0=hdf63c60_0
+  - libiconv=1.15=h516909a_1006
+  - liblapack=3.8.0=15_mkl
+  - liblapacke=3.8.0=15_mkl
+  - libllvm9=9.0.1=he513fc3_1
+  - libopencv=4.2.0=py38_6
+  - libpng=1.6.37=hbc83047_0
+  - libsodium=1.0.16=h1bed415_0
+  - libstdcxx-ng=9.1.0=hdf63c60_0
+  - libtiff=4.1.0=h2733197_0
+  - libuuid=2.32.1=h14c3975_1000
+  - libwebp=1.0.2=h56121f0_5
+  - libxcb=1.13=h14c3975_1002
+  - libxkbcommon=0.10.0=he1b5a44_0
+  - libxml2=2.9.9=hea5a465_1
+  - markupsafe=1.1.1=py38h7b6447c_0
+  - matplotlib=3.2.1=0
+  - matplotlib-base=3.2.1=py38h2af1d28_0
+  - mccabe=0.6.1=py38_1
+  - mistune=0.8.4=py38h7b6447c_1000
+  - mkl=2020.0=166
+  - mkl-service=2.3.0=py38he904b0f_0
+  - mkl_fft=1.0.15=py38ha843d7b_0
+  - mkl_random=1.1.0=py38h962f231_0
+  - nb_conda_kernels=2.2.3=py38_0
+  - nbconvert=5.6.1=py38_0
+  - nbformat=5.0.6=py_0
+  - ncurses=6.2=he6710b0_1
+  - nettle=3.4.1=h1bed415_1002
+  - networkx=2.4=py_0
+  - ninja=1.9.0=py38hfd86e86_0
+  - notebook=6.0.3=py38_0
+  - nspr=4.25=he1b5a44_0
+  - nss=3.47=he751ad9_0
+  - numpy=1.18.1=py38h4f9e942_0
+  - numpy-base=1.18.1=py38hde5b4d6_1
+  - olefile=0.46=py_0
+  - opencv=4.2.0=py38_6
+  - openh264=1.8.0=hdbcaa40_1000
+  - openssl=1.1.1g=h7b6447c_0
+  - pandas=1.0.4=py38hcb8c335_0
+  - pandoc=2.2.3.2=0
+  - pandocfilters=1.4.2=py38_1
+  - parso=0.7.0=py_0
+  - pcre=8.44=he1b5a44_0
+  - pexpect=4.8.0=py38_0
+  - pickleshare=0.7.5=py38_1000
+  - pillow=7.1.2=py38hb39fc2d_0
+  - pip=20.0.2=py38_1
+  - pixman=0.38.0=h516909a_1003
+  - prometheus_client=0.7.1=py_0
+  - prompt-toolkit=3.0.4=py_0
+  - prompt_toolkit=3.0.4=0
+  - pthread-stubs=0.4=h14c3975_1001
+  - ptyprocess=0.6.0=py38_0
+  - py-opencv=4.2.0=py38h23f93f0_6
+  - pygments=2.6.1=py_0
+  - pylint=2.5.2=py38_0
+  - pyparsing=2.4.7=pyh9f0ad1d_0
+  - pyqt=5.12.3=py38ha8c2ead_3
+  - pyrsistent=0.16.0=py38h7b6447c_0
+  - python=3.8.3=cpython_he5300dc_0
+  - python-dateutil=2.8.1=py_0
+  - python_abi=3.8=1_cp38
+  - pytorch=1.5.0=py3.8_cuda10.2.89_cudnn7.6.5_0
+  - pytz=2020.1=pyh9f0ad1d_0
+  - pywavelets=1.1.1=py38h7b6447c_0
+  - pyyaml=5.3.1=py38h7b6447c_0
+  - pyzmq=18.1.1=py38he6710b0_0
+  - qt=5.12.5=hd8c4c69_1
+  - readline=8.0=h7b6447c_0
+  - redis=5.0.3=h7b6447c_0
+  - scikit-image=0.16.2=py38h0573a6f_0
+  - scikit-learn=0.23.1=py38h423224d_0
+  - scipy=1.4.1=py38h0b6359f_0
+  - send2trash=1.5.0=py38_0
+  - setuptools=46.1.3=py38_0
+  - sip=4.19.13=py38he6710b0_0
+  - six=1.14.0=py38_0
+  - sqlite=3.31.1=h62c20be_1
+  - terminado=0.8.3=py38_0
+  - testpath=0.4.4=py_0
+  - threadpoolctl=2.1.0=pyh5ca1d4c_0
+  - tk=8.6.10=hed695b0_0
+  - toml=0.10.0=pyh91ea838_0
+  - toolz=0.10.0=py_0
+  - torchvision=0.6.0=py38_cu102
+  - tornado=6.0.4=py38h7b6447c_1
+  - traitlets=4.3.3=py38_0
+  - wcwidth=0.1.9=py_0
+  - webencodings=0.5.1=py38_1
+  - wheel=0.34.2=py38_0
+  - wrapt=1.11.2=py38h7b6447c_0
+  - x264=1!152.20180806=h14c3975_0
+  - xorg-kbproto=1.0.7=h14c3975_1002
+  - xorg-libice=1.0.10=h516909a_0
+  - xorg-libsm=1.2.3=h84519dc_1000
+  - xorg-libx11=1.6.9=h516909a_0
+  - xorg-libxau=1.0.9=h14c3975_0
+  - xorg-libxdmcp=1.1.3=h516909a_0
+  - xorg-libxext=1.3.4=h516909a_0
+  - xorg-libxrender=0.9.10=h516909a_1002
+  - xorg-renderproto=0.11.1=h14c3975_1002
+  - xorg-xextproto=7.3.0=h14c3975_1002
+  - xorg-xproto=7.0.31=h14c3975_1007
+  - xz=5.2.5=h7b6447c_0
+  - yaml=0.1.7=h96e3832_1
+  - zeromq=4.3.1=he6710b0_3
+  - zipp=3.1.0=py_0
+  - zlib=1.2.11=h7b6447c_3
+  - zstd=1.3.7=h0b5b093_0
+  - pip:
+    - pyqt5-sip==4.19.18
+    - pyqtchart==5.12
+    - pyqtwebengine==5.12.1
+prefix: /home/juboge/opt/anaconda3/envs/my_env
+
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/layer.py b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/layer.py
new file mode 100644
index 0000000000000000000000000000000000000000..ca24a1b63bdff8c460286ef6a7c1a7cfb8a33c1a
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/layer.py
@@ -0,0 +1,130 @@
+#
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+#   list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+#   this list of conditions and the following disclaimer in the documentation
+#   and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+#   contributors may be used to endorse or promote products derived from
+#   this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+#
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.npu
+import os
+NPU_CALCULATE_DEVICE = 0
+if os.getenv('NPU_CALCULATE_DEVICE') and str.isdigit(os.getenv('NPU_CALCULATE_DEVICE')):
+    NPU_CALCULATE_DEVICE = int(os.getenv('NPU_CALCULATE_DEVICE'))
+if torch.npu.current_device() != NPU_CALCULATE_DEVICE:
+    torch.npu.set_device(f'npu:{NPU_CALCULATE_DEVICE}')
+
+class MLP( nn.Module):
+    
+    def __init__(self, input_dimension, hidden_size , target_dimension = 1, activation_layer = 'LeakyReLU'):
+        super().__init__()
+
+        Activation = nn.LeakyReLU
+
+        # if activation_layer == 'DICE': pass
+        # elif activation_layer == 'LeakyReLU': pass
+
+        def _dense( in_dim, out_dim, bias = False):
+            return nn.Sequential(
+                nn.Linear( in_dim, out_dim, bias = bias),
+                nn.BatchNorm1d( out_dim),
+                Activation( 0.1 ))
+
+        dimension_pair = [input_dimension] + hidden_size
+        layers = [ _dense( dimension_pair[i], dimension_pair[i+1]) for i in range( len( hidden_size))]
+
+        layers.append( nn.Linear( hidden_size[-1], target_dimension))
+        layers.insert( 0, nn.BatchNorm1d( input_dimension) )
+
+        self.model = nn.Sequential( *layers )
+    
+    def forward( self, X): return self.model( X)
+
+
+class InputEmbedding( nn.Module):
+
+    def __init__(self, n_uid, n_mid, n_cid, embedding_dim ):
+        super().__init__()
+        self.user_embedding_unit = nn.Embedding( n_uid, embedding_dim)
+        self.material_embedding_unit = nn.Embedding( n_mid, embedding_dim)
+        self.category_embedding_unit = nn.Embedding( n_cid, embedding_dim)
+
+    def forward( self, user, material, category, material_historical, category_historical, 
+                                       material_historical_neg, category_historical_neg, neg_smaple = False ):
+
+        user_embedding = self.user_embedding_unit( user)
+
+        material_embedding = self.material_embedding_unit( material)
+        material_historical_embedding = self.material_embedding_unit( material_historical)
+
+        category_embedding = self.category_embedding_unit( category)
+        category_historical_embedding = self.category_embedding_unit( category_historical)
+
+        material_historical_neg_embedding = self.material_embedding_unit( material_historical_neg) if neg_smaple else None  
+        category_historical_neg_embedding = self.category_embedding_unit( category_historical_neg) if neg_smaple else None
+
+        ans = [ user_embedding, material_historical_embedding, category_historical_embedding, 
+            material_embedding, category_embedding, material_historical_neg_embedding, category_historical_neg_embedding ]
+        return tuple( map( lambda x: x.squeeze() if x != None else None , ans) )
+
+
+
+class AttentionLayer( nn.Module):
+
+    def __init__(self, embedding_dim, hidden_size, activation_layer = 'sigmoid'):
+        super().__init__()
+
+        Activation = nn.Sigmoid
+        if activation_layer == 'Dice': pass
+        
+        def _dense( in_dim, out_dim):
+            return nn.Sequential( nn.Linear( in_dim, out_dim), Activation() )
+        
+        dimension_pair = [embedding_dim * 8] + hidden_size
+        layers = [ _dense( dimension_pair[i], dimension_pair[i+1]) for i in range( len( hidden_size))]
+        layers.append( nn.Linear( hidden_size[-1], 1) )
+        self.model = nn.Sequential( *layers)
+    
+    def forward( self, query, fact, mask, return_scores = False):
+        B, T, D = fact.shape
+        
+        query = torch.ones((B, T, 1), device=f'npu:{NPU_CALCULATE_DEVICE}', dtype=torch.float16) * query.view( (B, 1, D)) 
+        # query = query.view(-1).expand( T, -1).view( T, B, D).permute( 1, 0, 2)
+
+        combination = torch.cat( [ fact, query, fact * query, query - fact ], dim = 2)
+
+        scores = self.model( combination).squeeze()
+        scores = torch.where( mask == 1, scores, torch.ones_like( scores) * ( -2 ** 31 ) )
+
+        scores = ( scores.softmax( dim = -1) * mask ).view( (B , 1, T))
+
+        if return_scores: return scores.squeeze()
+        return torch.matmul( scores, fact).squeeze()
\ No newline at end of file
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/main.py b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/main.py
new file mode 100644
index 0000000000000000000000000000000000000000..3f465f4b2f7f6086a85b38511ce3a2b9c19d3513
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/main.py
@@ -0,0 +1,211 @@
+#
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+#   list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+#   this list of conditions and the following disclaimer in the documentation
+#   and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+#   contributors may be used to endorse or promote products derived from
+#   this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+#
+import torch
+import torch.nn as nn
+import os
+import sys
+import pickle as pk
+import numpy as np
+import random
+
+from sklearn.metrics import roc_auc_score
+import time
+import torch.npu
+import os
+import apex
+from apex import amp
+NPU_CALCULATE_DEVICE = 0
+if os.getenv('NPU_CALCULATE_DEVICE') and str.isdigit(os.getenv('NPU_CALCULATE_DEVICE')):
+    NPU_CALCULATE_DEVICE = int(os.getenv('NPU_CALCULATE_DEVICE'))
+if torch.npu.current_device() != NPU_CALCULATE_DEVICE:
+    torch.npu.set_device(f'npu:{NPU_CALCULATE_DEVICE}')
+
+workspace_dir = '.'
+try:
+    from google.colab import drive
+    drive.mount( '/content/drive/' )
+
+    workspace_dir = os.path.join( '.' , 'drive', 'My Drive', 'DIN-pytorch')
+    sys.path.append( workspace_dir)
+    #! rm -rf data
+    #! tar zxf "{workspace_dir/npu/traindata/ID2837_CarPeting_Pytorch_DIN.tar.gz" -C ./
+    #! tar zxf "{workspace_dir}/loader.tar.gz" -C ./
+    #! ls -al data   
+except ImportError:
+    pass
+
+from model import DIN, DIEN, DynamicGRU
+from DataLoader import MyDataSet
+
+
+#Model hyper parameter
+MAX_LEN = 100
+EMBEDDING_DIM = 32
+# HIDDEN_SIZE_ATTENTION = [80, 40]
+# HIDDEN_SIZE_FC = [200, 80]
+# ACTIVATION_LAYER = 'LeakyReLU' # lr = 0.01
+
+
+# Adam
+LR = 1e-3
+BETA1 = 0.5
+BETA2 = 0.99
+
+# Train
+BATCH_SIZE = 128
+EPOCH_TIME = 20
+TEST_ITER = 1000
+
+RANDOM_SEED = 19940808
+
+USE_CUDA = True
+
+train_file = os.path.join( '/npu/traindata/ID2837_CarPeting_Pytorch_DIN', "local_train_splitByUser")
+test_file  = os.path.join( '/npu/traindata/ID2837_CarPeting_Pytorch_DIN', "local_test_splitByUser")
+uid_voc    = os.path.join( '/npu/traindata/ID2837_CarPeting_Pytorch_DIN', "uid_voc.pkl")
+mid_voc    = os.path.join( '/npu/traindata/ID2837_CarPeting_Pytorch_DIN', "mid_voc.pkl")
+cat_voc    = os.path.join( '/npu/traindata/ID2837_CarPeting_Pytorch_DIN', "cat_voc.pkl")
+
+if USE_CUDA and torch.npu.is_available():
+    print( "Cuda is avialable" )
+    device = torch.device(f'npu:{NPU_CALCULATE_DEVICE}')
+    dtype = torch.npu.FloatTensor
+else:
+    device = torch.device( f'npu:{NPU_CALCULATE_DEVICE}')
+    dtype = torch.FloatTensor
+
+# Stable the random seed
+def same_seeds(seed = RANDOM_SEED):
+    torch.manual_seed(seed)
+    if torch.npu.is_available():
+        torch.npu.manual_seed(seed)
+        torch.npu.manual_seed_all(seed)  # if you are using multi-GPU.
+    np.random.seed(seed)  
+    random.seed(seed) 
+    torch.backends.cudnn.benchmark = False
+    torch.backends.cudnn.deterministic = True
+
+# Initilize  parameters
+def weights_init( m):
+    try:
+        classname = m.__class__.__name__
+        if classname.find( 'BatchNorm') != -1:
+            nn.init.normal_( m.weight.data, 1.0, 0.02)
+            nn.init.constant_( m.bias.data, 0)
+        elif classname.find( 'Linear') != -1:
+            nn.init.normal_( m.weight.data, 0.0, 0.02)
+        elif classname.find( 'Embedding') != -1:
+            m.weight.data.uniform_(-1, 1)
+    except AttributeError:
+        print( "AttributeError:", classname)
+    
+
+
+def eval_output( scores, target, loss_function = torch.nn.functional.binary_cross_entropy_with_logits):
+    loss = loss_function( scores, target)
+
+    y_pred = scores.sigmoid().round()
+
+    accuracy = ( y_pred == target).type( dtype).mean()
+
+    auc = roc_auc_score( target.cpu().detach(), scores.cpu().detach() )
+    return loss, accuracy, auc
+
+# The dict mapping description(string) to type index(int) 
+# A more graceful api https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html#sklearn.preprocessing.LabelEncoder not used in this project
+
+user_map = pk.load( open( uid_voc, 'rb')); n_uid = len( user_map)
+material_map = pk.load( open( mid_voc, 'rb')); n_mid = len( material_map)
+category_map = pk.load( open( cat_voc, 'rb')); n_cat = len( category_map)
+
+same_seeds( RANDOM_SEED)
+
+dataset_train = MyDataSet( train_file, user_map, material_map, category_map, max_length = MAX_LEN)
+dataset_test = MyDataSet( test_file, user_map, material_map, category_map, max_length = MAX_LEN)
+
+loader_train = torch.utils.data.DataLoader( dataset_train, batch_size = BATCH_SIZE, shuffle = True, pin_memory=True)
+loader_test = torch.utils.data.DataLoader( dataset_test, batch_size = BATCH_SIZE, shuffle = False)
+
+# with open( 'data/loader.pk', 'rb') as fin:
+#     loader_train, loader_test = pk.load(fin) 
+
+# Get model and initialize it
+# model = DIEN(  n_uid, n_mid, n_cat, EMBEDDING_DIM).to( device)
+model = DIN(  n_uid, n_mid, n_cat, EMBEDDING_DIM ).to( f'npu:{NPU_CALCULATE_DEVICE}')
+model.apply( weights_init)
+
+# Set loss function and optimizer
+optimizer = apex.optimizers.NpuFusedAdam(model.parameters(), LR, (BETA1, BETA2))
+model, optimizer = amp.initialize(model, optimizer, opt_level="O2", loss_scale=128.0, combine_grad=True)
+model.train(); iter = 0
+for epoch in range( EPOCH_TIME):
+    for i, data in enumerate( loader_train):
+        if i >= 1000:pass
+        start_time = time.time()
+        iter += 1
+
+        # transform data to target device
+   
+        data = [ item.to( f'npu:{NPU_CALCULATE_DEVICE}', non_blocking=True) if item != None else None for item in data]
+        target = data.pop(-1)     
+        
+        model.zero_grad()
+
+        scores = model( data, neg_sample = False)
+        
+        loss, accuracy, auc = eval_output( scores, target)
+        with amp.scale_loss(loss, optimizer) as scaled_loss:
+            scaled_loss.backward()
+        optimizer.step( )
+        step_time = time.time() - start_time
+        FPS = BATCH_SIZE / step_time
+        print( "Epoch:{}, step:{}, loss:{:.4f}, Acc:{:.4f},Auc:{:.4f}, time/step(s):{:.4f},FPS:{:.3f}".format( epoch + 1, i + 1, loss.item(), accuracy.item(), auc.item(), step_time, FPS))
+
+        if iter % TEST_ITER == 0:
+            model.eval()
+            with torch.no_grad():
+                score_list = []; target_list = []
+                for data in  loader_test:
+                    data = [ item.to( f'npu:{NPU_CALCULATE_DEVICE}') if item != None else None for item in data]
+                    
+                    target = data.pop(-1)
+
+                    scores = model( data, neg_sample = False)
+                    score_list.append( scores)
+                    target_list.append( target)
+                scores = torch.cat( score_list, dim = -1)
+                target = torch.cat( target_list, dim = -1)
+                loss, accuracy, auc = eval_output( scores, target)
+                print( "\tTest Set\tloss:%.5f\tacc:%.5f\tauc:%.5f"%( loss.item(), accuracy.item(), auc.item() ) )
+            model.train()
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/model.py b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/model.py
new file mode 100644
index 0000000000000000000000000000000000000000..d84812b873531d87d31147478f16e36b1db0a751
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/model.py
@@ -0,0 +1,119 @@
+#
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+#   list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+#   this list of conditions and the following disclaimer in the documentation
+#   and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+#   contributors may be used to endorse or promote products derived from
+#   this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+#
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from layer import *
+from rnn import *
+import torch.npu
+import os
+NPU_CALCULATE_DEVICE = 0
+if os.getenv('NPU_CALCULATE_DEVICE') and str.isdigit(os.getenv('NPU_CALCULATE_DEVICE')):
+    NPU_CALCULATE_DEVICE = int(os.getenv('NPU_CALCULATE_DEVICE'))
+if torch.npu.current_device() != NPU_CALCULATE_DEVICE:
+    torch.npu.set_device(f'npu:{NPU_CALCULATE_DEVICE}')
+
+class DIN( nn.Module):
+    def __init__(self, n_uid, n_mid, n_cid, embedding_dim, ):
+        super().__init__()
+
+        self.embedding_layer = InputEmbedding( n_uid, n_mid, n_cid, embedding_dim )
+        self.attention_layer = AttentionLayer( embedding_dim, hidden_size = [ 80, 40], activation_layer='sigmoid')
+        # self.output_layer = MLP( embedding_dim * 9, [ 200, 80], 1, 'ReLU')
+        self.output_layer = MLP( embedding_dim * 7, [ 200, 80], 1, 'ReLU')
+
+    def forward( self, data, neg_sample = False):
+                            
+        user, material_historical, category_historical, mask, sequential_length , material, category, \
+            material_historical_neg, category_historical_neg = data
+        
+        user_embedding, material_historical_embedding, category_historical_embedding, \
+            material_embedding, category_embedding, material_historical_neg_embedding, category_historical_neg_embedding = \
+        self.embedding_layer(  user, material, category, material_historical, category_historical, material_historical_neg, category_historical_neg, neg_sample)
+
+        item_embedding = torch.cat( [ material_embedding, category_embedding], dim = 1)
+        item_historical_embedding = torch.cat( [ material_historical_embedding, category_historical_embedding], dim = 2 )
+
+        item_historical_embedding_sum = torch.matmul( mask.unsqueeze( dim = 1), item_historical_embedding).squeeze() / sequential_length.unsqueeze( dim = 1)
+
+
+        attention_feature = self.attention_layer( item_embedding, item_historical_embedding, mask)
+
+        # combination = torch.cat( [ user_embedding, item_embedding, item_historical_embedding_sum, attention_feature ], dim = 1)
+        combination = torch.cat( [ user_embedding, item_embedding, item_historical_embedding_sum, 
+                                    # item_embedding * item_historical_embedding_sum, 
+                                    attention_feature ], dim = 1)
+
+        scores = self.output_layer( combination)
+
+        return scores.squeeze()
+
+class DIEN( nn.Module):
+    def __init__(self, n_uid, n_mid, n_cid, embedding_dim):
+        super().__init__()
+
+        self.embedding_layer = InputEmbedding( n_uid, n_mid, n_cid, embedding_dim )
+        self.gru_based_layer = nn.GRU( embedding_dim * 2 , embedding_dim * 2, batch_first = True)
+        self.attention_layer = AttentionLayer( embedding_dim, hidden_size = [ 80, 40], activation_layer='sigmoid')
+        self.gru_customized_layer = DynamicGRU( embedding_dim * 2, embedding_dim * 2)
+
+        self.output_layer = MLP( embedding_dim * 9, [ 200, 80], 1, 'ReLU')
+        # self.output_layer = MLP( embedding_dim * 9, [ 200, 80], 1, 'ReLU')
+
+    def forward( self, data, neg_sample = False):
+                            
+        user, material_historical, category_historical, mask, sequential_length , material, category, \
+            material_historical_neg, category_historical_neg = data
+        
+        user_embedding, material_historical_embedding, category_historical_embedding, \
+            material_embedding, category_embedding, material_historical_neg_embedding, category_historical_neg_embedding = \
+        self.embedding_layer(  user, material, category, material_historical, category_historical, material_historical_neg, category_historical_neg, neg_sample)
+
+        item_embedding = torch.cat( [ material_embedding, category_embedding], dim = 1)
+        item_historical_embedding = torch.cat( [ material_historical_embedding, category_historical_embedding], dim = 2 )
+
+        item_historical_embedding_sum = torch.matmul( mask.unsqueeze( dim = 1), item_historical_embedding).squeeze() / sequential_length.unsqueeze( dim = 1)
+
+        output_based_gru, _ = self.gru_based_layer( item_historical_embedding)
+        attention_scores = self.attention_layer( item_embedding, output_based_gru, mask, return_scores = True)
+        output_customized_gru = self.gru_customized_layer( output_based_gru, attention_scores)
+
+        attention_feature = output_customized_gru[  range( len( sequential_length)), sequential_length - 1]
+
+        combination = torch.cat( [ user_embedding, item_embedding, item_historical_embedding_sum, item_embedding * item_historical_embedding_sum, attention_feature ], dim = 1)
+
+        scores = self.output_layer( combination)
+
+        return scores.squeeze()
\ No newline at end of file
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/modelzoo_level.txt b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/modelzoo_level.txt
new file mode 100644
index 0000000000000000000000000000000000000000..c45626e398eabe6022fe7b2e148f0ffce6400d6e
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/modelzoo_level.txt
@@ -0,0 +1,3 @@
+FuncStatus:OK
+PerfStatus:POK
+PrecisionStatus:OK
\ No newline at end of file
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/requirements.txt b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/rnn.py b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/rnn.py
new file mode 100644
index 0000000000000000000000000000000000000000..5b30b43b5049b2366edb87d1c41ad6f2740996aa
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/rnn.py
@@ -0,0 +1,82 @@
+#
+# BSD 3-Clause License
+#
+# Copyright (c) 2017 xxxx
+# All rights reserved.
+# Copyright 2021 Huawei Technologies Co., Ltd
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+#   list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+#   this list of conditions and the following disclaimer in the documentation
+#   and/or other materials provided with the distribution.
+#
+# * Neither the name of the copyright holder nor the names of its
+#   contributors may be used to endorse or promote products derived from
+#   this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# ============================================================================
+#
+import torch
+import torch.nn as nn
+import torch.npu
+import os
+NPU_CALCULATE_DEVICE = 0
+if os.getenv('NPU_CALCULATE_DEVICE') and str.isdigit(os.getenv('NPU_CALCULATE_DEVICE')):
+    NPU_CALCULATE_DEVICE = int(os.getenv('NPU_CALCULATE_DEVICE'))
+if torch.npu.current_device() != NPU_CALCULATE_DEVICE:
+    torch.npu.set_device(f'npu:{NPU_CALCULATE_DEVICE}')
+
+class AUGRUCell(nn.Module):
+    def __init__(self, input_dim, hidden_dim, bias = True):
+        super(AUGRUCell, self).__init__()
+
+        in_dim = input_dim + hidden_dim
+        self.reset_gate = nn.Sequential( nn.Linear( in_dim, hidden_dim, bias = bias), nn.Sigmoid())
+        self.update_gate = nn.Sequential( nn.Linear( in_dim, hidden_dim, bias = bias), nn.Sigmoid())
+        self.h_hat_gate = nn.Sequential( nn.Linear( in_dim, hidden_dim, bias = bias), nn.Tanh())
+
+
+    def forward(self, X, h_prev, attention_score):
+        temp_input = torch.cat( [ h_prev, X ] , dim = -1)
+        r = self.reset_gate( temp_input)
+        u = self.update_gate( temp_input)
+
+        h_hat = self.h_hat_gate( torch.cat( [ h_prev * r, X], dim = -1) )
+
+        u = attention_score.unsqueeze(1) * u
+        h_cur = (1. - u) * h_prev + u * h_hat
+
+        return h_cur
+
+
+class DynamicGRU(nn.Module):
+    def __init__(self, input_dim, hidden_dim, bias=True):
+        super().__init__()
+        self.input_dim = input_dim
+        self.hidden_dim = hidden_dim
+        self.rnn_cell = AUGRUCell( input_dim, hidden_dim, bias = True)
+
+    def forward(self, X, attenion_scores , h0 = None ):
+        B, T, D = X.shape
+        H = self.hidden_dim
+        
+        output = torch.zeros( B, T, H ).type( X.type() )
+        h_prev = torch.zeros( B, H ).type( X.type() ) if h0 == None else h0
+        for t in range( T): 
+            h_prev = output[ : , t, :] = self.rnn_cell( X[ : , t, :], h_prev, attenion_scores[ :, t] )
+        return output
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_full_1p.sh b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_full_1p.sh
new file mode 100644
index 0000000000000000000000000000000000000000..c47f7bc0e5ffd0a3bb309b7ad9348f1c6d37ed17
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_full_1p.sh
@@ -0,0 +1,186 @@
+#!/bin/bash
+
+#当前路径,不需要修改
+cur_path=`pwd`
+#export ASCEND_SLOG_PRINT_TO_STDOUT=1
+export NPU_CALCULATE_DEVICE=$ASCEND_DEVICE_ID
+
+#集合通信参数,不需要修改
+export RANK_SIZE=1
+export JOB_ID=10087
+RANK_ID_START=0
+
+
+# 数据集路径,保持为空,不需要修改
+data_path=""
+
+#基础参数，需要模型审视修改
+#网络名称，同目录名称
+Network="DIN_ID2837_for_PyTorch"
+#训练epoch
+train_epochs=20
+#训练batch_size
+batch_size=128
+#训练step
+#train_steps=`expr 1281167 / ${batch_size}`
+#学习率
+learning_rate=0.495
+
+#TF2.X独有，不需要修改
+#export NPU_LOOP_SIZE=${train_steps}
+
+#维测参数，precision_mode需要模型审视修改
+precision_mode="allow_mix_precision"
+#维持参数，以下不需要修改
+over_dump=False
+data_dump_flag=False
+data_dump_step="10"
+profiling=False
+autotune=False
+
+# 帮助信息，不h需要修改
+if [[ $1 == --help || $1 == -h ]];then
+    echo"usage:./train_full_1p.sh <args>"
+    echo " "
+    echo "parameter explain:
+    --precision_mode         precision mode(allow_fp32_to_fp16/force_fp16/must_keep_origin_dtype/allow_mix_precision)
+    --over_dump		         if or not over detection, default is False
+    --data_dump_flag	     data dump flag, default is False
+    --data_dump_step		 data dump step, default is 10
+    --profiling		         if or not profiling for performance debug, default is False
+    --data_path		         source data of training
+    -h/--help		         show help message
+    "
+    exit 1
+fi
+
+#参数校验，不需要修改
+for para in $*
+do
+    if [[ $para == --precision_mode* ]];then
+        precision_mode=`echo ${para#*=}`
+    elif [[ $para == --over_dump* ]];then
+        over_dump=`echo ${para#*=}`
+        over_dump_path=${cur_path}/output/overflow_dump
+        mkdir -p ${over_dump_path}
+    elif [[ $para == --data_dump_flag* ]];then
+        data_dump_flag=`echo ${para#*=}`
+        data_dump_path=${cur_path}/output/data_dump
+        mkdir -p ${data_dump_path}
+    elif [[ $para == --data_dump_step* ]];then
+        data_dump_step=`echo ${para#*=}`
+    elif [[ $para == --profiling* ]];then
+        profiling=`echo ${para#*=}`
+        profiling_dump_path=${cur_path}/output/profiling
+        mkdir -p ${profiling_dump_path}
+    elif [[ $para == --data_path* ]];then
+        data_path=`echo ${para#*=}`
+    fi
+done
+
+#校验是否传入data_path,不需要修改
+if [[ $data_path == "" ]];then
+    echo "[Error] para \"data_path\" must be confing"
+    exit 1
+fi
+
+#训练开始时间，不需要修改
+start_time=$(date +%s)
+
+#进入训练脚本目录，需要模型审视修改
+cd $cur_path/../
+
+
+sed -i "s|./data|$data_path|g" main.py
+#sed -i "s|EPOCH_TIME = 20|EPOCH_TIME = 1|g" main.py
+#sed -i "s|continue|break|g" main.py
+
+#python3 setup.py install
+#mkdir -p checkpoints
+#mkdir -p /root/.cache/torch/hub/checkpoints
+#cp $data_path/fcn_* /root/.cache/torch/hub/checkpoints
+
+for((RANK_ID=$RANK_ID_START;RANK_ID<$((RANK_SIZE+RANK_ID_START));RANK_ID++));
+do
+    #设置环境变量，不需要修改
+    echo "Device ID: $ASCEND_DEVICE_ID"
+    export RANK_ID=$RANK_ID
+    
+    
+    
+    #创建DeviceID输出目录，不需要修改
+    if [ -d ${cur_path}/output/${ASCEND_DEVICE_ID} ];then
+        rm -rf ${cur_path}/output/${ASCEND_DEVICE_ID}
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    else
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    fi
+    
+    #绑核，不需要绑核的模型删除，需要绑核的模型根据实际修改
+    #cpucount=`lscpu | grep "CPU(s):" | head -n 1 | awk '{print $2}'`
+    #cpustep=`expr $cpucount / 8`
+    #echo "taskset c steps:" $cpustep
+    #let a=RANK_ID*$cpustep
+    #let b=RANK_ID+1
+    #let c=b*$cpustep-1
+
+    #执行训练脚本，以下传参不需要修改，其他需要模型审视修改
+    nohup python3 main.py  > ${cur_path}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log 2>&1 &
+done 
+wait
+
+#恢复参数
+sed -i "s|$data_path|./data|g" main.py
+#sed -i "s|EPOCH_TIME = 1|EPOCH_TIME = 20|g" main.py
+#sed -i "s|break|continue|g" main.py
+
+#conda deactivate
+#训练结束时间，不需要修改
+end_time=$(date +%s)
+e2e_time=$(( $end_time - $start_time ))
+
+#结果打印，不需要修改
+echo "------------------ Final result ------------------"
+#输出性能FPS，需要模型审视修改
+FPS=`grep FPS  $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F "FPS:" '{print $2}'|tail -n +4|awk '{sum+=$1} END {print"",sum/NR}'|sed s/[[:space:]]//g`
+#FPS=`awk 'BEGIN{printf "%.2f\n",'${batch_size}'*'${perf}'}'`
+
+
+#打印，不需要修改
+echo "Final Performance images/sec : $FPS"
+
+#输出训练精度,需要模型审视修改
+train_accuracy=`grep "Test Set" $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F ':' '{print $3}' | awk -F 'auc' '{print $1}' | awk 'NR==1{max=$1;next}{max=max>$1?max:$1}END{print max}'|sed s/[[:space:]]//g`
+#打印，不需要修改
+echo "Final Train Accuracy : ${train_accuracy}"
+echo "E2E Training Duration sec : $e2e_time"
+
+#稳定性精度看护结果汇总
+#训练用例信息，不需要修改
+BatchSize=${batch_size}
+DeviceType=`uname -m`
+CaseName=${Network}_bs${BatchSize}_${RANK_SIZE}'p'_'acc'
+
+##获取性能数据
+#吞吐量，不需要修改
+ActualFPS=${FPS}
+#单迭代训练时长，不需要修改
+TrainingTime=`awk 'BEGIN{printf "%.2f\n",'${BatchSize}'*1000/'${FPS}'}'`
+
+#从train_$ASCEND_DEVICE_ID.log提取Loss到train_${CaseName}_loss.txt中，需要根据模型审视
+grep "FPS" $cur_path/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log|awk -F "," '{print $3}'| awk -F "loss:" '{print $2}' >> $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt
+
+#最后一个迭代loss值，不需要修改
+ActualLoss=`awk 'END {print}' $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt`
+
+#关键信息打印到${CaseName}.log中，不需要修改
+echo "Network = ${Network}" > $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "RankSize = ${RANK_SIZE}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "BatchSize = ${BatchSize}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "DeviceType = ${DeviceType}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "CaseName = ${CaseName}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualFPS = ${ActualFPS}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "TrainingTime = ${TrainingTime}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "TrainAccuracy = ${train_accuracy}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualLoss = ${ActualLoss}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "E2ETrainingTime = ${e2e_time}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p.sh b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p.sh
new file mode 100644
index 0000000000000000000000000000000000000000..b3c181b9b8dd7510dba822295b00c4ec61512a9d
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p.sh
@@ -0,0 +1,186 @@
+#!/bin/bash
+
+#当前路径,不需要修改
+cur_path=`pwd`
+export ASCEND_SLOG_PRINT_TO_STDOUT=0
+export NPU_CALCULATE_DEVICE=$ASCEND_DEVICE_ID
+
+#集合通信参数,不需要修改
+export RANK_SIZE=1
+export JOB_ID=10087
+RANK_ID_START=0
+
+
+# 数据集路径,保持为空,不需要修改
+data_path=""
+
+#基础参数，需要模型审视修改
+#网络名称，同目录名称
+Network="DIN_ID2837_for_PyTorch"
+#训练epoch
+train_epochs=1
+#训练batch_size
+batch_size=128
+#训练step
+#train_steps=`expr 1281167 / ${batch_size}`
+#学习率
+learning_rate=0.495
+
+#TF2.X独有，不需要修改
+#export NPU_LOOP_SIZE=${train_steps}
+
+#维测参数，precision_mode需要模型审视修改
+precision_mode="allow_mix_precision"
+#维持参数，以下不需要修改
+over_dump=False
+data_dump_flag=False
+data_dump_step="10"
+profiling=False
+autotune=False
+
+# 帮助信息，不h需要修改
+if [[ $1 == --help || $1 == -h ]];then
+    echo"usage:./train_full_1p.sh <args>"
+    echo " "
+    echo "parameter explain:
+    --precision_mode         precision mode(allow_fp32_to_fp16/force_fp16/must_keep_origin_dtype/allow_mix_precision)
+    --over_dump		         if or not over detection, default is False
+    --data_dump_flag	     data dump flag, default is False
+    --data_dump_step		 data dump step, default is 10
+    --profiling		         if or not profiling for performance debug, default is False
+    --data_path		         source data of training
+    -h/--help		         show help message
+    "
+    exit 1
+fi
+
+#参数校验，不需要修改
+for para in $*
+do
+    if [[ $para == --precision_mode* ]];then
+        precision_mode=`echo ${para#*=}`
+    elif [[ $para == --over_dump* ]];then
+        over_dump=`echo ${para#*=}`
+        over_dump_path=${cur_path}/output/overflow_dump
+        mkdir -p ${over_dump_path}
+    elif [[ $para == --data_dump_flag* ]];then
+        data_dump_flag=`echo ${para#*=}`
+        data_dump_path=${cur_path}/output/data_dump
+        mkdir -p ${data_dump_path}
+    elif [[ $para == --data_dump_step* ]];then
+        data_dump_step=`echo ${para#*=}`
+    elif [[ $para == --profiling* ]];then
+        profiling=`echo ${para#*=}`
+        profiling_dump_path=${cur_path}/output/profiling
+        mkdir -p ${profiling_dump_path}
+    elif [[ $para == --data_path* ]];then
+        data_path=`echo ${para#*=}`
+    fi
+done
+
+#校验是否传入data_path,不需要修改
+if [[ $data_path == "" ]];then
+    echo "[Error] para \"data_path\" must be confing"
+    exit 1
+fi
+
+#训练开始时间，不需要修改
+start_time=$(date +%s)
+
+#进入训练脚本目录，需要模型审视修改
+cd $cur_path/../
+
+
+sed -i "s|./data|$data_path|g" main.py
+sed -i "s|EPOCH_TIME = 20|EPOCH_TIME = 1|g" main.py
+sed -i "s|if i >= 1000:pass|if i >= 1000:break|g" main.py
+
+#python3 setup.py install
+#mkdir -p checkpoints
+#mkdir -p /root/.cache/torch/hub/checkpoints
+#cp $data_path/fcn_* /root/.cache/torch/hub/checkpoints
+
+for((RANK_ID=$RANK_ID_START;RANK_ID<$((RANK_SIZE+RANK_ID_START));RANK_ID++));
+do
+    #设置环境变量，不需要修改
+    echo "Device ID: $ASCEND_DEVICE_ID"
+    export RANK_ID=$RANK_ID
+    
+    
+    
+    #创建DeviceID输出目录，不需要修改
+    if [ -d ${cur_path}/output/${ASCEND_DEVICE_ID} ];then
+        rm -rf ${cur_path}/output/${ASCEND_DEVICE_ID}
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    else
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    fi
+    
+    #绑核，不需要绑核的模型删除，需要绑核的模型根据实际修改
+    #cpucount=`lscpu | grep "CPU(s):" | head -n 1 | awk '{print $2}'`
+    #cpustep=`expr $cpucount / 8`
+    #echo "taskset c steps:" $cpustep
+    #let a=RANK_ID*$cpustep
+    #let b=RANK_ID+1
+    #let c=b*$cpustep-1
+
+    #执行训练脚本，以下传参不需要修改，其他需要模型审视修改
+    nohup python3 main.py  > ${cur_path}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log 2>&1 &
+done 
+wait
+
+#恢复参数
+sed -i "s|$data_path|./data|g" main.py
+sed -i "s|EPOCH_TIME = 1|EPOCH_TIME = 20|g" main.py
+sed -i "s|if i >= 1000:break|if i >= 1000:pass|g" main.py
+
+#conda deactivate
+#训练结束时间，不需要修改
+end_time=$(date +%s)
+e2e_time=$(( $end_time - $start_time ))
+
+#结果打印，不需要修改
+echo "------------------ Final result ------------------"
+#输出性能FPS，需要模型审视修改
+FPS=`grep FPS  $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F "FPS:" '{print $2}'|tail -n +4|awk '{sum+=$1} END {print"",sum/NR}'|sed s/[[:space:]]//g`
+#FPS=`awk 'BEGIN{printf "%.2f\n",'${batch_size}'*'${perf}'}'`
+
+
+#打印，不需要修改
+echo "Final Performance images/sec : $FPS"
+
+#输出训练精度,需要模型审视修改
+#train_accuracy=`grep eval_accuracy $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|grep -v mlp_log|awk 'END {print $5}'| sed 's/,//g' |cut -c 1-5`
+#打印，不需要修改
+#echo "Final Train Accuracy : ${train_accuracy}"
+#echo "E2E Training Duration sec : $e2e_time"
+
+#稳定性精度看护结果汇总
+#训练用例信息，不需要修改
+BatchSize=${batch_size}
+DeviceType=`uname -m`
+CaseName=${Network}_bs${BatchSize}_${RANK_SIZE}'p'_'perf'
+
+##获取性能数据
+#吞吐量，不需要修改
+ActualFPS=${FPS}
+#单迭代训练时长，不需要修改
+TrainingTime=`awk 'BEGIN{printf "%.2f\n",'${BatchSize}'*1000/'${FPS}'}'`
+
+#从train_$ASCEND_DEVICE_ID.log提取Loss到train_${CaseName}_loss.txt中，需要根据模型审视
+grep "FPS" $cur_path/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log|awk -F "," '{print $3}'| awk -F "loss:" '{print $2}' >> $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt
+
+#最后一个迭代loss值，不需要修改
+ActualLoss=`awk 'END {print}' $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt`
+
+#关键信息打印到${CaseName}.log中，不需要修改
+echo "Network = ${Network}" > $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "RankSize = ${RANK_SIZE}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "BatchSize = ${BatchSize}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "DeviceType = ${DeviceType}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "CaseName = ${CaseName}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualFPS = ${ActualFPS}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "TrainingTime = ${TrainingTime}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+#echo "TrainAccuracy = ${train_accuracy}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualLoss = ${ActualLoss}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "E2ETrainingTime = ${e2e_time}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
diff --git a/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p_success.sh b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p_success.sh
new file mode 100644
index 0000000000000000000000000000000000000000..b0c942106dc664528d46c9b46a86910bae0500ce
--- /dev/null
+++ b/PyTorch/dev/cv/image_classification/DIN_ID2837_for_PyTorch/test/train_performance_1p_success.sh
@@ -0,0 +1,186 @@
+#!/bin/bash
+
+#当前路径,不需要修改
+cur_path=`pwd`
+export ASCEND_SLOG_PRINT_TO_STDOUT=1
+export NPU_CALCULATE_DEVICE=$ASCEND_DEVICE_ID
+
+#集合通信参数,不需要修改
+export RANK_SIZE=1
+export JOB_ID=10087
+RANK_ID_START=0
+
+
+# 数据集路径,保持为空,不需要修改
+data_path=""
+
+#基础参数，需要模型审视修改
+#网络名称，同目录名称
+Network="DIN_ID2837_for_PyTorch"
+#训练epoch
+train_epochs=1
+#训练batch_size
+batch_size=128
+#训练step
+#train_steps=`expr 1281167 / ${batch_size}`
+#学习率
+learning_rate=0.495
+
+#TF2.X独有，不需要修改
+#export NPU_LOOP_SIZE=${train_steps}
+
+#维测参数，precision_mode需要模型审视修改
+precision_mode="allow_mix_precision"
+#维持参数，以下不需要修改
+over_dump=False
+data_dump_flag=False
+data_dump_step="10"
+profiling=False
+autotune=False
+
+# 帮助信息，不h需要修改
+if [[ $1 == --help || $1 == -h ]];then
+    echo"usage:./train_full_1p.sh <args>"
+    echo " "
+    echo "parameter explain:
+    --precision_mode         precision mode(allow_fp32_to_fp16/force_fp16/must_keep_origin_dtype/allow_mix_precision)
+    --over_dump		         if or not over detection, default is False
+    --data_dump_flag	     data dump flag, default is False
+    --data_dump_step		 data dump step, default is 10
+    --profiling		         if or not profiling for performance debug, default is False
+    --data_path		         source data of training
+    -h/--help		         show help message
+    "
+    exit 1
+fi
+
+#参数校验，不需要修改
+for para in $*
+do
+    if [[ $para == --precision_mode* ]];then
+        precision_mode=`echo ${para#*=}`
+    elif [[ $para == --over_dump* ]];then
+        over_dump=`echo ${para#*=}`
+        over_dump_path=${cur_path}/output/overflow_dump
+        mkdir -p ${over_dump_path}
+    elif [[ $para == --data_dump_flag* ]];then
+        data_dump_flag=`echo ${para#*=}`
+        data_dump_path=${cur_path}/output/data_dump
+        mkdir -p ${data_dump_path}
+    elif [[ $para == --data_dump_step* ]];then
+        data_dump_step=`echo ${para#*=}`
+    elif [[ $para == --profiling* ]];then
+        profiling=`echo ${para#*=}`
+        profiling_dump_path=${cur_path}/output/profiling
+        mkdir -p ${profiling_dump_path}
+    elif [[ $para == --data_path* ]];then
+        data_path=`echo ${para#*=}`
+    fi
+done
+
+#校验是否传入data_path,不需要修改
+if [[ $data_path == "" ]];then
+    echo "[Error] para \"data_path\" must be confing"
+    exit 1
+fi
+
+#训练开始时间，不需要修改
+start_time=$(date +%s)
+
+#进入训练脚本目录，需要模型审视修改
+cd $cur_path/../
+
+
+sed -i "s|./data|$data_path|g" main.py
+sed -i "s|EPOCH_TIME = 20|EPOCH_TIME = 1|g" main.py
+sed -i "s|continue|break|g" main.py
+
+#python3 setup.py install
+#mkdir -p checkpoints
+#mkdir -p /root/.cache/torch/hub/checkpoints
+#cp $data_path/fcn_* /root/.cache/torch/hub/checkpoints
+
+for((RANK_ID=$RANK_ID_START;RANK_ID<$((RANK_SIZE+RANK_ID_START));RANK_ID++));
+do
+    #设置环境变量，不需要修改
+    echo "Device ID: $ASCEND_DEVICE_ID"
+    export RANK_ID=$RANK_ID
+    
+    
+    
+    #创建DeviceID输出目录，不需要修改
+    if [ -d ${cur_path}/output/${ASCEND_DEVICE_ID} ];then
+        rm -rf ${cur_path}/output/${ASCEND_DEVICE_ID}
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    else
+        mkdir -p ${cur_path}/output/$ASCEND_DEVICE_ID/ckpt
+    fi
+    
+    #绑核，不需要绑核的模型删除，需要绑核的模型根据实际修改
+    #cpucount=`lscpu | grep "CPU(s):" | head -n 1 | awk '{print $2}'`
+    #cpustep=`expr $cpucount / 8`
+    #echo "taskset c steps:" $cpustep
+    #let a=RANK_ID*$cpustep
+    #let b=RANK_ID+1
+    #let c=b*$cpustep-1
+
+    #执行训练脚本，以下传参不需要修改，其他需要模型审视修改
+    nohup python3 main.py  > ${cur_path}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log 2>&1 &
+done 
+wait
+
+#恢复参数
+sed -i "s|$data_path|./data|g" main.py
+sed -i "s|EPOCH_TIME = 1|EPOCH_TIME = 20|g" main.py
+sed -i "s|break|continue|g" main.py
+
+#conda deactivate
+#训练结束时间，不需要修改
+end_time=$(date +%s)
+e2e_time=$(( $end_time - $start_time ))
+
+#结果打印，不需要修改
+echo "------------------ Final result ------------------"
+#输出性能FPS，需要模型审视修改
+FPS=`grep FPS  $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F "FPS:" '{print $2}'|tail -n +2|awk '{sum+=$1} END {print"",sum/NR}'|sed s/[[:space:]]//g`
+#FPS=`awk 'BEGIN{printf "%.2f\n",'${batch_size}'*'${perf}'}'`
+
+
+#打印，不需要修改
+echo "Final Performance images/sec : $FPS"
+
+#输出训练精度,需要模型审视修改
+#train_accuracy=`grep eval_accuracy $cur_path/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|grep -v mlp_log|awk 'END {print $5}'| sed 's/,//g' |cut -c 1-5`
+#打印，不需要修改
+#echo "Final Train Accuracy : ${train_accuracy}"
+#echo "E2E Training Duration sec : $e2e_time"
+
+#稳定性精度看护结果汇总
+#训练用例信息，不需要修改
+BatchSize=${batch_size}
+DeviceType=`uname -m`
+CaseName=${Network}_bs${BatchSize}_${RANK_SIZE}'p'_'perf'
+
+##获取性能数据
+#吞吐量，不需要修改
+ActualFPS=${FPS}
+#单迭代训练时长，不需要修改
+TrainingTime=`awk 'BEGIN{printf "%.2f\n",'${BatchSize}'*1000/'${FPS}'}'`
+
+#从train_$ASCEND_DEVICE_ID.log提取Loss到train_${CaseName}_loss.txt中，需要根据模型审视
+grep "FPS" $cur_path/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log|awk -F "Loss :" '{print $2}'|awk -F "," '{print $1}' >> $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt
+
+#最后一个迭代loss值，不需要修改
+ActualLoss=`awk 'END {print}' $cur_path/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt`
+
+#关键信息打印到${CaseName}.log中，不需要修改
+echo "Network = ${Network}" > $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "RankSize = ${RANK_SIZE}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "BatchSize = ${BatchSize}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "DeviceType = ${DeviceType}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "CaseName = ${CaseName}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualFPS = ${ActualFPS}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "TrainingTime = ${TrainingTime}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+#echo "TrainAccuracy = ${train_accuracy}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "ActualLoss = ${ActualLoss}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log
+echo "E2ETrainingTime = ${e2e_time}" >> $cur_path/output/$ASCEND_DEVICE_ID/${CaseName}.log