update docs (#186)

.github/ISSUE_TEMPLATE/bug_report.md

@@ -20,7 +20,7 @@ Steps to reproduce the behavior:
 **Operating environment(运行环境):**
  - python version [e.g. 3.5, 3.6]
  - torch version [e.g. 1.6.0, 1.7.0]
- - deepctr-torch version [e.g. 0.2.4,]
+ - deepctr-torch version [e.g. 0.2.7,]
 
 **Additional context**
 Add any other context about the problem here.

.github/ISSUE_TEMPLATE/question.md

@@ -17,4 +17,4 @@ Add any other context about the problem here.
 **Operating environment(运行环境):**
  - python version [e.g. 3.6]
  - torch version [e.g. 1.7.0,]
- - deepctr-torch version [e.g. 0.2.4,]
+ - deepctr-torch version [e.g. 0.2.7,]

README.md

@@ -9,7 +9,7 @@
 
 [![Documentation Status](https://readthedocs.org/projects/deepctr-torch/badge/?version=latest)](https://deepctr-torch.readthedocs.io/)
 ![CI status](https://github.com/shenweichen/deepctr-torch/workflows/CI/badge.svg)
-[![codecov](https://codecov.io/gh/shenweichen/DeepCTR-Torch/branch/master/graph/badge.svg)](https://codecov.io/gh/shenweichen/DeepCTR-Torch)
+[![codecov](https://codecov.io/gh/shenweichen/DeepCTR-Torch/branch/master/graph/badge.svg?token=m6v89eYOjp)](https://codecov.io/gh/shenweichen/DeepCTR-Torch)
 [![Disscussion](https://img.shields.io/badge/chat-wechat-brightgreen?style=flat)](./README.md#disscussiongroup)
 [![License](https://img.shields.io/github/license/shenweichen/deepctr-torch.svg)](https://github.com/shenweichen/deepctr-torch/blob/master/LICENSE)
 
@@ -41,6 +41,8 @@ Let's [**Get Started!**](https://deepctr-torch.readthedocs.io/en/latest/Quick-St
 |                IFM                 | [IJCAI 2019][An Input-aware Factorization Machine for Sparse Prediction](https://www.ijcai.org/Proceedings/2019/0203.pdf)   |
 |                DCN V2                    | [arxiv 2020][DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems](https://arxiv.org/abs/2008.13535)   |
 |                DIFM                 | [IJCAI 2020][A Dual Input-aware Factorization Machine for CTR Prediction](https://www.ijcai.org/Proceedings/2020/0434.pdf)   |
+|                AFN                 | [AAAI 2020][Adaptive Factorization Network: Learning Adaptive-Order Feature Interactions](https://arxiv.org/pdf/1909.03276)   |
+
 
 
 ## DisscussionGroup & Related Projects
@@ -49,7 +51,7 @@ Let's [**Get Started!**](https://deepctr-torch.readthedocs.io/en/latest/Quick-St
     <table style="margin-left: 20px; margin-right: auto;">
         <tr>
             <td>
-                公众号：<b>浅梦的学习笔记</b><br><br>
+                公众号：<b>浅梦学习笔记</b><br><br>
                 <a href="https://github.com/shenweichen/deepctr-torch">
   <img align="center" src="./docs/pics/code.png" />
 </a>
@@ -74,7 +76,7 @@ Let's [**Get Started!**](https://deepctr-torch.readthedocs.io/en/latest/Quick-St
 
 
 
-## Contributors([welcome to join us!](./CONTRIBUTING.md))
+## Main Contributors([welcome to join us!](./CONTRIBUTING.md))
 
 <table border="0">
   <tbody>
@@ -125,18 +127,18 @@ Let's [**Get Started!**](https://deepctr-torch.readthedocs.io/en/latest/Quick-St
         <p>Dev<br>
         NetEase <br>  <br>  </p>​
       </td>
+      <td>
+        ​ <a href="https://github.com/WeiyuCheng"><img width="70" height="70" src="https://github.com/WeiyuCheng.png?s=40" alt="pic"></a><br>
+        ​ <a href="https://github.com/WeiyuCheng">Cheng Weiyu</a> ​
+        <p>Dev<br>
+        Shanghai Jiao Tong University</p>​
+      </td>
       <td>
         ​ <a href="https://github.com/tangaqi"><img width="70" height="70" src="https://github.com/tangaqi.png?s=40" alt="pic"></a><br>
         ​ <a href="https://github.com/tangaqi">Tang</a>
         <p>Test<br>
         Tongji University <br> <br>  </p>​
       </td>
-      <td>
-        ​ <a href="https://github.com/uestc7d"><img width="70" height="70" src="https://github.com/uestc7d.png?s=40" alt="pic"></a><br>
-        ​ <a href="https://github.com/uestc7d">Xu Qidi</a> ​
-        <p>Dev<br>
-        University of <br> Electronic  Science  and <br> Technology of China</p>​
-      </td>
     </tr>
   </tbody>
 </table>

deepctr_torch/__init__.py

@@ -2,5 +2,5 @@
 from . import models
 from .utils import check_version
 
-__version__ = '0.2.6'
+__version__ = '0.2.7'
 check_version(__version__)

deepctr_torch/layers/activation.py

@@ -17,6 +17,7 @@ class Dice(nn.Module):
         - [Zhou G, Zhu X, Song C, et al. Deep interest network for click-through rate prediction[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. ACM, 2018: 1059-1068.](https://arxiv.org/pdf/1706.06978.pdf)
         - https://github.com/zhougr1993/DeepInterestNetwork, https://github.com/fanoping/DIN-pytorch
     """
+
     def __init__(self, emb_size, dim=2, epsilon=1e-8, device='cpu'):
         super(Dice, self).__init__()
         assert dim == 2 or dim == 3
@@ -41,18 +42,16 @@ def forward(self, x):
             x_p = self.sigmoid(self.bn(x))
             out = self.alpha * (1 - x_p) * x + x_p * x
             out = torch.transpose(out, 1, 2)
-
         return out
 
 
 class Identity(nn.Module):
 
-
     def __init__(self, **kwargs):
         super(Identity, self).__init__()
 
-    def forward(self, X):
-        return X
+    def forward(self, inputs):
+        return inputs
 
 
 def activation_layer(act_name, hidden_size=None, dice_dim=2):

deepctr_torch/layers/interaction.py

@@ -130,7 +130,7 @@ def __init__(self, filed_size, embedding_size, bilinear_type="interaction", seed
                 self.bilinear.append(
                     nn.Linear(embedding_size, embedding_size, bias=False))
         elif self.bilinear_type == "interaction":
-            for i, j in itertools.combinations(range(filed_size), 2):
+            for _, _ in itertools.combinations(range(filed_size), 2):
                 self.bilinear.append(
                     nn.Linear(embedding_size, embedding_size, bias=False))
         else:
@@ -487,9 +487,9 @@ def __init__(self, in_features, low_rank=32, num_experts=4, layer_num=2, device=
         self.bias = nn.Parameter(torch.Tensor(self.layer_num, in_features, 1))
 
         init_para_list = [self.U_list, self.V_list, self.C_list]
-        for i in range(len(init_para_list)):
-            for j in range(self.layer_num):
-                nn.init.xavier_normal_(init_para_list[i][j])
+        for para in init_para_list:
+            for i in range(self.layer_num):
+                nn.init.xavier_normal_(para[i])
 
         for i in range(len(self.bias)):
             nn.init.zeros_(self.bias[i])

deepctr_torch/layers/sequence.py

@@ -117,7 +117,7 @@ def forward(self, query, keys, keys_length, mask=None):
         Output shape
           - 3D tensor with shape: ``(batch_size, 1, embedding_size)``.
         """
-        batch_size, max_length, dim = keys.size()
+        batch_size, max_length, _ = keys.size()
 
         # Mask
         if self.supports_masking:
@@ -176,16 +176,16 @@ def __init__(self, k, axis, device='cpu'):
         self.axis = axis
         self.to(device)
 
-    def forward(self, input):
-        if self.axis < 0 or self.axis >= len(input.shape):
+    def forward(self, inputs):
+        if self.axis < 0 or self.axis >= len(inputs.shape):
             raise ValueError("axis must be 0~%d,now is %d" %
-                             (len(input.shape) - 1, self.axis))
+                             (len(inputs.shape) - 1, self.axis))
 
-        if self.k < 1 or self.k > input.shape[self.axis]:
+        if self.k < 1 or self.k > inputs.shape[self.axis]:
             raise ValueError("k must be in 1 ~ %d,now k is %d" %
-                             (input.shape[self.axis], self.k))
+                             (inputs.shape[self.axis], self.k))
 
-        out = torch.topk(input, k=self.k, dim=self.axis, sorted=True)[0]
+        out = torch.topk(inputs, k=self.k, dim=self.axis, sorted=True)[0]
         return out
 
 
@@ -220,11 +220,11 @@ def __init__(self, input_size, hidden_size, bias=True):
             self.register_parameter('bias_ih', None)
             self.register_parameter('bias_hh', None)
 
-    def forward(self, input, hx, att_score):
-        gi = F.linear(input, self.weight_ih, self.bias_ih)
+    def forward(self, inputs, hx, att_score):
+        gi = F.linear(inputs, self.weight_ih, self.bias_ih)
         gh = F.linear(hx, self.weight_hh, self.bias_hh)
-        i_r, i_z, i_n = gi.chunk(3, 1)
-        h_r, h_z, h_n = gh.chunk(3, 1)
+        i_r, _, i_n = gi.chunk(3, 1)
+        h_r, _, h_n = gh.chunk(3, 1)
 
         reset_gate = torch.sigmoid(i_r + h_r)
         # update_gate = torch.sigmoid(i_z + h_z)
@@ -266,8 +266,8 @@ def __init__(self, input_size, hidden_size, bias=True):
             self.register_parameter('bias_ih', None)
             self.register_parameter('bias_hh', None)
 
-    def forward(self, input, hx, att_score):
-        gi = F.linear(input, self.weight_ih, self.bias_ih)
+    def forward(self, inputs, hx, att_score):
+        gi = F.linear(inputs, self.weight_ih, self.bias_ih)
         gh = F.linear(hx, self.weight_hh, self.bias_hh)
         i_r, i_z, i_n = gi.chunk(3, 1)
         h_r, h_z, h_n = gh.chunk(3, 1)
@@ -293,25 +293,25 @@ def __init__(self, input_size, hidden_size, bias=True, gru_type='AGRU'):
         elif gru_type == 'AUGRU':
             self.rnn = AUGRUCell(input_size, hidden_size, bias)
 
-    def forward(self, input, att_scores=None, hx=None):
-        if not isinstance(input, PackedSequence) or not isinstance(att_scores, PackedSequence):
+    def forward(self, inputs, att_scores=None, hx=None):
+        if not isinstance(inputs, PackedSequence) or not isinstance(att_scores, PackedSequence):
             raise NotImplementedError("DynamicGRU only supports packed input and att_scores")
 
-        input, batch_sizes, sorted_indices, unsorted_indices = input
+        inputs, batch_sizes, sorted_indices, unsorted_indices = inputs
         att_scores, _, _, _ = att_scores
 
         max_batch_size = int(batch_sizes[0])
         if hx is None:
             hx = torch.zeros(max_batch_size, self.hidden_size,
-                             dtype=input.dtype, device=input.device)
+                             dtype=inputs.dtype, device=inputs.device)
 
-        outputs = torch.zeros(input.size(0), self.hidden_size,
-                              dtype=input.dtype, device=input.device)
+        outputs = torch.zeros(inputs.size(0), self.hidden_size,
+                              dtype=inputs.dtype, device=inputs.device)
 
         begin = 0
         for batch in batch_sizes:
             new_hx = self.rnn(
-                input[begin:begin + batch],
+                inputs[begin:begin + batch],
                 hx[0:batch],
                 att_scores[begin:begin + batch])
             outputs[begin:begin + batch] = new_hx

deepctr_torch/models/afn.py

@@ -2,6 +2,7 @@
 """
 Author:
     Weiyu Cheng, [email protected]
+
 Reference:
     [1] Cheng, W., Shen, Y. and Huang, L. 2020. Adaptive Factorization Network: Learning Adaptive-Order Feature
          Interactions. Proceedings of the AAAI Conference on Artificial Intelligence. 34, 04 (Apr. 2020), 3609-3616.
@@ -14,7 +15,7 @@
 
 
 class AFN(BaseModel):
-    """Instantiates the Adaptive Factorization Network architecture. 
+    """Instantiates the Adaptive Factorization Network architecture.
     
     In DeepCTR-Torch, we only provide the non-ensembled version of AFN for the consistency of model interfaces. For the ensembled version of AFN+, please refer to https://github.com/WeiyuCheng/DeepCTR-Torch (Pytorch Version) or https://github.com/WeiyuCheng/AFN-AAAI-20 (Tensorflow Version).
 
@@ -38,7 +39,7 @@ class AFN(BaseModel):
 
     def __init__(self,
                  linear_feature_columns, dnn_feature_columns,
-                 ltl_hidden_size=600, afn_dnn_hidden_units=(400, 400, 400),
+                 ltl_hidden_size=256, afn_dnn_hidden_units=(256, 128),
                  l2_reg_linear=0.00001, l2_reg_embedding=0.00001, l2_reg_dnn=0,
                  init_std=0.0001, seed=1024, dnn_dropout=0, dnn_activation='relu',
                  task='binary', device='cpu', gpus=None):

deepctr_torch/models/dien.py

@@ -319,7 +319,7 @@ def __init__(self,
     @staticmethod
     def _get_last_state(states, keys_length):
         # states [B, T, H]
-        batch_size, max_seq_length, hidden_size = states.size()
+        batch_size, max_seq_length, _ = states.size()
 
         mask = (torch.arange(max_seq_length, device=keys_length.device).repeat(
             batch_size, 1) == (keys_length.view(-1, 1) - 1))

deepctr_torch/models/ifm.py

@@ -66,7 +66,7 @@ def __init__(self,
         self.to(device)
 
     def forward(self, X):
-        sparse_embedding_list, dense_value_list = self.input_from_feature_columns(X, self.dnn_feature_columns,
+        sparse_embedding_list, _ = self.input_from_feature_columns(X, self.dnn_feature_columns,
                                                                                   self.embedding_dict)
         if not len(sparse_embedding_list) > 0:
             raise ValueError("there are no sparse features")

docs/source/Features.md

@@ -261,6 +261,14 @@ Dual Inputaware Factorization Machines (DIFM) can adaptively reweight the origin
 
 [Lu W, Yu Y, Chang Y, et al. A Dual Input-aware Factorization Machine for CTR Prediction[C]//IJCAI. 2020: 3139-3145.](https://www.ijcai.org/Proceedings/2020/0434.pdf)
 
+### AFN(Adaptive Factorization Network: Learning Adaptive-Order Feature Interactions)
+
+Adaptive Factorization Network (AFN) can learn arbitrary-order cross features adaptively from data. The core of AFN is a logarith- mic transformation layer to convert the power of each feature in a feature combination into the coefficient to be learned.
+[**AFN Model API**](./deepctr_torch.models.afn.html)  
+
+![AFN](../pics/AFN.jpg)
+
+[Cheng, W., Shen, Y. and Huang, L. 2020. Adaptive Factorization Network: Learning Adaptive-Order Feature Interactions. Proceedings of the AAAI Conference on Artificial Intelligence. 34, 04 (Apr. 2020), 3609-3616.](https://arxiv.org/pdf/1909.03276)
 
 
 ## Layers

docs/source/History.md

@@ -1,5 +1,6 @@
 # History
-- 04/04/2021 : [v0.2.6](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.6) released.Add add [IFM](./Features.html#ifm-input-aware-factorization-machine) and [DIFM](./Features.html#difm-dual-input-aware-factorization-machine);Support multi-gpus running([example](./FAQ.html#how-to-run-the-demo-with-multiple-gpus)).
+- 06/14/2021 : [v0.2.7](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.6) released.Add [AFN](./Features.html#afn-adaptive-factorization-network-learning-adaptive-order-feature-interactions) and fix some bugs.
+- 04/04/2021 : [v0.2.6](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.6) released.Add [IFM](./Features.html#ifm-input-aware-factorization-machine) and [DIFM](./Features.html#difm-dual-input-aware-factorization-machine);Support multi-gpus running([example](./FAQ.html#how-to-run-the-demo-with-multiple-gpus)).
 - 02/12/2021 : [v0.2.5](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.5) released.Fix bug in DCN-M.
 - 12/05/2020 : [v0.2.4](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.4) released.Imporve compatibility & fix issues.Add History callback.([example](https://deepctr-torch.readthedocs.io/en/latest/FAQ.html#set-learning-rate-and-use-earlystopping)).
 - 10/18/2020 : [v0.2.3](https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.3) released.Add [DCN-M](./Features.html#dcn-deep-cross-network)&[DCN-Mix](./Features.html#dcn-mix-improved-deep-cross-network-with-mix-of-experts-and-matrix-kernel).Add EarlyStopping and ModelCheckpoint callbacks([example](https://deepctr-torch.readthedocs.io/en/latest/FAQ.html#set-learning-rate-and-use-earlystopping)).

docs/source/conf.py

@@ -26,7 +26,7 @@
 # The short X.Y version
 version = ''
 # The full version, including alpha/beta/rc tags
-release = '0.2.6'
+release = '0.2.7'
 
 
 # -- General configuration ---------------------------------------------------

docs/source/deepctr_torch.models.afn.rst

@@ -0,0 +1,7 @@
+deepctr\_torch.models.afn module
+================================
+
+.. automodule:: deepctr_torch.models.afn
+    :members:
+    :no-undoc-members:
+    :no-show-inheritance:

docs/source/deepctr_torch.models.rst

@@ -23,6 +23,7 @@ Submodules
    deepctr_torch.models.dien
    deepctr_torch.models.ifm
    deepctr_torch.models.difm
+   deepctr_torch.models.afn
 
 Module contents
 ---------------

docs/source/index.rst

@@ -34,16 +34,16 @@ You can read the latest code at https://github.com/shenweichen/DeepCTR-Torch and
 
 News
 -----
+06/14/2021 : Add `AFN <./Features.html#afn-adaptive-factorization-network-learning-adaptive-order-feature-interactions>`_ and fix some bugs.  `Changelog <https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.7>`_
+
 04/04/2021 : Add `IFM <./Features.html#ifm-input-aware-factorization-machine>`_ and `DIFM <./Features.html#difm-dual-input-aware-factorization-machine>`_ . Support multi-gpus running(`example <./FAQ.html#how-to-run-the-demo-with-multiple-gpus>`_). `Changelog <https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.6>`_
 
 02/12/2021 : Fix bug in DCN-M. `Changelog <https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.4>`_
 
-12/05/2020 : Imporve compatibility & fix issues.Add History callback(`example <https://deepctr-torch.readthedocs.io/en/latest/FAQ.html#set-learning-rate-and-use-earlystopping>`_). `Changelog <https://github.com/shenweichen/DeepCTR-Torch/releases/tag/v0.2.3>`_
-
 DisscussionGroup
 -----------------------
 
-公众号：**浅梦的学习笔记**  wechat ID: **deepctrbot**
+公众号：**浅梦学习笔记**  wechat ID: **deepctrbot**
 
 .. image:: ../pics/code.png
 

setup.py

@@ -9,7 +9,7 @@
 
 setuptools.setup(
     name="deepctr-torch",
-    version="0.2.6",
+    version="0.2.7",
     author="Weichen Shen",
     author_email="[email protected]",
     description="Easy-to-use,Modular and Extendible package of deep learning based CTR(Click Through Rate) prediction models with PyTorch",

tests/models/AFN_test.py

@@ -1,18 +1,16 @@
 # -*- coding: utf-8 -*-
 import pytest
 
-from deepctr_torch.callbacks import EarlyStopping, ModelCheckpoint
 from deepctr_torch.models import AFN
 from tests.utils import get_test_data, SAMPLE_SIZE, check_model, get_device
 
 
 @pytest.mark.parametrize(
     'afn_dnn_hidden_units, sparse_feature_num, dense_feature_num',
-    [((256, 128), 3, 0), 
-    ((256, 128), 3, 3),
-     ((256, 128), 0,3)]
+    [((256, 128), 3, 0),
+     ((256, 128), 3, 3),
+     ((256, 128), 0, 3)]
 )
-
 def test_AFN(afn_dnn_hidden_units, sparse_feature_num, dense_feature_num):
     model_name = 'AFN'
     sample_size = SAMPLE_SIZE
@@ -23,5 +21,6 @@ def test_AFN(afn_dnn_hidden_units, sparse_feature_num, dense_feature_num):
 
     check_model(model, model_name, x, y)
 
+
 if __name__ == '__main__':
     pass