zhougr1993/DeepInterestNetwork

Issues

• TypeError: iter() returned non-iterator of type 'DataInputTest'

  #87 opened 3 years ago by ucasiggcas
  1

• 在attention里面，vec1*vec2的背后逻辑是什么？

  #54 opened 5 years ago by lijian10086
  12

• Can not reproduce the DIN result in README.md

  #92 opened 2 years ago by liyangliu
  0

• 我实现的mini-batch aware regularization，求拍砖

  #82 opened 4 years ago by bound2020
  1

• 关于DIN复现的结果问题

  #57 opened 5 years ago by hit-joseph
  2

• DIN模型中self.j

  #78 opened 4 years ago by xiaopp123
  3

• 对代码与论文模型图中的几个疑问

  #90 opened 3 years ago by LXXiaogege
  1

• Attention 中为什么用 softmax 把权重归一？ 和原论文的表述不一致

  #91 opened 2 years ago by wenmin-wu
  1

• 模拟随机数据 auc 特别高0.877

  #66 opened 5 years ago by LHFCOD
  3

• train速度太慢，是否有多卡运行版本

  #89 opened 3 years ago by lljjgg
  0

• wide_deep 中 wide 部分的实现

  #88 opened 3 years ago by mikudehuane
  0

• cat aa ab ac 与python运行脚本生成的dataset.pkl不相同？大小差了两倍左右？！

  #62 opened 5 years ago by Ming-Qin-tech
  4

• DIN model中为何没有拼接user vector

  #64 opened 5 years ago by zhiweishen
  4

• 感觉input.py那里hist_i用0补全是有问题的

  #85 opened 4 years ago by miaotianmiaodi
  0

• 请问代码当中的self attention与self all attention 与shine是什么意思？可以给我看看相关文章吗？

  #86 opened 4 years ago by GuoshenLi
  0

• 关于代码中的bn实现部分的疑问

  #83 opened 4 years ago
  2

• 关于可以运行的Electronics数据集

  #84 opened 4 years ago by BelindaYanyiwuhua
  0

• 输入问题

  #81 opened 4 years ago by huozijie1990
  0

• why use sigmoid in Attention

  #80 opened 4 years ago by huaweiboy
  1

• attention_multi_items 的问题？

  #79 opened 4 years ago by Peterisfar
  0

• 关于负采样的问题

  #73 opened 5 years ago by Yanglazi
  3

• max_sl 变量（用户点击序列的长度）在每个 batch 里都是不一样的值，模型结构在每个 batch 里是不固定的？

  #69 opened 5 years ago by lcshr123
  3

• DIN模型处理dense feature的相关问题

  #47 opened 5 years ago by eckolemon
  1

• din模型实际业务应用

  #44 opened 5 years ago by andrew-zzz
  4

• 该模型是针对一个广告做训练吗？Is the model training for an AD？

  #58 opened 5 years ago by manji460
  1

• movielen数据集如何预处理？

  #75 opened 4 years ago by Qian-Hao
  1

• 代码存在严重bug: label泄露

  #76 opened 4 years ago by starspringcloud
  2

• 不定长的hist如果不进行padding是否更合适

  #77 opened 4 years ago by DSXiangLi
  2

• DIN的输入只包括商品、类目以及历史行为商品吗？

  #48 opened 5 years ago by waterbeach
  1

• 原论文在介绍attention时候，叙述说放弃了softmax的归一化，但在代码中依旧保留了softmax，请问这是怎么回事？

  #74 opened 4 years ago by lcwy220
  2

• Low AUC after fixed BN bugs

  #63 opened 4 years ago by liuhu-bigeye
  1

• Different AUC implementation with the paper

  #71 opened 4 years ago by liuhu-bigeye
  3

• Mini-batch Aware Regularization正则化在代码哪里体现了呀。各位大佬

  #60 opened 4 years ago by cuiweizuishuai
  1

• 代码的命名格式太差

  #61 opened 4 years ago by wind-meta
  2

• Questions about Wide and deep model

  #45 opened 4 years ago by czisok
  1

• 关于attention的实现

  #50 opened 4 years ago by ZhengYukai666
  1

• 代码写的太差了，p8大佬

  #67 opened 4 years ago by weiyu322
  1

• 有关movielens数据集的问题

  #68 opened 5 years ago by yuanzhiyu
  1

• Why Scale

  #72 opened 5 years ago by LittleMisss
  0

• DIN使用的都是现有（假设100个）itemID训练，如果出现新的（第101个）itemID,这个模型是否就不能用了？需要重新训练？

  #55 opened 5 years ago by lijian10086
  2

• 请问必须显存>=10G才能跑吗~

  #46 opened 5 years ago by Slayerxxx
  2

• Not reusing batch_norm between hist_i and hist_j may

  #65 opened 5 years ago by liuhu-bigeye
  0

• 我save model后model特别大可能是什么问题

  #59 opened 5 years ago by cuiweizuishuai
  0

• DIN的训练结果

  #52 opened 5 years ago by Rosehan
  2

• DataInput类hist的疑问

  #56 opened 5 years ago by Slayerxxx
  0

• 使用其他模型（如LR）作为对比试验时，是否也需要输入历史ID信息

  #43 opened 5 years ago by tinkle1129
  2

• DeepFM代码中的一个问题

  #53 opened 5 years ago by zhangbei121
  0

• 关于实际问题中的序列构造问题

  #49 opened 5 years ago by caojiajia
  0

• 为什么GAUC的值一直为0？请问self.mf_auc = tf.reduce_mean(tf.to_float(x > 0))这一句代码的含义是？

  #51 opened 5 years ago by eckolemon
  0

• 关于GAUC的计算

  #42 opened 5 years ago by patricxu
  2