tensorflow-DeepFM icon indicating copy to clipboard operation
tensorflow-DeepFM copied to clipboard

Tensorflow implementation of DeepFM for CTR prediction.

Results 47 tensorflow-DeepFM issues
Sort by recently updated
recently updated
newest added
trafficstars

老哥,我对你代码的理解,first_order是F维,second_order是K维,Deep层输出维度。这三个维度concat之后再加权输出? 这个跟论文好像不一样呀……

can't import this module which version of tf conclude this module

请问 # "ps_calc_01", "ps_calc_02", "ps_calc_03", "ps_calc_04", # "ps_calc_05", "ps_calc_06", "ps_calc_07", "ps_calc_08", # "ps_calc_09", "ps_calc_10", "ps_calc_11", "ps_calc_12", # "ps_calc_13", "ps_calc_14", # "ps_calc_15_bin", "ps_calc_16_bin", "ps_calc_17_bin", # "ps_calc_18_bin", "ps_calc_19_bin", "ps_calc_20_bin" ] 这里的数据为什么要忽略呢,是不是加上就过拟合了??因为这都是向量特征,加上不是更合适吗?

我用如下代码导出DeepFM模型,但是发现freeze_model_dir/variables为空,请问应该如何导出DeepFM的pb模型? `def freeze_model(self): freeze_model_dir = "freeze_model_dir" save_dir = 'checkpoints/' save_path = os.path.join(save_dir, 'best_validation') start_time = `time()` print(tf.trainable_variables()) print("freeze model...") SIGNATURE_NAME = "serving_default" builder = tf.saved_model.builder.SavedModelBuilder(freeze_model_dir) inputs = {'feat_index': tf.saved_model.utils.build_tensor_info(self.feat_index), 'feat_value':...

https://github.com/ChenglongChen/tensorflow-DeepFM/blob/a43dd5ff1f61a275c8a6e0fd659e200af64093cc/DeepFM.py#L89

我也有此疑问, 按论文中的意思,y_FM= reduce_sum(first_order,1) + reduce_sum(second_order,1) y_DNN = reduce_sum(y_deep,1),这个和 concat([first_order, second_order, y_deep]) X weights["concat_projection"])是不等价的吧,毕竟weights["concat_projection"]是不全为1的向量(变量),而且只有wx和DNN最后一层需要乘,second_order的xixj项不需要乘weight 不知道是不是我理解不对? concat之后再输出,在计算结果上,和论文中 sigmoid(y_FM+y_DNN) 单独计算再加和是一样的。 我觉得first order乘以feature_bias是多余的。因为embedding的结果与deep、second order拼接最后接一个projection layer,只看feat_value-projection这一块就已经是等价LR 的形式(论文中的公式2),在前面乘以一个feature_bias又不加非线性激活函数完全没必要。 PS:在gayhub上讨论,是不是还是用英语更合适? _Originally posted by @futureer in https://github.com/ChenglongChen/tensorflow-DeepFM/issues/32#issuecomment-451852547_

If do so,what`s the meaning of categorical data?

hi, 你好,非常喜欢DeepFM的想法,但是有一个疑问就是,FM的embedding size通常设置的不大比如4或者8这个量级,但是在深度学习中一般设置的会相对大,比如32、64,128。DeepFM中这两者为同一个的话,会不会说某种程度抑制了深度部分的学习能力?请问有做过相关的实验吗?

请问各位 : TypeError: Expected string passed to parameter 'tensor_names' of op 'SaveV2', got ['Variable', 'Variable/Adam', 'Variable/Adam_1', 'Variable_1', 'Variable_1/Adam', 'Variable_1/Adam_1', 'Variable_2', 'Variable_2/Adam', 'Variab 这个错误是为什么? 完整的错误信息是这样: Traceback (most recent call last): File...