Command for evaluating NeuralFM model

[rezacsedu]

Hi there,

Could you please provide the command for evaluating NeuralFM model? -i.e. I was expecting something similar presented at https://github.com/hexiangnan/attentional_factorization_machine.

Best,

[hexiangnan]

When you run the codes, it evaluates the model after each training epoch. See: https://github.com/hexiangnan/neural_factorization_machine/blob/master/NeuralFM.py#L285

…

On Sat, Sep 16, 2017 at 10:01 AM, Md. Rezaul Karim ***@***.*** > wrote: Hi there, Could you please provide the command for evaluating NeuralFM model? -i.e. I was expecting there will something similar presented at https://github.com/hexiangnan/attentional_factorization_machine. Best, — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABGxjqbZNygTPEeBksSJZOhaz0aH0zpfks5siywKgaJpZM4PZq4X> .

-- Best Regards, Xiangnan He

[rezacsedu]

@hexiangnan, thanks so much.
I have seperated the test set from train/validation and did the evalaution seperately:

print("RMSE: ")
print(model.evaluate(data.Test_data)) # evaluate the results for an input set

BTW, could you please upload the script you used to convert the raw MovieLens (or frappe) dataset into LibFM format?

[hexiangnan]

Sorry I can not find the script. Should be easy to create one. Check the libFM manual Section 2.1.1: http://www.libfm.org/libfm-1.42.manual.pdf

…

On Sat, Sep 16, 2017 at 9:53 PM, Md. Rezaul Karim ***@***.***> wrote: @hexiangnan <https://github.com/hexiangnan>, thanks so much. I have seperated the test set from train/validation and did the evalaution seperately: Model evaluation print("RMSE: ") print(model.evaluate(data.Test_data)) # evaluate the results for an input set BTW, could you please upload the script you used to convert the raw MovieLens (or frappe) dataset into LibFM format? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABGxjiW38lDYIsyuWzKSPF6x-IKQgaRWks5si9LggaJpZM4PZq4X> .

-- Best Regards, Xiangnan He