How to add a long feature vector as a feature to the model? about deepctr HOT 6 CLOSED

@Cong-Zou print the shape of train['video_feature'].values and test['video_feature'].values,make sure their shape are(#records,128) .

from deepctr.

@Cong-Zou please refer to

https://github.com/shenweichen/Bytedance_ICME2019_challenge_baseline/blob/master/model.py

line 20,21 define a input for continue 128-dimension feature

you can perform the desired operation on video_input tensor

when fit the model,the input to the model is [original input list] + [video_input_array]

from deepctr.

I have read about the code..
But sorry, I'm a little confused about it.
After I add the two sentences (delete the annotations) in model.py as u suggest,
in train.py, as shown below:

do I just need to extend the train and test model input lists with the video feature arrays?

from deepctr.

yes~

from deepctr.

Hi Weichen,
I have tried like this..

The train['video_feature'].values is a ndarray(D=#records) and every entry in it is also an array(D=128).
However, I run the code and get the error as shown below:

May I ask the reason about it?
Thanks a lot!

from deepctr.

Great! It works now..
Thank you again!

from deepctr.