jungyhuk / latent-attention Goto Github PK

Hello.
What is the format of the pre-processed dataset of IFTTT ?
On unpickling the msr_data.pkl file, the data in it is not making much sense to me... because the attributes like correct_action_param, correct_trigger_param, label_names, words, url etc. are occurring together again and again, as a bundle, and after them is another bundle of some other attributes , namely label_types, trigger_chans, action_chans, action_funcs etc and then is a bundle of action and trigger fields, which is followed by some fields which I don't understand but I think are the fields which require user input and are initialized as NULL and then... I don't understand.

Can you tell me how to make sense of the data ? Also, it would be pretty helpful if you can tell me how to see the data in a pretty-formatted format.

Where is the LabelEncoder() that can be used to map prediction numbers to real instructions.
E.G.
'label_names': ['Instagram',
'Instagram.Any_new_photo_by_you',
'Dropbox',
'Dropbox.Add_file_from_URL'],
'labels': [45, 193, 11, 20],

Where can I find the map between label_names and labels ?

Thanks a lot. I am trying to use your code to play with prediction.