nicolas-ivanov
commented
5 years ago So, do I understand correctly that you are actually not detecting any emotion from the user text input but rather hardwire the emotion in the response to the emotion input category, no matter what the user's emotion in the input text is?
Yes, that's correct.
Looks like you are multiplying the emotion condition (from the input) with the condition ids that you gathered from the tokenized user text. What do these condition ids actually relate to?
I see that this piece of code is written poorly, since the condition_ids_num can be simply replaced with 1. Thanks for pointing this out, will be fixed in future release to the following:
condition_ids = transform_conditions_to_ids([emotion], _cakechat_model.condition_to_index, n_dialogs=1)Explanation of the current code: condition_ids_num equals to the number of contexts to be processed in the same batch, i.e. it is a batch_size. Since the api function get_response processes only one dialog context at a time, condition_ids_num always equals to 1. If get_response function had to process a batch of contexts with the same emotion used for conditioning, the current code would make sense.
tiimoS
From what I can make out in the code (get_response in cakechat.api.response), you are using the input emotion category (that the user can set - {joy, anger, sadness etc.}) to condition the response. So, do I understand correctly that you are actually not detecting any emotion from the user text input but rather hardwire the emotion in the response to the emotion input category, no matter what the user's emotion in the input text is?
Looks like you are multiplying the emotion condition (from the input) with the condition ids that you gathered from the tokenized user text. What do these condition ids actually relate to?
Thanks in advance for the clarifications 👍