Question about NLP models/conversation bot.
Currently my OSRS bot uses an external API call for its chatbot and some hand coded phrases.
This is really low cpu intensity since the whole processes is reduced on my end to a single API call.
I want to move it client side, so I will write my own bespoke (probably deep learned) and continually updated chatbot.
I will likely ripoff someone elses code skeleton and hopefully something thats a pretrained base.
My question is: how CPU intense is it to call to the model to get responses once itrs mostly trained, (is this completely trivial and same cpu demand as calling an external api over the internet)
and is updating in batches of convos cpu intense assuming its 99% trained already...?
Currently my OSRS bot uses an external API call for its chatbot and some hand coded phrases.
This is really low cpu intensity since the whole processes is reduced on my end to a single API call.
I want to move it client side, so I will write my own bespoke (probably deep learned) and continually updated chatbot.
I will likely ripoff someone elses code skeleton and hopefully something thats a pretrained base.
My question is: how CPU intense is it to call to the model to get responses once itrs mostly trained, (is this completely trivial and same cpu demand as calling an external api over the internet)
and is updating in batches of convos cpu intense assuming its 99% trained already...?
