Deployment of Chatbots | Learn AI Together | Page 1

calm bluff May 23, 2023, 5:49 PM

#

Hey guys I'm pretty new to ML. I have been trying to deploy GPT-J as a chatbot for months now but all of the websites I've found and services (Google Cloud, AWS, HuggingFace, "easy" API services, etc..) are either restrictive, unreliable, very confusing, contain very difficult to understand fees (ahem ahem AWS ahem), or are just plain expensive. I've decided on GPT-J since it's open-source and seems the best for human-like interactions but unfortunately this has led me down a path of the issues mentioned above.

What cheap way could there possibly be to deploy a model and have API access, especially for prototypes? Do you guys have any recommendations?
There doesn't seem to be much information online.

mighty storm May 25, 2023, 3:14 AM

#

I prefer to deploy a model by own linux server, but you need some coding skills and a server with a IP adress, if you are poor at coding , you can try https://ora.ai/

1-click chatbot | ora.ai

Create your own chatbot with just a description and a click

calm bluff May 25, 2023, 2:23 PM

#

mighty storm I prefer to deploy a model by own linux server, but you need some coding skills...

Thank you for the recommendation. Out of curiousity, what service do you use for your linux server hosting?

mighty storm May 25, 2023, 4:27 PM

#

I am not using any third-party service I just deploy the model with pytorch or some deep learning framework and then provide web api with flask

#Deployment of Chatbots