The most horrible large language model on the internet: GPT4chan
https://playground-beta.gpt4chan.com/
It was trained using 3 years worth of /pol/ posts.
https://github.com/yk/gpt-4chan
https://zenodo.org/record/3606810
you can call the api directly at https://playground-beta.gpt4chan.com/api/complete
example POST payload:
{"prompt":"-----\n--- 231316735\n","length":128,"top_p":0.9,"temperature":0.8}
It is running off of a single TPU, and If there are too many requests in the queue, the request will time out and return no results or error message to the user.
https://playground-beta.gpt4chan.com/
It was trained using 3 years worth of /pol/ posts.
https://github.com/yk/gpt-4chan
https://zenodo.org/record/3606810
you can call the api directly at https://playground-beta.gpt4chan.com/api/complete
example POST payload:
{"prompt":"-----\n--- 231316735\n","length":128,"top_p":0.9,"temperature":0.8}
It is running off of a single TPU, and If there are too many requests in the queue, the request will time out and return no results or error message to the user.