Go to http://gpu.gohai.xyz:7860/ (needs NYUSH VPN)

Note: this software doesn’t work well (yet) with multiple users at the same time. You might want to try this some random time during the week for best results…

In the Models tab you can load a number of different models. (Only one model can be loaded at the same time.)

In the Parameters > Generation tab you can select between different presents that that control the generation behavior of the model. (”Divine Intellect” is one, “simple-1” the default.)

For autocompleting models (like base models, such as Llama 2), it’s best to use the Default tab to generate text with the Prompt set to None.

For instruction-tuned models (such as chat-bots), you want instead want to use the Chat tab to talk 🤖.