Carl Kenner

This addresses Issue #1383 Adds StableLM Instruction Following support in Chat mode and Notebook mode for Tuned models. Adds support for 4096 tokens with StableLM models. I tested the 3...

Galactica was trained to answer questions in a variety of special formats. I included examples of some of them from the documentation. References: * [Galactica Prompt Book ](https://github.com/paperswithcode/galai/blob/main/docs/PROMPTBOOK.md) * [Galactica...

Support for MPT, INCITE, WizardLM, StableLM, Galactica, Vicuna, Guanaco, and Baize instruction following

10

This fixes Issue https://github.com/oobabooga/text-generation-webui/issues/1383 * Adds characters and prompts for MPT, INCITE, WizardLM, StableLM, Galactica, Vicuna, Guanaco, and Baize. * _Uses the Alpaca format for LaMini models_ * Adds presets...

Gpt4all looks like it is instruction-tuned, but I can't find anywhere what instruction format it uses or was trained on. How are the questions and answers tagged in the text...

If we include special tokens in the input prompt, we should print them in our debug prints. Debug prints should match what is actually being sent to the model.