license: other
Alpac(ino) stands for Alpaca Integrated Narrative Optimization.
This model is a triple model merge of (Alpaca+(CoT+Storytelling)), resulting in a comprehensive boost in Alpaca's reasoning and story writing capabilities. Alpaca was chosen as the backbone of this merge to ensure Alpaca's instruct format remains dominant.
Below is one use case example of what is possible when using Alpac(ino) in a convenient language model interface such as Text-Generation-WebUI or KoboldAI: An infinite text-based adventure game can be setup by switching to chat mode (in this case username is Player and AI is Narrator) and tailoring the following template in the context/memory section of the interface:
Make Narrator function as a text based adventure game that responds with verbose, detailed, and creative descriptions of what happens next after Player's response. Make Player function as the player input for Narrator's text based adventure game, controlling a character named (insert character name here, their short bio, and whatever quest or other information to keep consistent in the interaction).
If using either KAI or TGUI for model inference, testing subjectively suggests ideal presets are "Storywriter" (temp raised to 1.1) or "Godlike" with context tokens maxed at 2048 and max generation tokens ~680 or greater. This model is intelligent enough to determine when to stop writing, and will rarely use half those tokens; however verbosity will vary depending on provided instructions.
Legalese: This release contains modified weights from the LLaMA-13b model. This model is under a non-commercial license. You should only use this repository if you have been granted access to the model by filling out the form below- https://docs.google.com/forms/d/e/1FAIpQLSfqNECQnMkycAp2jP4Z9TFX0cGR4uf7b_fBxjY_OjhJILlKGA/viewform Llama models (7b, 13b, 33b, 65b) are all property of Meta AI and are intended for research purposes. This release is commensurate with good faith that those who download and/or utilize this model have been granted research access through the provided form.
Obligatory: This model may output offensive text and/or infactual statements; do not use this model for advice in any domain, especially medical or mental health advice. Meta AI and I are not liable for improper use or any damages, percieved or otherwise.
Sourced LoRA Credits:
ChanSung's exellently made Alpaca LoRA https://huggingface.co/chansung/alpaca-lora-13b https://huggingface.co/datasets/yahma/alpaca-cleaned https://github.com/gururise/AlpacaDataCleaned
magicgh's valuable CoT LoRA https://huggingface.co/magicgh/llama13b-lora-cot https://huggingface.co/datasets/QingyiSi/Alpaca-CoT https://github.com/PhoebusSi/alpaca-CoT
GamerUntouch's unique Storytelling LoRA https://huggingface.co/GamerUntouch/Storytelling-LLaMa-LoRAs