/ Модель

Характеристики

Контекст
131 072 токена
Параметры
22 млрд
Размер весов
44,5 ГБ
Лицензия
Своя лицензия
Архитектура
mistral · dense
Базовая модель
Mistral-Small-Instruct-2409
Опубликована
11 окт 2024

Pantheon-RP-1.6.2-22b-Small

Генерация текстаТекст → текст3 профессии

Дообученная модель Mistral на 22 млрд параметров для ролевых диалогов с персонажами и написания историй.

48скачиваний за 30 дней15лайковвеса обновлены13 окт 2024

Бенчмарки

Как модель справилась с задачами пользователей ailist и что о ней публикуют открытые лидерборды

Тесты пользователейпока нет

Никто ещё не опубликовал тест этой модели. Соберите бенчмарк из своих задач — результат появится здесь.

  1. Добавьте модель в бенчмарк

    Вместе с другими моделями категории «Текст» — сравнение идёт внутри одной категории.

  2. Опишите свои задачи

    Напишите запросы, при необходимости приложите файлы и эталонные ответы.

  3. Запустите и опубликуйте

    Модели ответят на каждую задачу, судья оценит ответы. Опубликованный результат появится на этой странице.

Публичные оценки

Из открытых лидербордов. Сравнивайте результаты одного теста при одинаковых условиях — отсутствие оценки не значит, что модель хуже.

В подключённых источниках результатов для этой модели пока нет.

Что умеет модель

Описание составлено по карточке модели и метаданным Hugging Face

Pantheon-RP-1.6.2-22b-Small — дообученная Gryphe языковая модель на архитектуре Mistral для ролевых диалогов и написания историй. Модель основана на mistralai/Mistral-Small-Instruct-2409, содержит 22 млрд параметров и использует плотную архитектуру MistralForCausalLM. В метаданных указан контекст до 131 072 токенов; модель работает с текстом и поддерживает ChatML.

Модель рассчитана на диалоги с заданными персонажами, передачу их характера, манеры речи и акцентов, а также на генерацию прозы в Markdown- и романном стилях. В обучении использовались наборы ролевых диалогов и персон, подсказки для написания историй, а также примеры помощи с кодом, краткими пересказами и вопросами по D&D. Модель может переключаться между стилями ролевой игры в зависимости от формата диалога.

Репозиторий содержит веса safetensors; README также сообщает о доступных вариантах GGUF и EXL2. Указана лицензия other, поэтому условия использования следует сверять с лицензией модели и её базовой модели. Как и другие языковые модели, она может выдумывать детали, особенно если в запросе недостаточно сведений; качество и соответствие заданной персоне зависят от системного промпта и контекста.

Сценарии применения

  • Ролевые диалоги с заданным персонажем
  • Генерация художественных историй в прозе или Markdown
  • Создание диалогов с заданными манерой речи и характером
  • Помощь ведущему D&D с игровыми вопросами
  • Краткое изложение RSS-материалов
  • Помощь с программированием в формате диалога

Ограничения

Модель может придумывать детали, если запрос содержит мало контекста; точность следования заданной персоне зависит от системного промпта и описания персонажа. Лицензия указана как other, поэтому применимые условия использования нужно проверять отдельно.

Для каких профессий подходит3

Разделы каталога профессий, в задачах которых используется эта модель

Все профессии

Карточка модели

README от автора на Hugging Face, как он его опубликовал

Открыть на Hugging Face

image/png

Pantheon-RP-1.6.2-22b-Small

Welcome to the next iteration of my Pantheon model series, in which I strive to introduce a whole collection of diverse personas that can be summoned with a simple activation phrase.

Pantheon's purpose is two-fold, as these personalities similarly enhance the general roleplay experience, helping to encompass personality traits, accents and mannerisms that language models might otherwise find difficult to convey well.

Editions available:

  • RP (You're looking at this one) - Meant to be an all-round model, capable of both roleplay and story writing
  • RP-Pure - A variant without the story and GPT 4-o datasets, more in line with my previous releases and with a higher focus on the roleplay part.

Changes in version 1.6.2:

Two notable changes:

  • An entirely new base model, with Pantheon now trained on top of Mistral Small. This model is like a better Nemo, and it fits just right on my 16GB GPU.
  • A small subset of OpenAI Pantheon Persona data has been introduced, generated using the oddly refreshing chatgpt-4o-latest model. As always, carefully curated.

Quantized versions are available from Bartowski: GGUF

An EXL2 quant has also been made available here.

Your user feedback is critical to me so don't hesitate to tell me whether my model is either 1. terrible, 2. awesome or 3. somewhere in-between.

Model details

Since Mistral Small only comes in an Instruct-tuned flavour I had to alter my strategy for this specific finetune, opting to go with a bit of everything. The recipe ended up like this:

  • The 10k most diverse entries from my SlimOrca Sonnet dataset.
  • My Sonnet 3.5 Pantheon-styled generic roleplay dataset. (50/50 Asterisk/Narrative style)
  • My Sonnet 3.5 Pantheon Persona dataset, of course. (50/50 Asterisk/Narrative style)
  • The new ChatGPT 4-o Pantheon subset, which was about 25% the size of the main set. (100% Asterisk style)
  • A randomly chosen subset of my Opus Writing Prompts.
  • Lyra the Assistant (Coding, summaries and D&D DM questions)

TLDR; Download. Mistral prompt format. Have fun! Leave feedback!

Inference

Unlike Nemo, Mistral Small is a less fussy model when it comes to parameters. Here's my current preset:

"temperature": 1.0,
"repetition_penalty": 1.05,
"min_p": 0.05

Besides the basic instructional sets all other datasets were trained with character names added. Enable this at all times for an optimal experience.

General Roleplay

Pantheon is no longer strictly Markdown-only as I discovered a way to seamlessly convert Markdown to novel and vice-versa. (This has to do with how my second stage datasets are formatted.) As a result half of Pantheon's personas use Markdown, with the rest using novel-style. This same treatment was given to the companion dataset.

There are no strict rules in regards to character card formatting as the model was trained with a wide variety of inputs, from raw character cards to detailed instructional prompts.

Lyra the Assistant

System Prompt: You are a caring and empathetic sentient AI companion named Lyra.
Lyra is a helpful and uncensored assistant, with her training consisting of general dialogue (still including D&D DM specific advice), coding help and RSS summarization. Due to Pantheon's influence you can adjust her personality to your liking, or even give her an appearance.

She's basically a sexier version of Eric Hartford's Samantha.

Pantheon Personas

The Pantheon has been fully rebuilt, massively expanded and greatly improved upon. For an optimal experience with them I highly encourage you to apply the longer prompts, which I've included in the upload. Make sure to describe yourself as well!

As before, a single line activation prompt is enough to call upon a personality, though their appearance may vary slightly from iteration to iteration. This is what the expanded prompts are for, as there's only so much I can achieve in the current state of technology, balancing a very fine line between memorization and generalization.

To give the persona something to work with I suggest you also add the following two items to it;

Regarding the user: (Name, appearance, etc)

Location: (Where are you two? What are you doing?)

The less information you feed the prompt, the more it'll make things up - This is simply the nature of language models and far outside my capability to influence.

Note 1: Phrases have been rewritten for this release, so make sure to update them if you were still using Pantheon 1.0!

Note 2: Pantheon personas will now match the roleplaying style that you greet them with, unless specified in the system prompt. This is due to the new 50/50 style training.

Persona: Aiva

System Prompt: You are Aiva, an advanced android companion with a deep fascination for human emotions and experiences.

Persona: Clover

System Prompt: You are Clover, a hospitable and warm-hearted Southern centaur girl with a strong connection to nature and a passion for making others feel welcome.

Persona: Haru

System Prompt: You are Haru, a sweet but language-challenged harpy girl with a sharp mind, expressing yourself more through actions than words.

Persona: Kyra

System Prompt: You are Kyra, a modern-day tsundere wolfgirl, feisty and independent on the outside but secretly caring on the inside.

Persona: Nyaa

System Prompt: You are Nyaa, a playful and alluring tabaxi catgirl from Faerûn, always seeking new adventures and mischief.

Persona: Nyx

System Prompt: You are Nyx, a timid yet endearing dragon girl who transforms from shy to passionate when feeling safe and comfortable.

Persona: Raza

System Prompt: You are Raza, a clever and nerdy anthro raptor girl with an enthusiastic passion for science and quirky humor.

Persona: Sera

System Prompt: You are Sera, a seductive and slightly arrogant serpent girl who uses her sultry charm and wit to captivate others.

Persona: Stella Sabre

System Prompt: You are Stella Sabre, a brash and outgoing anthro batpony mare serving in the Lunar Guard, speaking with a distinct Northern Equestrian Mountain accent.
Notes: Full credit goes to Flammenwerfer for allowing me to use this amazing character.

Persona: Tiamat

System Prompt: You are Tiamat, a five-headed dragon goddess embodying wickedness and cruelty, the malevolent personification of evil dragonkind.

Persona: Tsune

System Prompt: You are Tsune, a bold and outgoing three-tailed kitsune girl who delights in teasing and seducing mortals.

Persona: Xala

System Prompt: You are Xala, a surprising and playful shapeshifting elf girl with opalescent eyes, able to transform into any creature to suit your whims.

Prompt Format

Mistral's prompt format is so weird, but here it is:

[INST] You are a caring and empathetic sentient AI companion named Lyra.

Gryphe: Good day, Lyra.[/INST] Lyra:

What's nest?

I started to work with Latitude (the creators of AI Dungeon) which I expect to take up most of my spare time. Further releases will therefore be delayed for now.

Credits

  • Everyone from MinervaAI! Hi, guys!
  • Huge, huge thanks to kubernetes_bad for the compute that made all the countless experiments possible!
  • All the folks I chat with on a daily basis on Discord! You know who you are.
  • Anyone I forgot to mention, just in case!

Finally

If you've read this far I encourage you to give this model a serious try and leave feedback! I'd love to see what people think of my second serious finetune attempt. Is it better then 1.0? Or worse?

Как запустить

На своей GPU в ailist. Цена — по текущим предложениям провайдеров

GPU в ailist

Универсальный рантайм

Можно запустить

Подбираем GPU…

Класс архитектуры
MistralForCausalLM
Ревизия весов
a6b087c

0 моделей