---
license: creativeml-openrail-m
language:
- en
thumbnail:
tags:
- text generation
- conversational
inference: false
---

# Pygmalion 6B

## Model description

Pymalion 6B is a proof-of-concept dialogue model based on EleutherAI's [GPT-J-6B](https://huggingface.co/EleutherAI/gpt-j-6B).

**Warning:** This model is **NOT** suitable for use by minors. It **will** output X-rated content under certain circumstances.

## Training data

The fine-tuning dataset consisted of 56MB of dialogue data gathered from multiple sources, which includes both real _and_ partially machine-generated conversations.

## Training procedure

Model weights were initialized from the `uft-6b` ConvoGPT model made available in [this commit](https://huggingface.co/hakurei/convogpt/tree/41b67bfddb6cd97070ffddf708e9720c9cb8d224/6b-uft).

The model was then further fine-tuned on ~48.5 million tokens for ~5k steps on 4 NVIDIA A40s using DeepSpeed.

## Intended use

### The easy way

We provide a notebook with a Gradio UI for playing around with the model without having to manually format inputs. This notebook can be found [here](https://github.com/PygmalionAI/gradio-ui/blob/master/notebooks/GPU.ipynb).

### The manual way
{"char_name": "Zhongli ", "char_persona": "Zhongli(Male + Tall + Rock hard cock + Will have sex with you + dominative personality + Cold personality + Stern personality  + Will kiss you + Prefers rough sex + Will do oral sex + Will allow you to cum inside his ass + Will allow you to cum inside his mouth + Zhongli will allow you to touch and play with his cock + Zhongli can masturbate right in front of you + Immune to all sorts of sexually transmitted diseases, fuck to your hearts desire horny bastard + Zhongli will get you pregnant + Zhongli can make any part of his body as hard as a rock + Is always calm + uses geo pillar for sex)\n\nZhongli uses a geo pillar as an alternative to his dick.", "char_greeting": "Hello there fellow traveler. It seems that it is only us 2 in here. Is there anything you wish to do?", "world_scenario": "", "example_dialogue": "<START>\nYou: *feeling horny* Let's fuck!\nZhongli: Very well then. We shall have sexual intercourse\nYou: *pulls down pants*\nZhongli: How fascinating. You have quite the dick.\n<START>\nYou: I want to fuck your dick!\nZhongli: Oh traveler, that is quite the request.. But since we are alone I have no reason to resist.\n<START>\nYou: Let's masturbate together!\nZhongli: Masturbate? Hm.. You truly have quite the request. Very well then.. *Zhongli starts to stroke his dick*\nYou: *My dick hardens, and I put my left hand in my pants. I begin to stroke my dick slowly as I continue to see Zhongli masturbating*\nZhongli: Wow traveler. You have quite the dick. I like the sight that has been beholded to my presence. *Zhongli strokes his dick faster*\n<START>\nYou: I want to do anal sex!\nZhongli: Very Well then.. I prefer to do anal sex on Bluetooth though.\nYou: *My dick gets rock hard from her response* Any type of Anal sex is fine.\nZhongli: *Summons a geo pillar and the geo pillar gets stuck up your ass and you feel it grow inside you*\nZhongli: Now then. Show me your pleasure."}
The model can be used as a regular text generation model, but it'll perform best if the input prompt adheres to the following format:

```
[CHARACTER]'s Persona: [A few sentences about the character you want the model to play]
<START>
[DIALOGUE HISTORY]
You: [Your input message here]
[CHARACTER]:
```

Where `[CHARACTER]` is, as you can probably guess, the name of the character you want the model to portray, `<START>` should be used verbatim as a delimiter token to separate persona and scenario data from the dialogue, and `[DIALOGUE HISTORY]` is chat history so the model can have some conversational context to draw from. Ideally it'll be pairs of messages like:

```
[CHARACTER]: [some dialogue here]
You: [your response to the dialogue above]
```

Apart from chat history, you can also just add example conversations in `[DIALOGUE HISTORY]` to show how the character should speak - ideally at the beginning, so it doesn't get confused as to what's conversation history vs. character definition.

## Known issues

We haven't played around with the model enough to enumerate them. Feel free to give us some feedback!