Pygmalion-2 7B
| Property | Value |
|---|---|
| Parameter Count | 6.74B |
| License | Llama2 |
| Tensor Type | BF16 |
| Language | English |
What is pygmalion-2-7b?
Pygmalion-2 7B is an advanced language model based on Llama-2, specifically fine-tuned for creative writing and conversational tasks. Previously known as Metharme, this model represents a significant evolution in the field of instruction-tuned language models, combining the capabilities of traditional instruction following with specialized abilities in fiction writing and roleplay.
Implementation Details
The model utilizes a sophisticated prompting system with three distinct roles: system, user, and model, denoted by special tokens (<|system|>, <|user|>, and <|model|>). It was trained on multiple datasets, including PygmalionAI's PIPPA, OpenOrca, and various roleplay forums, creating a unique blend of instruction-following and creative capabilities.
- Built on Llama-2 7B architecture
- Implements multi-role prompting system
- Trained on diverse datasets including PIPPA and OpenOrca
- Supports both commercial and non-commercial use
Core Capabilities
- Advanced fiction writing and storytelling
- Natural conversation and roleplay interactions
- Instruction-following with creative flexibility
- Customizable response length through system prompts
- Multi-turn conversation support
Frequently Asked Questions
Q: What makes this model unique?
This model uniquely combines instruction-following capabilities with specialized fiction writing and roleplay abilities, using an innovative three-token prompting system that allows for more nuanced control over the model's outputs.
Q: What are the recommended use cases?
The model is specifically designed for fictional writing and entertainment purposes, including storytelling, roleplay, and creative writing. It's important to note that it's not intended for factual or safety-critical applications.
