alpindale/pygmalion-instruct
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:May 24, 2023License:mitArchitecture:Transformer0.0K Open Weights Cold
alpindale/pygmalion-instruct is an experimental 7 billion parameter instruction-tuned causal language model developed by alpindale. It is trained using a combination of the Pygmalion and WizardLM datasets. This model is specifically designed to combine complex instruction-following capabilities with role-playing functionalities, making it suitable for guiding conversational styles.
Loading preview...
Model Overview
alpindale/pygmalion-instruct is an experimental 7 billion parameter model that integrates the instruction-following strengths of WizardLM with the role-playing capabilities derived from the Pygmalion dataset. This unique combination aims to provide a versatile model for interactive applications.
Key Capabilities
- Complex Instruction Following: Designed to process and respond to intricate instructions, leveraging the WizardLM training.
- Role-Playing (RP) Functionality: Incorporates the conversational style and character interaction features from Pygmalion.
- Guiding Conversation Style: Intended to allow users to steer the bot towards specific conversational styles or personas.
Intended Use Cases
- Role-Playing with Instruct Prompts: The primary intended use is for scenarios where users want to engage in role-playing while also providing specific instructions to guide the interaction.
Limitations and Out-of-Scope Uses
- Assistant Bot: Not recommended for general assistant tasks, as it may provide incorrect instructions.
- Complex Multi-Character Chat: The model is not optimized for managing intricate conversations involving multiple distinct characters simultaneously.
- Potential for Harmful Content: Users should exercise caution as the model can generate potentially harmful or NSFW outputs.