Blue Orchid 2x7b

Introduction

Blue-Orchid-2x7b is a roleplaying-focused Mixture of Experts (MoE) model designed for text generation tasks. It combines expertise in roleplaying and storywriting, making it versatile for both applications. The base model used is SanjiWatsuki/Kunoichi-DPO-v2-7B.

Architecture

This model utilizes a Mixture of Experts approach, where:

• Expert 1 merges models focused on roleplaying, including LimaRP, Limamono, Noromaid 0.4 DPO, and good-robot.
• Expert 2 merges models focused on storywriting, including Erebus, Holodeck, Dans-AdventurousWinds-Mk2, Opus, Ashhwriter, and good-robot.

Training

Blue-Orchid-2x7b is built by merging specific models to optimize performance for roleplaying and storywriting tasks. The training details, including datasets and specific methodologies, are not provided.

Guide: Running Locally

To run Blue-Orchid-2x7b locally, follow these steps:

1. Clone the repository containing the model from Hugging Face.
2. Install dependencies such as the transformers and safetensors libraries.
3. Load the model using your preferred Python environment.
4. Configure your input using the provided prompt templates (LIMARP or Alpaca) for optimal results.
5. Execute the model to generate responses based on your input.

For efficient performance, consider using cloud GPU services such as AWS EC2, Google Cloud Platform, or Azure.

License

The Blue-Orchid-2x7b model is licensed under the Apache 2.0 License, allowing for free use, modification, and distribution under the terms stipulated in the license.