Llama 3.1 8 B Stheno v3.4

Introduction

Llama-3.1-8B-Stheno-v3.4 is a language model that has undergone finetuning through multiple stages, focusing on conversational-instruction and creative writing/roleplay datasets. It aims to introduce unique characteristics while improving multi-turn conversation coherency.

Architecture

The model uses the LLAMA architecture, leveraging multiple datasets, namely Setiaku/Stheno-v3.4-Instruct and Setiaku/Stheno-3.4-Creative-2, which include a mixture of human and Claude data. The model primarily supports English.

Training

The model's training process involved two key stages:

1. Multi-turn Conversational-Instruct finetuning.
2. Creative Writing/Roleplay with additional creative-based instruct datasets.

These stages incorporated enhancements such as improved prompts and answers, increased roleplaying examples filtered from Gryphe's Charcard RP Sets, and datasets targeting system prompt adherence and reasoning/spatial awareness.

Guide: Running Locally

To run the model locally, follow these steps:

1. Install Required Libraries: Ensure you have Python and necessary libraries like Hugging Face Transformers installed.

   pip install transformers
   

2. Clone the Repository: Access the model's files and versions via Hugging Face's model card link.

3. Load the Model: Use the Transformers library to load the model.

   from transformers import AutoModel
   
   model = AutoModel.from_pretrained("Sao10K/Llama-3.1-8B-Stheno-v3.4")
   

4. Inference: Implement the desired inference or task using the model.

For optimal performance, consider using cloud-based GPU services such as AWS, Google Cloud, or Azure.

License

The model is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (cc-by-nc-4.0), which permits use, sharing, and adaptation for non-commercial purposes, provided appropriate credit is given.