georgesung/llama3_8b_chat_uncensored
georgesung/llama3_8b_chat_uncensored is an 8 billion parameter Llama-3 based language model fine-tuned by georgesung. It utilizes QLoRA fine-tuning on an uncensored/unfiltered Wizard-Vicuna conversation dataset. This model is designed for chat applications requiring less restrictive content generation. It is available in both fp32 HuggingFace and quantized 4-bit GGUF formats.
Loading preview...
Model Overview
georgesung/llama3_8b_chat_uncensored is an 8 billion parameter language model built upon the Meta-Llama-3-8B architecture. This model has been fine-tuned using QLoRA with a specific focus on an uncensored and unfiltered Wizard-Vicuna conversation dataset. The primary goal of this fine-tuning was to create a chat model with fewer content restrictions compared to its base model.
Key Features
- Base Model: Meta-Llama-3-8B, providing a robust foundation for conversational AI.
- Fine-tuning Method: QLoRA, enabling efficient fine-tuning with reduced memory requirements.
- Training Data: An uncensored/unfiltered Wizard-Vicuna conversation dataset, influencing its less restrictive output.
- Availability: Provided as a full precision (fp32) HuggingFace model and a quantized 4-bit q4_0 GGUF version for optimized local inference.
Usage and Deployment
The model is designed for conversational interactions, following a specific prompt style (### HUMAN:, ### RESPONSE:). Instructions are provided for local inference using Ollama with the GGUF version, including steps to create a custom Modelfile. The training code used for this fine-tuning is publicly available here, allowing for reproducibility and further experimentation.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.