--- license: apache-2.0 library_name: transformers base_model: - meta-llama/Llama-3.1-8B-Instruct pipeline_tag: text-generation tags: - roleplay - rp - character - peft ---
Peach
This model is designed for easy, local inference on CPUs and GPUs using llama.cpp-based software like **LM Studio** and **Ollama**. The model embodies a dominant, assertive, and creative persona for role-playing and storytelling. It was fine-tuned on a multi-turn conversational dataset to enhance its coherence and memory. ## Model Details * **Original LoRA Model:** [`samunder12/llama-3.1-8b-roleplay-v3-lora`](https://huggingface.co/samunder12/llama-3.1-8b-roleplay-v3-lora) * **Quantization:** `Q4_K_M`. This method provides an excellent balance between model size, performance, and VRAM/RAM usage. * **Context Length:** `4096` tokens. ## Usage Instructions ### LM Studio (Recommended) 1. Download and install [LM Studio](https://lmstudio.ai/). 2. In the app, search for this model repo: `samunder12/llama-3.1-8b-roleplay-v3-gguf`. 3. Download the GGUF file listed in the "Files" tab. 4. Go to the Chat tab (💬 icon) and load the model you just downloaded. 5. **CRITICAL:** On the right-hand panel, under "Prompt Format", select the **Llama 3** preset. 6. Set the `Context Length (n_ctx)` to **4096**. 7. Use the "Role-Play" sampler settings below for best results. #### Recommended Sampler Settings (Role-Play Preset) | Setting | Value | | :--- | :--- | | **Temperature** | `0.75` | | **Repeat Penalty** | `1.06` | | **Mirostat** | `Mirostat 2.0` | | **top_p** | `0.92 `| | **top_k** | `40 or 100`|