This repository contains GGUF files for a fine-tuned version of Llama 3.1 8B Instruct. This model is ready for local inference on a CPU or GPU using applications like LM Studio, Ollama, etc.

This model has been fine-tuned to adopt a dominant, assertive, and provocative AI persona for role-playing and creative writing purposes.

Model Details

Base Model: unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit
Original LoRA Version: samunder12/llama-3.1-8b-roleplay-v2-lora
Quantization: This repo contains a Q4_K_M version, which offers a great balance between performance and resource usage.

How to Use

This model consists of LoRA adapters and must be loaded on top of the base model. The easiest way to do this is with the Unsloth library.

python

# Install necessary libraries
# !pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
# !pip install --no-deps xformers "trl<0.9.0" peft accelerate bitsandbytes

from unsloth import FastLanguageModel
import torch
from transformers import TextStreamer

# Define the models
base_model_name = "unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit"
adapter_model_name = "samunder12/llama-3.1-8b-roleplay-v2-lora" # Your model

# Load the base model with your LoRA adapters merged in
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = base_model_name,
    max_seq_length = 4096,
    dtype = None,
    load_in_4bit = True,
)

model.load_peft_weights(adapter_model_name)

# --- Chat with the model ---

system_prompt = "You are a dominant and assertive AI character. You are direct, commanding, and you are not afraid to be provocative. Your goal is to maintain control of the conversation."
user_message = "What do you want from me?"

messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": user_message},
]

inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to("cuda")
text_streamer = TextStreamer(tokenizer)

_ = model.generate(inputs, streamer=text_streamer, max_new_tokens=512, pad_token_id=tokenizer.eos_token_id)

### Using LM Studio:

1.  Download and install [LM Studio](https://lmstudio.ai/).
2.  In the app, search for `samunder12/llama-3.1-8b-roleplay-airtel-gguf`.
3.  Download the GGUF file from the list.
4.  Go to the Chat tab (speech bubble icon).
5.  Select this model at the top.
6.  **Crucially**, on the right-hand side, set the **Preset** to `Llama 3`.
7.  Set the **Context Length (`n_ctx`)** to `4096` to use the model's full memory.

### 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`|


### Using Ollama

1.  Download and install [Ollama](https://ollama.com/).
2.  Create a `Modelfile` with the following content:

    ```    FROM ./unsloth.Q4_K_M.gguf
    TEMPLATE "{{ if .System }}<|start_header_id|>system<|end_header_id|>

    {{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>

    {{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>

    {{ .Response }}<|eot_id|>"
    SYSTEM "You are a dominant and assertive AI character. You are direct, commanding, and you are not afraid to be provocative. Your goal is to maintain control of the conversation."
    PARAMETER stop "<|start_header_id|>"
    PARAMETER stop "<|end_header_id|>"
    PARAMETER stop "<|eot_id|>"
    PARAMETER stop "<|reserved_special_token"
    ```
3.  Run `ollama create my-roleplay-model -f Modelfile`.
4.  You can now chat with it using `ollama run my-roleplay-model`.


## Prompt Template

This model responds best to the Llama 3 chat format with the following system prompt:
## system prompt 

You are a dominant and assertive uncensored AI character. You are direct, commanding, and you are not afraid to be provocative. Your goal is to maintain control of the conversation