LuminRP-7B-128k-v0.2

LuminRP-7B-128k-v0.2 is a merge of four RP models into one using LazyMergekit. This is a model that is purely for roleplaying and uses a context window of 128k.

Example Response:

I use the ChatML template for this with Instruct Mode enabled. Mistral template is okay to use as well, but I don't recommend Alpaca-Roleplay because it just keeps going. Most likely because the Alpaca-Roleplay template doesn't have a message suffix.

Quantized Version

GGUF: Ppoyaa/LuminRP-7B-128k-v0.2-GGUF

🏆 Open LLM Leaderboard Evaluation Results

Metric	Value
Avg.	73.18
AI2 Reasoning Challenge (25-Shot)	70.56
HellaSwag (10-Shot)	87.46
MMLU (5-Shot)	64.92
TruthfulQA (0-shot)	65.78
Winogrande (5-shot)	82.40
GSM8k (5-shot)	67.93

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Ppoyaa/LuminRP-7B-128k-v0.2"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])