language:
- en
license: apache-2.0
library_name: transformers
tags:
- merge
- mergekit
- lazymergekit
base_model:
- bunnycore/Llama-3.1-8B-TitanFusion-Test
- vicgalle/Roleplay-Hermes-3-Llama-3.1-8B
- vicgalle/Humanish-Roleplay-Llama-3.1-8B
- bunnycore/Llama-3.1-8B-TitanFusion-Mix
- kromeurus/L3.1-Siithamo-v0.4-8B
pipeline_tag: text-generation
model-index:
- name: Llama-3.1-8B-SpecialTitanFusion
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: IFEval (0-Shot)
type: HuggingFaceH4/ifeval
args:
num_few_shot: 0
metrics:
- type: inst_level_strict_acc and prompt_level_strict_acc
value: 74.02
name: strict accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: BBH (3-Shot)
type: BBH
args:
num_few_shot: 3
metrics:
- type: acc_norm
value: 34.82
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MATH Lvl 5 (4-Shot)
type: hendrycks/competition_math
args:
num_few_shot: 4
metrics:
- type: exact_match
value: 23.34
name: exact match
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GPQA (0-shot)
type: Idavidrein/gpqa
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 6.6
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MuSR (0-shot)
type: TAUR-Lab/MuSR
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 7.49
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU-PRO (5-shot)
type: TIGER-Lab/MMLU-Pro
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 29.12
name: accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ZeroXClem/Llama-3.1-8B-SpecialTitanFusion
name: Open LLM Leaderboard
🏆 ZeroXClem-Llama-3.1-8B-SpecialTitanFusion 🏆
A powerful fusion of Titan-level models, designed for enhanced roleplay, creativity, and intelligence.
📌 Overview
ZeroXClem-Llama-3.1-8B-SpecialTitanFusion is a meticulously crafted model merge leveraging state-of-the-art transformer architectures. Using mergekit, we combined multiple high-performance Llama-3.1 models to enhance context retention, creativity, and nuanced text generation.
This model is based on kromeurus/L3.1-Siithamo-v0.4-8B, with carefully selected models merged using the model_stock method.
🛠 Merge Details
🔄 Merge Method: model_stock
This model was merged using the model_stock method, ensuring a balanced and optimized blend of all contributing architectures.
📑 Models Merged
The following models contributed to this fusion:
- 🔷 kromeurus/L3.1-Siithamo-v0.4-8B
- 🦾 bunnycore/Llama-3.1-8B-TitanFusion-Test
- 🎭 vicgalle/Roleplay-Hermes-3-Llama-3.1-8B
- 💡 vicgalle/Humanish-Roleplay-Llama-3.1-8B
- 🔥 bunnycore/Llama-3.1-8B-TitanFusion-Mix
⚙ Configuration
name: ZeroXClem-Llama-3.1-8B-SpecialTitanFusion
base_model: kromeurus/L3.1-Siithamo-v0.4-8B
dtype: bfloat16
merge_method: model_stock
models:
- model: bunnycore/Llama-3.1-8B-TitanFusion-Test
- model: vicgalle/Roleplay-Hermes-3-Llama-3.1-8B
- model: vicgalle/Humanish-Roleplay-Llama-3.1-8B
- model: bunnycore/Llama-3.1-8B-TitanFusion-Mix
tokenizer_source: kromeurus/L3.1-Siithamo-v0.4-8B
🌟 Features & Capabilities
🔹 Highly dynamic writing – Perfect for storytelling, world-building, and creative applications.
🔹 Refined roleplay abilities – Enhanced persona handling, deep emotional responses, and immersive dialogue generation.
🔹 Better structured recall – Improved consistency across large-context conversations.
🔹 Balanced & non-restrictive responses – Adaptable across different use cases.
🛠 How to Use
🔥 Ollama (Quick Inference)
You can run the model using Ollama for direct testing:
ollama run hf.co/ZeroXClem-Llama-3.1-8B-SpecialTitanFusion
🤗 Hugging Face Transformers (Python)
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import torch
model_name = "ZeroXClem-Llama-3.1-8B-SpecialTitanFusion"
# Load tokenizer & model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.bfloat16,
device_map="auto"
)
# Initialize text generation pipeline
text_generator = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
torch_dtype=torch.bfloat16,
device_map="auto"
)
# Example prompt
prompt = "Describe the significance of AI ethics in modern technology."
# Generate output
outputs = text_generator(
prompt,
max_new_tokens=200,
do_sample=True,
temperature=0.7,
top_k=50,
top_p=0.95
)
print(outputs[0]["generated_text"])
🔧 Recommended Usage
📜 Prompting Style
For best results, use system prompts similar to Llama-3.1 Instruct.
Example system message:
Think step by step with a logical reasoning and intellectual sense before you provide any response.
For enhanced creativity in roleplay, try:
### Instruction:
You are an advanced roleplaying assistant. Maintain deep character consistency and immersive storytelling.
🏗 Model Settings
For optimal output quality, use the following settings:
Temperature: 1.2
Min P: 0.1
Repeat Penalty: 1.05
Repeat Penalty Tokens: 256
Smooth Sampling: 0.18
🔥 Disclaimer
🔹 Use responsibly!
This model follows Meta’s Llama-3.1 Community License Agreement. It is an uncensored model, meaning that alignment should be implemented based on individual use cases.
🔹 You are responsible for the content you generate.
Please ensure compliance with ethical AI guidelines when deploying this model in production environments.
💬 Feedback & Contributions
If you have suggestions or improvements, feel free to open a discussion on Hugging Face! Let's continue improving the Llama-3.1 merging meta-game! 🚀
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 29.23 |
| IFEval (0-Shot) | 74.02 |
| BBH (3-Shot) | 34.82 |
| MATH Lvl 5 (4-Shot) | 23.34 |
| GPQA (0-shot) | 6.60 |
| MuSR (0-shot) | 7.49 |
| MMLU-PRO (5-shot) | 29.12 |