---
license: apache-2.0
tags:
- merge
- mergekit
- lazymergekit
- Chainbase-Labs/Theia-Llama-3.1-8B-v1
- EpistemeAI/Fireball-Meta-Llama-3.2-8B-Instruct-agent-003-128k-code-DPO
- aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored
- DeepAutoAI/ldm_soup_Llama-3.1-8B-Inst
---

# ZeroXClem/Llama3.1-TheiaFire-DarkFusion-8B

ZeroXClem/Llama3.1-TheiaFire-DarkFusion-8B is a merge of the following models using [mergekit](https://github.com/cg123/mergekit):
* [Chainbase-Labs/Theia-Llama-3.1-8B-v1](https://huggingface.co./Chainbase-Labs/Theia-Llama-3.1-8B-v1)
* [EpistemeAI/Fireball-Meta-Llama-3.2-8B-Instruct-agent-003-128k-code-DPO](https://huggingface.co./EpistemeAI/Fireball-Meta-Llama-3.2-8B-Instruct-agent-003-128k-code-DPO)
* [aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored](https://huggingface.co./aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored)
* [DeepAutoAI/ldm_soup_Llama-3.1-8B-Inst](https://huggingface.co./DeepAutoAI/ldm_soup_Llama-3.1-8B-Inst)

## 🧩 Configuration

```yaml
models:
  - model: Chainbase-Labs/Theia-Llama-3.1-8B-v1
    parameters:
      density: 0.4  # Balancing technical vision and crypto capabilities
      weight: 0.3
  - model: EpistemeAI/Fireball-Meta-Llama-3.2-8B-Instruct-agent-003-128k-code-DPO
    parameters:
      density: 0.6  # Giving priority to code-based reasoning and agentic capabilities
      weight: 0.4
  - model: aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored
    parameters:
      density: 0.5  # Focus on creativity and uncensored roleplay flexibility
      weight: 0.2
  - model: DeepAutoAI/ldm_soup_Llama-3.1-8B-Inst
    parameters:
      density: 0.5  # Blending latent diffusion capabilities for unseen tasks
      weight: 0.1

merge_method: ties
base_model: DeepAutoAI/ldm_soup_Llama-3.1-8B-Inst  # Using the ldm_soup as a base for optimal diffusion of strengths
dtype: bfloat16
parameters:
  normalize: true  # Ensuring consistent scaling of layer outputs
out_dtype: float16  # Maintaining efficiency and precision

```