|
--- |
|
license: apache-2.0 |
|
tags: |
|
- merge |
|
- mergekit |
|
- della-linear |
|
- Hermes3 |
|
- SuperNova |
|
- Purosani |
|
- Llama3.1 |
|
- Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B |
|
- instruction-following |
|
- long-form-generation |
|
- storytelling |
|
base_model: |
|
- ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B |
|
--- |
|
|
|
# **L3SAO-Mix-SuperHermes-NovaPurosani-8B** |
|
|
|
**L3SAO-Mix-SuperHermes-NovaPurosani-8B** is an innovative merged model that combines high-performance elements from two prominent models to create a powerhouse capable of excelling in a wide range of tasks. Whether it's for **instruction-following**, **roleplaying**, or **complex storytelling**, this model is designed for adaptability and precision. |
|
|
|
## 🌟 **Family Tree** |
|
|
|
This model is a **hybrid** of the following: |
|
|
|
- [**ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B**](<https://huggingface.co/ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B>) |
|
- [**Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B-V1_fp32-merge-calc**](<https://huggingface.co/Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B-V1_fp32-merge-calc>) |
|
|
|
These models are themselves built upon a solid foundation of advanced AI architectures, ensuring a model that’s both **robust** and **versatile** for multiple applications. |
|
|
|
## 🌳 **Model Family Genealogy** |
|
|
|
This model represents the fusion of **Hermes3**'s instruction-following prowess and **bluuwhale's** rich contextual understanding, making it perfect for tasks that require **long-form generation** and **complex contextual analysis**. |
|
|
|
--- |
|
|
|
## 🧬 **Detailed Model Lineage** |
|
|
|
### **A: ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B** |
|
|
|
This model is built from: |
|
|
|
- **NousResearch/Hermes-3-Llama-3.1-8B**: Known for its strong instruction-following capabilities and contextual understanding. |
|
- **THUDM/LongWriter-llama3.1-8B**: Focused on **long-form content generation**, capable of handling over 10,000 words in a single pass, making it perfect for detailed content creation. |
|
|
|
### **B: Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B-V1** |
|
|
|
This model incorporates components from: |
|
|
|
- **Sao10K/L3-8B-Stheno-v3.2** |
|
- **Sao10K/L3-8B-Tamamo-v1** |
|
- **Sao10K/L3-8B-Lunaris-v1** |
|
|
|
Its primary strengths lie in **instructional roleplaying** and **creative content generation**. |
|
|
|
--- |
|
|
|
## 🛠️ **Merge Details** |
|
|
|
This model was merged using the **Della Linear** method with **bfloat16** precision. The process involved merging key elements from both parent models to balance **instruction-following** with **creative contextual analysis**. |
|
|
|
The following YAML configuration was used during the merge: |
|
|
|
```yaml |
|
merge_method: della_linear |
|
dtype: bfloat16 |
|
parameters: |
|
epsilon: 0.1 |
|
lambda: 1.0 |
|
int8_mask: true |
|
normalize: true |
|
|
|
base_model: ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B |
|
models: |
|
- model: ZeroXClem/Llama3.1-Hermes3-SuperNova-8B-L3.1-Purosani-2-8B |
|
parameters: |
|
weight: 1 |
|
density: 0.5 |
|
- model: Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B-V1_fp32-merge-calc |
|
parameters: |
|
weight: 1 |
|
density: 0.55 |
|
|
|
``` |
|
|
|
--- |
|
|
|
## 🎯 **Extended Roleplay & Storytelling Features** |
|
|
|
With its heritage from **SuperNova** and **bluuwhale**, this model excels in **immersive storytelling** and **dynamic roleplay scenarios**. It can handle: |
|
|
|
- **Long-form character development**: Crafting rich, nuanced personalities for interactive narratives. |
|
- **World-building & lore**: Generating detailed worlds and interconnected lore on the fly. |
|
- **Dynamic dialogues**: Perfect for game development, this model can generate complex, believable conversations for NPCs in real-time. |
|
|
|
--- |
|
|
|
## 🚀 **Key Features & Capabilities** |
|
|
|
### **1. Long-Form Content Generation** |
|
|
|
This model is ideal for generating large bodies of text without losing coherence, making it perfect for: |
|
|
|
- **Research papers** |
|
- **Novels** |
|
- **Detailed reports** |
|
|
|
### **2. Advanced Instruction-Following** |
|
|
|
Thanks to its **Hermes3** roots, this model can effectively follow complex instructions for: |
|
|
|
- **Task automation** |
|
- **AI assistants** |
|
- **Research and summarization tasks** |
|
|
|
### **3. Roleplay and Storytelling** |
|
|
|
The model’s ability to handle both short and long interactions makes it perfect for: |
|
|
|
- **Roleplaying games** |
|
- **Interactive storytelling** |
|
- **Narrative creation** |
|
|
|
--- |
|
|
|
## 📜 **License** |
|
|
|
This model is available under the **Apache-2.0 License**, allowing users to utilize and modify it freely with attribution. |
|
|
|
## 💡 **Tags** |
|
|
|
- `merge` |
|
- `mergekit` |
|
- `Hermes3` |
|
- `SuperNova` |
|
- `Purosani` |
|
- `Llama3.1` |
|
- `instruction-following` |
|
- `long-form-generation` |
|
- `storytelling` |
|
|
|
--- |
|
|
