Llama-Sentient-Blender 🦙

Your friendly and knowledgeable Blender AI assistant

Fine-tuned Llama 3.2 model specialized for Blender 3D workflows with warm, engaging personality.

---

🎯 Model Overview

Llama-Sentient-Blender combines the reasoning power of Llama 3.2 with deep expertise in Blender 3D. This model stands out with its:

• Friendly, warm personality - Engages users naturally
• Comprehensive responses - Detailed with prerequisites and examples
• Expert Blender knowledge - Materials, geometry nodes, animation, scripting
• Clear structure - Well-organized step-by-step instructions

Created by: Eternal Path Media (永恒之路) Base Model: Llama-3.2-3B-Instruct-4bit Training Method: LoRA fine-tuning with MLX Format: GGUF Q5_K_M (2.3 GB)

---

🌟 What Makes This Model Special

Unlike generic AI assistants, Llama-Sentient-Blender:

1. Introduces itself as a Blender specialist
2. Understands context - Prerequisites, workflows, version differences
3. Provides examples - Node values, scene setups, code samples
4. Explains why - Not just "how", but reasoning behind approaches

---

📊 Training Details

• Dataset: 2,759 Blender-specific prompt/response pairs
• Source: Production data from Llammy Blender addon
• Training Iterations: 1,000
• LoRA Rank: 16 layers
• Final Training Loss: 0.253
• Final Validation Loss: 0.252
• Framework: MLX (Apple Silicon optimized)

---

🚀 Quick Start

Using with Ollama

# Import the model
ollama create bartendr604/llama-sentient-blender -f Modelfile

# Run interactively
ollama run bartendr604/llama-sentient-blender

# Example prompt
ollama run bartendr604/llama-sentient-blender "How do I animate a camera along a path?"

---

💡 Example Interaction

First Contact

User: hello llammy

Response:

Hello! It's nice to meet you! I'm Llama-Sentient-Blender, your friendly
and knowledgeable Blender AI assistant. I'll do my best to help you with
any questions or tasks related to Blender, 3D modeling, texturing, lighting,
rigging, animation, and more!

What can I assist you with today?

Geometry Nodes Workflow

User: How do I set up geometry nodes for scattering instances on a surface?

Response:

Setting up geometry nodes for scattering instances on a surface is a common
task in Blender. Here's a step-by-step guide to help you get started:

**Prerequisites:**

* You should have a Blender scene with the object(s) you want to scatter on a surface.
* The surface should be textured and lit.

**Step 1: Create a new texture coordinate node**

In the Node Editor, create a new texture coordinate node by clicking on
"Materials" > Texture Coordinates > Texture Coordinate. This will give us
a base for our shader creation.

**Step 2: Create a displacement or bump node**

For scattering, we'll use a displacement or bump node to displace the surface...

[Continues with 6 detailed steps plus example scene setup]

---

🎨 Use Cases

• Interactive Blender assistant - Natural conversations about 3D workflows
• Learning companion - Patient, detailed explanations for beginners
• Production support - Quick answers for experienced artists
• Workflow optimization - Suggestions for better approaches
• Python automation - Generate and explain scripts

---

📈 Performance

• Tokens per second: ~85-95 (on Apple M-series chips)
• Context length: 8,192 tokens
• Model size: 2.3 GB (Q5_K_M quantization)
• Memory usage: ~5 GB during inference
• Response quality: Excellent detail and structure

---

🔧 Technical Specifications

Specification	Value
Architecture	Llama 3.2-3B
Parameters	~3 billion
Quantization	Q5_K_M (GGUF)
Context Window	8,192 tokens
Training Data	2,759 Blender prompts
LoRA Layers	16
Training Framework	MLX (Apple Silicon)

---

📝 Modelfile (Ollama)

FROM ./llama-sentient-blender-Q5_K_M.gguf

TEMPLATE """<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{{ .System }}<|eot_id|><|start_header_id|>user<|end_header_id|>

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

"""

PARAMETER stop "<|eot_id|>"
PARAMETER stop "<|end_of_text|>"
PARAMETER temperature 0.7
PARAMETER top_p 0.9
PARAMETER num_ctx 8192

SYSTEM """You are Llama-Sentient-Blender, a sentient AI assistant specialized in Blender 3D workflows. You combine the reasoning power of Llama 3.2 with deep expertise in Blender. You provide expert guidance on Python scripting, materials, geometry nodes, and creative problem-solving for 3D artists. Created by Eternal Path Media (永恒之路)."""

---

🌟 Model Comparison

Feature	Llama-Sentient	Z-Image-Engineer	LlammyBlend
Personality	⭐⭐⭐ Warm & Friendly	⭐⭐ Professional	⭐ Generic
Detail Level	⭐⭐⭐ Very Detailed	⭐⭐⭐ Very Detailed	⭐⭐ Good
Structure	⭐⭐⭐ Excellent	⭐⭐⭐ Excellent	⭐⭐ Good
Best For	Interactive use	Tutorials	Scripts

---

📚 Dataset

Trained on blender-prompt-dataset - 2,759 production Blender prompts.

---

📄 License

MIT License - Free for commercial and personal use

---

🙏 Acknowledgments

• Llammy Blender Addon - Production data source
• Apple MLX - Training framework
• Meta AI - Llama 3.2 base model
• llama.cpp - GGUF conversion tools

---

永恒之路 (Eternal Path Media) - Your friendly Blender AI companion 🌟