---
datasets:
- Sweaterdog/Andy-4-base-2
- Sweaterdog/Andy-4-ft
language:
- en
base_model:
- unsloth/Qwen2.5-1.5B-Instruct-unsloth-bnb-4bit
tags:
- gaming
- minecraft
- mindcraft
---

# 🤏 Andy‑4‑micro 🧠

![image/png](https://cdn-uploads.huggingface.co/production/uploads/66960602f0ffd8e3a381106a/NFDS9XMy9YMM8KSn1Jjcd.png)

**Andy‑4‑micro** is a lightweight Minecraft-tuned AI model derived from the Andy‑4 architecture. Built for responsiveness and portability, it’s ideal for local testing, light inference, and experimentation within the **Mindcraft** framework.

**The current version of Andy-4-micro is `Andy-4-micro-0516`**, All previous versions of Andy-4-micro can still be found on my huggingface page.

> 💡 Trained on a **single RTX 3070** over **four days**, Andy‑4‑micro maintains strong performance while staying efficient.

> ⚠️ **Certification:**  
> Andy‑4‑micro is **not yet certified** by the Mindcraft developers. Use in production at your own discretion.

---

## 📊 Model Overview

- **Base Architecture:** Qwen 2.5
- **Parameter Count:** 1.5 B
- **Training Duration:** ~4 days
- **Training GPU:** 1 × NVIDIA RTX 3070
- **Total Tokens Used:** ~42M
- **License:** [Andy 1.1 License](LICENSE)
- **Repository:** https://huggingface.co/Sweaterdog/Andy-4-micro

---

## 🚀 Installation

First, you need to choose your quantization, this chart is with the base of `8192` set as the context window

| Quantization | VRAM Required |
|--------------|---------------|
|--------------|---------------|
| F16          | 5 GB       |
| Q8_0       | 3 GB+         |
| Q5_K_M       | 2 GB+         |
| Q3_K_M       | 1GB or CPU   |

**NOTE:** GPUs made before 2017 will have *significantly slower speeds* than newer GPUs, also, CPU inference will be extremely slow.

### 1. Installation directly on Ollama

1. Visit [Andy-4 on Ollama](https://ollama.com/Sweaterdog/Andy-4)
2. Copy the command after choosing model type / quantization
3. Run the command in the terminal
4. Set the profile's model to be what you installed, such as `ollama/sweaterdog/andy-4:latest`

### 2. Manual Download & Setup

1. **Download**
   - Visit the Hugging Face **Files** tab.
   - Download the `.GGUF` quantization weights (e.g. `Andy-4-micro.Q4_K_M.gguf`).
   - Grab the provided `Modelfile`.

2. **Edit `Modelfile`**

Change the path placeholder:

```text
FROM YOUR/PATH/HERE
```

to:

```text
FROM /path/to/Andy-4-micro.Q4_K_M.gguf
```

*Optional*: Adjust `num_ctx` for longer context windows if your system supports it.

3. **Create Model**

```bash
ollama create andy-4-micro -f Modelfile
```

This registers Andy‑4‑micro locally with Ollama.

---

If you lack a GPU, check the [Mindcraft Discord guide](https://ptb.discord.com/channels/1303399789995626667/1347027684768878644/1347027684768878644) for free cloud setups.


## 🔧 Context‑Window Quantization

To lower VRAM use for context windows:

#### **Windows**

1. Close Ollama.  
2. In **System Properties → Environment Variables**, add:  
   ```text
   OLLAMA_FLASH_ATTENTION=1  
   OLLAMA_KV_CACHE_TYPE=q8_0   # or q4_0 for extra savings, but far more unstable
   ```  
3. Restart Ollama.

#### **Linux/macOS**

```bash
export OLLAMA_FLASH_ATTENTION=1
export OLLAMA_KV_CACHE_TYPE="q8_0"   # or "q4_0", but far more unstable
ollama serve
```

---

## 📌 Acknowledgments

<details>
<summary>Click to expand</summary>

- **Data & Model by:** @Sweaterdog  
- **Framework:** Mindcraft (https://github.com/kolbytn/mindcraft)  
- **LoRA Weights:** https://huggingface.co/Sweaterdog/Andy-4-micro-LoRA

</details>

---

## ⚖️ License

See [Andy 1.0 License](LICENSE).  

*This work uses data and models created by @Sweaterdog.*