README.md

---
license: apache-2.0
datasets:
- Sweaterdog/Andy-3.5
language:
- en
base_model:
- deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
- deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
tags:
- Minecraft
- MindCraft
---

# 🚀 Welcome to a new generation of Minecraft with Andy 3.5 🚀

## Andy 3.5 is a collection of LOCAL LLM's designed for playing Minecraft
*Andy 3.5 is designed to be used with MindCraft, and is not designed nor intended to be used for any other applications*

*Also note Andy-3.5 has a newer version, Andy-3.6 which can be found [here](https://huggingface.co/Sweaterdog/Andy-3.6)*


>  # Please note! [!WARNING]
> 
> Andy-3.5 was trained on older data, and not the newest and latest versions of Mindcraft.
> 
> I **cannot** guarantee that Andy-3.5 will work on future versions as the model was tuned to play MindCraft with a specific version!
> 
> For the rest of the Andy-3.5 generation, this model will **ONLY** be supported on the version of Mindcraft in [this github repo!](https://github.com/Sweaterdog/Mindcraft-for-Andy-3.5)
>
> For more info, as well as the supported version of Mindcraft, please follow [this link to github](https://github.com/Sweaterdog/Mindcraft-for-Andy-3.5)


## Andy-3.5 is a great model, but wanted an updated version? Capable of *more?*
Andy-3.6 is an updated version of Andy-3.5, trained with more Epochs, and a bigger, and better dataset.

Andy-3.6 was trained on **4 epochs**, instead of one *(35,853 steps over 4,000)*, trained with a dataset of **24,000** examples instead of 11,000

Andy-3.6 also has a new feature, **case-by-case reasoning**, this means Andy-3.6 can reason when it deems a task needing of it, Andy-3.6 can be prompted to always reason

Why wait? Do **you** want a better model? you can find Andy-3.6 [Here](https://huggingface.co/Sweaterdog/Andy-3.6)

# How to Install / Setup

1. Select the model you would like to use *(The regular model, as well as the small model is recommended)*
2. Download the Modelfile
3. Once downloaded, open Modelfile in a text editor, and change the path to the download location of the gguf file
4. When changed, save the file, and open command terminal
5. *(Optional if CMD isn't opened via file explorer)* Navigate to the correct directory using "cd"
6. Run the command ```ollama create sweaterdog/Andy-3.5 -f Modelfile``` If you want multiple models, include a tag afterwards. Example: sweaterdog/Andy-3.5:mini-fp16 or sweaterdog/Andy-3.5:q2_k
7. Go to a profile in MindCraft
8. Change the model to be ```sweaterdog/Andy-3.5``` *Or whatever you named your model*
9. Ensure you have the emdedding tag set to Ollama, like below
```
{
    "name": "andy-3.5",

    "model": "Sweaterdog/Andy-3.5",

    "embedding": "ollama"

}
```

10. Enjoy playing with an AI that you are hosting!

> # Ollama Support [!NOTE]
> On Huggingface, there is an option to download GGUF models via Ollama
>
> However, this method **DOES NOT WORK** for models other than the base model of Andy-3.5!


# How was model trained?

The model was trained on the [MindCraft dataset](https://huggingface.co/datasets/Sweaterdog/Andy-3.5) for Andy-3.5, which includes ~12,000 prompts, featuring all things Minecraft.

# What are capabilities and Limitations?

Andy-3.5 was trained on EVERYTHING regarding Minecraft and MindCraft, it knows how to use commands natively without a system prompt.
Andy-3.5 also knows how to build / use !newAction to perform commands, it was trained on lots of building, as well as, using !newAction to do tasks like manually making something or strip mining.

# What models can I choose?

There are going to be 3 odel sizes avaliable, Regular, Small, and Mini
* Regular is a 7B parameter model, tuned from [Deepseek-R1 Distilled](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B)
* Small is a 3B parameter model, tuned from [Qwen2.5 3B](Qwen/Qwen2.5-3B-Instruct)
* Mini is a 1.5B parameter model, also tuned from [Deepseek-R1 Distilled](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B)

Small has a dedicated **"reasoning"** version released, Regular won't have a reasoning tune, Andy-3.6 will have built in case-by-case reasoning.

Out of all of the models, Teensy had the largest percent of parameters tuned, being 1/2 the models total size

# Safety and FAQ

Q: Is this model safe to use?

A. Yes, this model is non-volatile, and cannot generate malicous content

Q. Can this model be used on a server?

A. Yes, In theory and practice the model is only capable of building and performing manual tasks via newAction

Q. Who is responsible if this model does generate malicous content?

A. You are responsible, even though the model was never trained to be able to make malicous content, there is a ***very very slight chance*** it still generates malicous code.

Q. If I make media based on this model, like photos / videos, do I have to mention the Creator?

A. No, if you are making a post about MindCraft, and using this model, you only have to mention the creator if you mention the model being used.

# 🔥UPDATE🔥

**All models have their own folder, besides the main version of Andy-3.5**

To find models such as reasoning or mini, go into files and search inside the folder

**There is an Andy-3.5-reasoning-preview model, designed to demonstrate reasoning abilities in small language models to improve Minecraft skills**

Remember that this is a ***preview*** model and is **not** guaranteed to work, nor perform better or the same as Andy-3.5-*(Base)* 

When the full Andy-3.5-reasoning model is released, there will be the regular 7B model, as well as the small model, which is 3B parameters.

For future updates and generations there will **not** be a mini and a teensy version, of course the name may stay, but there wil **not** be a 1.5B **nor** a 360M model

> # I want to thank all supporters! [!NOTE]
> I would love to thank everyone who supported this project, there is a list of supporters in the files section.
> 
> You can find all of the supporters [here](https://huggingface.co/Sweaterdog/Andy-3.5/blob/main/Supporters.txt)

# Performance Metrics

These benchmarks are a-typical, since most standard benchmarks don't apply to Minecraft

The benchmarks below include models via API that are cheap, and other fine-tuned local models *(Excluding Andy-v2 and Andy-v3, since they are poor in quality)*

## Zero info Prompting
*How fast can a model collect 16 oak logs, and convert them all into sticks*

![image/png](https://cdn-uploads.huggingface.co/production/uploads/66960602f0ffd8e3a381106a/IEw1Gydg943qVSNGAL3RW.png)


Currently, Andy-3.5, Andy-3.5-small, and Andy-3.5-mini are the **ONLY** models that can play without command documentation, or any other instruction, and Andy-3.5-Mini *sometimes* fares better ***without*** the unnecessary data.
Test this for yourself using [this profile](https://huggingface.co/Sweaterdog/Andy-3.5/blob/main/local_demo.json)


## Time to get a stone pickaxe

![image/png](https://cdn-uploads.huggingface.co/production/uploads/66960602f0ffd8e3a381106a/frrT9IcJsNeUOLhszFrOq.png)

I am sure other models like Deepseek-R1 may be faster at getting a stone pickaxe, however the Demo was to show the performance of Andy-3.5

*For Andy-3.5-mini, I used the FP16 model, I had enough VRAM to do so*

*For Andy-3.5, I used the Q4_K_M quantization*

*For Andy-3.5-small, I used the Q8_0 quantization*

*Andy-3.5-reasoning-small was able to be the most efficient model producing the lowest amount of messages, but took a whopping 34.5 minutes to get a stone pickaxe.*

*For Andy-3.5-Teensy, I used the FP16 quantization*

*For Mineslayerv1 and Mineslayerv2, I used the default (and only) quantization, Q4_K_M*

## Notes about the benchmarks

**Zero Info Prompting**

Andy-3.5-Teensy was able to use one command successfully, but was not able to afterwards

Andy-3.5-Mini collected 32 oak_log instead of 16 oak_log

Andy-3.5-small *No notes*

Andy-3.5 attempted to continue playing, and make a wooden_pickaxe after the goal was done.

Both Mineslayerv1 and Mineslayerv2 hallucinated commands, like !chop or !grab

**Time to get a stone pickaxe**

Andy-3.5-teensy hallucinates too much for stable gameplay *(It is a 360M parameter model, what can be expected)*

Andy-3.5-Mini was unable to make itself a stone pickaxe, however it collected enough wood, but then got stuck on converting logs to planks, it kept trying "!craftRecipe("wooden_planks", 6) instead of oak_planks

Andy-3.5-small kept trying to make a stone_pickaxe first

Andy-3.5 Made a stone pickaxe the fastest out of all models, including GPT-4o-mini and Claude-3.5-Haiku

Mineslayerv1 Was unable to use !collectBlocks, instead kept trying !collectBlock

Mineslayerv2 Was unable to play, it kept hallucinating on the first command