Skip to content
Other

Le Dog

AI models for smart wolves in Minecraft

2 min readupdated

A package of two small ONNX models that replace standard Minecraft wolf behavior with realistic dog habits. The models run directly inside the game via ONNX Runtime and manage entity decisions based on temperament and surroundings. The main model was created using distillation (knowledge transfer) from the Claude Sonnet large language model. This development will interest players using the Le Dog mod and anyone curious about lightweight AI model inference in Java.

Key features

  • Two included models: a core brain for action planning and a separate module for executing commands.
  • High speed: the main model takes about 1 ms per decision on a CPU.
  • Temperament tracking: wolf behavior depends on its personality (curious, brave, energetic, social, playful).
  • Reinforcement learning: the trick module was trained using PPO (Proximal Policy Optimization, a reinforcement learning method) inside actual Minecraft.

What you can do

  • Install in the game: use the models alongside the Le Dog mod for Minecraft version 26.3.
  • Train virtual pets: give wolves commands like sit, stay, wait, and come at distances up to 40 blocks.
  • Explore ONNX integration: see how lightweight AI models run inside Java applications.

Example

The models download automatically through the mod using the /ledog update command, and input and output formats are described in the mod source code within the StudentSource.java and TrickBrain.java files.

Read moreDetails
Size of student.onnx: 4.8 MB (1.18 million parameters), rl_tricks.onnx: 27 KB.
Performance: median speed of the core brain is under 1.1 ms per decision on a CPU.
Limitations: the models were trained and tested exclusively in Minecraft on symbolic observations and are not intended for real-world robots.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # KuanKuanKuan/le-dog
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download KuanKuanKuan/le-dog --local-dir ~/models/le-dog
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to le-dog
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of le-dog (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

KuanKuanKuan/le-dog
Author
KuanKuanKuan
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments