---
license: apache-2.0
base_model:
- Qwen/Qwen3-VL-4B-Instruct
language:
- en
pipeline_tag: image-text-to-text
library_name: transformers
tags:
  - minecraft
  - qwen3-vl
---


# Qwen3-VL-4B-Instruct — Minecraft Single-Agent SFT

Paper: [Learning Self-Orchestration in Decentralized Agent Swarms](https://arxiv.org/abs/2610.22322)

Code: [https://github.com/Embodied-Minds-Lab/MPT](https://github.com/Embodied-Minds-Lab/MPT)

For the post-trained model initialized from this checkpoint, see [`Xkev/mc_single`](https://huggingface.co/Xkev/mc_single).

For the three-agent counterpart, see [`Xkev/mc_society_sft`](https://huggingface.co/Xkev/mc_society_sft).

## Training

- **Base model**: [`Qwen/Qwen3-VL-4B-Instruct`](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct)
- **Dataset**: Minecraft iron-pickaxe tasks
- **Framework**: [transformers](https://github.com/huggingface/transformers) `Trainer`
- **Hyperparameters**: `lr=1e-5`, `lr_warmup_ratio=0.03`, cosine schedule, `epochs=2`

## Intended use

Research on multi-agent post-training and embodied agents. Not intended for general dialog or production.

## License

Apache-2.0.