---
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 RL

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 SFT cold start, see [`Xkev/mc_single_sft`](https://huggingface.co/Xkev/mc_single_sft).

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

## Training

- **Initialization**: [`Xkev/mc_single_sft`](https://huggingface.co/Xkev/mc_single_sft)
- **Dataset**: Minecraft iron-pickaxe tasks
- **Framework**: [verl](https://github.com/volcengine/verl)
- **Algorithm**: monolithic REINFORCE
- **Hyperparameters**: `lr=1e-6`, `lr_warmup_ratio=0`, constant schedule

## Intended use

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

## License

Apache-2.0.