Instructions to use Rtian/MiRL-Qwen3.5-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Rtian/MiRL-Qwen3.5-4B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Rtian/MiRL-Qwen3.5-4B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Rtian/MiRL-Qwen3.5-4B")
model = AutoModelForMultimodalLM.from_pretrained("Rtian/MiRL-Qwen3.5-4B", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Rtian/MiRL-Qwen3.5-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Rtian/MiRL-Qwen3.5-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rtian/MiRL-Qwen3.5-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Rtian/MiRL-Qwen3.5-4B

SGLang

How to use Rtian/MiRL-Qwen3.5-4B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Rtian/MiRL-Qwen3.5-4B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rtian/MiRL-Qwen3.5-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Rtian/MiRL-Qwen3.5-4B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rtian/MiRL-Qwen3.5-4B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Rtian/MiRL-Qwen3.5-4B with Docker Model Runner:
```
docker model run hf.co/Rtian/MiRL-Qwen3.5-4B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

MiRL-Qwen3.5-4B

Model Checkpoint for paper "MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Agent Misbehaviors".

A terminal agent trained from Qwen/Qwen3.5-4B with reinforcement learning on 189 terminal tasks.

Terminal agents often repeat the same misbehaviors like making false claims about tests and repeating unnecessary checks. Misbehavior-driven RL (MiRL) first profiles the errors of a policy to find these misbehaviors. It then trains the policy with RL, starting rollouts at the steps where these misbehaviors occur and rewarding a rollout only if it passes the tests and avoids the misbehavior.

Usage

vllm serve Rtian/MiRL-Qwen3.5-4B --language-model-only --max-model-len 262144 \
  --enable-auto-tool-choice --tool-call-parser qwen3_xml

The repository holds the language model only (no vision or MTP weights), so --language-model-only is needed. The config, tokenizer and chat template are unchanged from Qwen3.5-4B.

For more details, see https://github.com/runchu-tian/MiRL.

Citation

@misc{tian2026mirlerrortargetedreinforcementlearning,
      title={MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Agent Misbehaviors},
      author={Runchu Tian and Zaid Khan and Zun Wang and Justin Chih-Yao Chen and Mohit Bansal},
      year={2026},
      eprint={2610.35697},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2610.35697},
}