Instructions to use twothsnd/ReTerm-9B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use twothsnd/ReTerm-9B with Transformers:

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

pipe = pipeline("image-text-to-text", model="twothsnd/ReTerm-9B")
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("twothsnd/ReTerm-9B")
model = AutoModelForMultimodalLM.from_pretrained("twothsnd/ReTerm-9B", 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 twothsnd/ReTerm-9B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "twothsnd/ReTerm-9B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "twothsnd/ReTerm-9B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/twothsnd/ReTerm-9B

SGLang

How to use twothsnd/ReTerm-9B 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 "twothsnd/ReTerm-9B" \
    --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": "twothsnd/ReTerm-9B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "twothsnd/ReTerm-9B" \
        --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": "twothsnd/ReTerm-9B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use twothsnd/ReTerm-9B with Docker Model Runner:
```
docker model run hf.co/twothsnd/ReTerm-9B
```

twothsnd commited on 17 days ago

Commit

7858be7

verified ·

1 Parent(s): 952c6ae

Refine model card language

Browse files

Files changed (1) hide show

README.md +3 -3

README.md CHANGED Viewed

@@ -28,12 +28,12 @@ In this example, a failed `import pystan` leads the GRPO-trained agent to repeat
 ## 🏆 Performance
-The final checkpoint completes **31/89 Terminal-Bench 2.0 tasks (34.8% Pass@1)** and reaches **58.4% Pass@1 on SWE-bench Verified** under the evaluation settings in our study. These are first-attempt scores, not results after a second recovery attempt. Comparisons require the same agent harness, task images, verifier, inference budget, and sampling configuration.
 ## 🔧 Using the model
-These are model weights in Hugging Face format, not a standalone terminal agent. A harness must supply the task instruction, a bash tool, command execution, observations, and the task environment. Use the Qwen3.5 chat template and matching tool-call protocol. The [training repository](https://github.com/twothsnd/ReTerm) describes the interaction and state-preservation interfaces used by ReTerm.
 ## ⚠️ Limitations
-The model can issue incorrect or destructive commands, modify files unexpectedly, and judge a task complete when it is not. Run it in an isolated environment with resource and permission limits. Benchmark performance does not guarantee correctness on new tasks or under a different harness.

 ## 🏆 Performance
+The final checkpoint completes **31/89 Terminal-Bench 2.0 tasks (34.8% Pass@1)** and reaches **58.4% Pass@1 on SWE-bench Verified** under the evaluation settings in our study. Comparisons use the same agent harness, task images, verifier, inference budget, and sampling configuration.
 ## 🔧 Using the model
+The Hugging Face weights run as a terminal agent when paired with a harness that supplies the task instruction, a bash tool, command execution, observations, and the task environment. Use the Qwen3.5 chat template and matching tool-call protocol. The [training repository](https://github.com/twothsnd/ReTerm) describes the interaction and state-preservation interfaces used by ReTerm.
 ## ⚠️ Limitations
+The model can issue incorrect or destructive commands, modify files unexpectedly, and declare completion prematurely. Run it in an isolated environment with resource and permission limits, and independently verify its work on new tasks.