Instructions to use naos-ku/GraphTokenLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use naos-ku/GraphTokenLM with Transformers:

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

pipe = pipeline("text-generation", model="naos-ku/GraphTokenLM", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("naos-ku/GraphTokenLM", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use naos-ku/GraphTokenLM with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/naos-ku/GraphTokenLM

SGLang

How to use naos-ku/GraphTokenLM 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 "naos-ku/GraphTokenLM" \
    --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": "naos-ku/GraphTokenLM",
		"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 "naos-ku/GraphTokenLM" \
        --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": "naos-ku/GraphTokenLM",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use naos-ku/GraphTokenLM with Docker Model Runner:
```
docker model run hf.co/naos-ku/GraphTokenLM
```

GraphTokenLM

File size: 871 Bytes

a342f8b

{
  "architectures": [
    "GraphTokenLM"
  ],
  "base_model": "Qwen/Qwen3-4B-Base",
  "bos_token_id": 151643,
  "dtype": "float32",
  "enable_lora": false,
  "eos_token_id": 151643,
  "freeze_llm": true,
  "gnn_hidden": 64,
  "gnn_hidden_dim": 64,
  "gnn_out": 64,
  "gnn_out_dim": 64,
  "gnn_type": "GIN",
  "graph_pooling": [
    "mean"
  ],
  "hidden_size": 2560,
  "llm_name": "Qwen/Qwen3-4B-Base",
  "lpe_dim": 8,
  "model_type": "graph_token_lm",
  "node_feat_dim": 8,
  "node_pos_emb_dim": 8,
  "num_attention_heads": 32,
  "num_gnn_layers": 3,
  "num_graph_tokens": 4,
  "num_hidden_layers": 36,
  "num_max_nodes": 20,
  "num_proj_layers": 2,
  "pos_emb_dim": 8,
  "transformers_version": "4.57.1",
  "use_degree_emb": false,
  "vocab_size": 151936,
  "auto_map": {
    "AutoConfig": "glm.GraphTokenLMConfig",
    "AutoModelForCausalLM": "glm.GraphTokenLM"
  }
}