Instructions to use INCModel4/granite-4.1-8b-W4A16-AutoRound with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use INCModel4/granite-4.1-8b-W4A16-AutoRound with Transformers:

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

pipe = pipeline("text-generation", model="INCModel4/granite-4.1-8b-W4A16-AutoRound")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("INCModel4/granite-4.1-8b-W4A16-AutoRound")
model = AutoModelForCausalLM.from_pretrained("INCModel4/granite-4.1-8b-W4A16-AutoRound", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use INCModel4/granite-4.1-8b-W4A16-AutoRound with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "INCModel4/granite-4.1-8b-W4A16-AutoRound"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "INCModel4/granite-4.1-8b-W4A16-AutoRound",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/INCModel4/granite-4.1-8b-W4A16-AutoRound

SGLang

How to use INCModel4/granite-4.1-8b-W4A16-AutoRound 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 "INCModel4/granite-4.1-8b-W4A16-AutoRound" \
    --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": "INCModel4/granite-4.1-8b-W4A16-AutoRound",
		"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 "INCModel4/granite-4.1-8b-W4A16-AutoRound" \
        --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": "INCModel4/granite-4.1-8b-W4A16-AutoRound",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use INCModel4/granite-4.1-8b-W4A16-AutoRound with Docker Model Runner:
```
docker model run hf.co/INCModel4/granite-4.1-8b-W4A16-AutoRound
```

Model Details

This model is an int4 (W4A16, group_size 128, symmetric) model of ibm-granite/granite-4.1-8b generated by intel/auto-round with 200 tuning iterations. Please follow the license of the original model.

Quantized (W4A16): attention and MLP projections
Kept in BF16: token embeddings, lm_head, normalization layers

How to Use

Environment

uv pip install vllm==0.33.0
uv pip install transformers==5.20.2

vLLM Usage

vllm serve INCModel4/granite-4.1-8b-W4A16-AutoRound \
    --kv-cache-dtype auto \
    --dtype bfloat16

Evaluation

Benchmark	BF16	W4A16	Recovery
PIQA	79.05	78.78	99.66%
HellaSwag	62.84	61.94	98.57%
MMLU	71.68	71.21	99.35%
GSM8K	92.49	91.28	98.69%
Average	76.52	75.80	99.07%

lm_eval

lm_eval --model local-completions \
  --model_args model=INCModel4/granite-4.1-8b-W4A16-AutoRound,tokenizer=ibm-granite/granite-4.1-8b,tokenizer_backend=huggingface,base_url=http://127.0.0.1:8000/v1/completions,tokenized_requests=True,num_concurrent=32,max_retries=50,timeout=1800,max_length=131073,max_gen_toks=2048,add_bos_token=False \
  --tasks gsm8k --batch_size 32 --output_path results/gsm8k --log_samples --seed 42 \
  --num_fewshot 5 --apply_chat_template --fewshot_as_multiturn --gen_kwargs max_gen_toks=2048

lm_eval --model local-completions \
  --model_args model=INCModel4/granite-4.1-8b-W4A16-AutoRound,tokenizer=ibm-granite/granite-4.1-8b,tokenizer_backend=huggingface,base_url=http://127.0.0.1:8000/v1/completions,tokenized_requests=True,num_concurrent=32,max_retries=50,timeout=1800,max_length=131073,max_gen_toks=2048,add_bos_token=False \
  --tasks piqa,mmlu,hellaswag --batch_size 32 --output_path results/piqa_mmlu_hellaswag --log_samples --seed 42

Generate the Model

auto-round \
    --model_name ibm-granite/granite-4.1-8b \
    --scheme W4A16 \
    --iters 200 \
    --nsamples 128 \
    --ignore_layers lm_head \
    --format auto_round \
    --output_dir granite-4.1-8b-W4A16-AutoRound \
    --seqlen 2048 \
    --group_size 128 \
    --disable_trust_remote_code

Ethical Considerations and Limitations

The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs. Therefore, before deploying any applications of the model, developers should perform safety testing.

Caveats and Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Here are a couple of useful links to learn more about Intel's AI software:

Disclaimer

The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.

Cite

@article{cheng2023optimize,
  title={Optimize weight rounding via signed gradient descent for the quantization of llms},
  author={Cheng, Wenhua and Zhang, Weiwei and Shen, Haihao and Cai, Yiyang and He, Xin and Lv, Kaokao and Liu, Yi},
  journal={arXiv preprint arXiv:2309.05516},
  year={2023}
}

arxiv github

Signed-off-by: Sun, Xuehao xuehao.sun@intel.com

Downloads last month: 300

Safetensors

Model size

1B params

Tensor type

I32

BF16

F16

Model tree for INCModel4/granite-4.1-8b-W4A16-AutoRound

Base model

ibm-granite/granite-4.1-8b

Quantized

(83)

this model

Paper for INCModel4/granite-4.1-8b-W4A16-AutoRound

Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs

Paper • 2309.05516 • Published Sep 11, 2023 • 15