Instructions to use fhamborg/phi-4-4bit-gptq with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fhamborg/phi-4-4bit-gptq with Transformers:

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

pipe = pipeline("text-generation", model="fhamborg/phi-4-4bit-gptq", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("fhamborg/phi-4-4bit-gptq", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("fhamborg/phi-4-4bit-gptq", trust_remote_code=True)
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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use fhamborg/phi-4-4bit-gptq with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/fhamborg/phi-4-4bit-gptq

SGLang

How to use fhamborg/phi-4-4bit-gptq 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 "fhamborg/phi-4-4bit-gptq" \
    --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": "fhamborg/phi-4-4bit-gptq",
		"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 "fhamborg/phi-4-4bit-gptq" \
        --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": "fhamborg/phi-4-4bit-gptq",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use fhamborg/phi-4-4bit-gptq with Docker Model Runner:
```
docker model run hf.co/fhamborg/phi-4-4bit-gptq
```

Phi-4 GPTQ (4-bit Quantized)

Model Description

This is a 4-bit GPTQ-quantized version of the Phi-4 transformer model, optimized for efficient inference while maintaining performance.

Base Model: Phi-4
Quantization: GPTQ (4-bit)
Format: safetensors
Tokenizer: Uses standard vocab.json and merges.txt

Intended Use

Fast inference with minimal VRAM usage
Deployment in resource-constrained environments
Optimized for low-latency text generation

Model Details

Attribute	Value
Model Name	Phi-4 GPTQ
Quantization	4-bit (GPTQ)
File Format	`.safetensors`
Tokenizer	`phi-4-tokenizer.json`
VRAM Usage	~X GB (depending on batch size)

Downloads last month: -

Safetensors

Model size

3B params

Tensor type

I32

F16

Model tree for fhamborg/phi-4-4bit-gptq

Base model

microsoft/phi-4

Quantized

(149)

this model