Instructions to use StoyanGanchev/gemma-2-9b-nf4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use StoyanGanchev/gemma-2-9b-nf4 with Transformers:

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

pipe = pipeline("text-generation", model="StoyanGanchev/gemma-2-9b-nf4")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("StoyanGanchev/gemma-2-9b-nf4")
model = AutoModelForCausalLM.from_pretrained("StoyanGanchev/gemma-2-9b-nf4")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use StoyanGanchev/gemma-2-9b-nf4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "StoyanGanchev/gemma-2-9b-nf4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "StoyanGanchev/gemma-2-9b-nf4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/StoyanGanchev/gemma-2-9b-nf4

SGLang

How to use StoyanGanchev/gemma-2-9b-nf4 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 "StoyanGanchev/gemma-2-9b-nf4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "StoyanGanchev/gemma-2-9b-nf4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "StoyanGanchev/gemma-2-9b-nf4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "StoyanGanchev/gemma-2-9b-nf4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use StoyanGanchev/gemma-2-9b-nf4 with Docker Model Runner:
```
docker model run hf.co/StoyanGanchev/gemma-2-9b-nf4
```

Gemma-2-9B quantized in NormalFloat4

Description

This repository contains a 4-bit quantized version of the Gemma-2-9B model. Designed to minimize memory consumption and speed up inference.

Benchmark results

Gemma-2-9b	Wiki	C4	PIQA	ARC-E	ARC-C	HellaSwag	Wino	Avg.
	0-shot	0-shot	0-shot	0-shot	25-shot	0-shot	0-shot
Unquantized	6.88	10.12	81.39	87.25	64.33	61.27	74.11	73.67
NF4	7.05	11.04	81.45	86.78	64.62	60.87	74.51	73.65

Benchmark scores are computed with lm-evaluation-harness.

How to use:

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("StoyanGanchev/gemma-2-9b-nf4")
model = AutoModelForCausalLM.from_pretrained("StoyanGanchev/gemma-2-9b-nf4")

All files could be accesed on repository

Downloads last month: 1

Safetensors

Model size

9B params

Tensor type

F32

F16

Collection including StoyanGanchev/gemma-2-9b-nf4

Gemma-2-quantized

Collection

Gemma-2-2B and Gemma-2-9B Quantized in low-bit • 4 items • Updated Aug 2, 2024