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

Libraries

How to use radm/gemma-2-9b-it-SimPO-rudpo with Transformers:

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

pipe = pipeline("text-generation", model="radm/gemma-2-9b-it-SimPO-rudpo")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("radm/gemma-2-9b-it-SimPO-rudpo", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use radm/gemma-2-9b-it-SimPO-rudpo with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "radm/gemma-2-9b-it-SimPO-rudpo"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "radm/gemma-2-9b-it-SimPO-rudpo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/radm/gemma-2-9b-it-SimPO-rudpo

SGLang

How to use radm/gemma-2-9b-it-SimPO-rudpo 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 "radm/gemma-2-9b-it-SimPO-rudpo" \
    --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": "radm/gemma-2-9b-it-SimPO-rudpo",
		"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 "radm/gemma-2-9b-it-SimPO-rudpo" \
        --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": "radm/gemma-2-9b-it-SimPO-rudpo",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use radm/gemma-2-9b-it-SimPO-rudpo with Docker Model Runner:
```
docker model run hf.co/radm/gemma-2-9b-it-SimPO-rudpo
```

Configuration Parsing Warning:Config file config.json cannot be fetched (too big)

Configuration Parsing Warning:Config file tokenizer_config.json cannot be fetched (too big)

Gemma-2-9b-it-SimPO-rudpo

Model Information

Improved quality of Russian language compared to the base model. Tested on arena-hard questions in Russian

gemma-2-9b-it-SimPO-rudpo           | score: 91.9  | 95% CI:   (-0.9, 0.9)   | average #tokens: 1013   <---- THIS MODEL
gemma-2-9b-it-SimPO                 | score: 90.9  | 95% CI:   (-0.9, 1.1)   | average #tokens: 1065
gemma-2-27b-it-FP8                  | score: 82.0  | 95% CI:   (-1.3, 1.5)   | average #tokens: 799
gemma-2-9b-it                       | score: 67.0  | 95% CI:   (-1.8, 1.7)   | average #tokens: 760
gemma-2-2b-it-abl-rudpo             | score: 61.6  | 95% CI:   (-1.7, 2.2)   | average #tokens: 1121
gemma-2-2b-it-abl                   | score: 48.8  | 95% CI:   (-1.9, 2.0)   | average #tokens: 783
gemma-2b-it                         | score:  8.8  | 95% CI:   (-1.1, 1.0)   | average #tokens: 425