Instructions to use Gensyn/open-1b-sft with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Gensyn/open-1b-sft with Transformers:

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

pipe = pipeline("text-generation", model="Gensyn/open-1b-sft", 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("Gensyn/open-1b-sft", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Gensyn/open-1b-sft with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Gensyn/open-1b-sft"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Gensyn/open-1b-sft",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Gensyn/open-1b-sft

SGLang

How to use Gensyn/open-1b-sft 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 "Gensyn/open-1b-sft" \
    --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": "Gensyn/open-1b-sft",
		"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 "Gensyn/open-1b-sft" \
        --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": "Gensyn/open-1b-sft",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Gensyn/open-1b-sft with Docker Model Runner:
```
docker model run hf.co/Gensyn/open-1b-sft
```

Question about parameter count 1B

by pengwenzhi - opened Sep 16

Discussion

pengwenzhi

Sep 16

Hi! I noticed that the model is named open-1b-sft, but Hugging Face shows 2B parameters. Could you clarify the actual parameter count and why it differs from the model name?

jbonneville

Gensyn org Sep 16

Hey @pengwenzhi good question! open-1b is 1.61B total parameters -- 1.08B non-embedding, which is the figure in the model card farther down.

The 2B parameters in the sidebar is HF's rounded display # of the count it reads from the Safetensors header, so it is rounding 1.61b -> 2b. The 1B in the name itself is the size class which matches the non-embedding count of 1.08B. Other models adopt this convention too -- for example Llama 3.2 1B is actually 1.24B (rounds down).

You can check this in a couple ways: on the file tab you'll see the model.safetensors file is 6.43gb which, stored at F32, is 1.61B parameters. You can also use a curl command on HF's API to see how it computed the number from the header:

curl -s "https://huggingface.co/api/models/Gensyn/open-1b-sft?expand=safetensors" | jq .safetensors

Which returns the parameters object + a 'total'. That total is what gets rounded up to the "2B Params" in that pill you're looking at.

pengwenzhi

Sep 17

got it !

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment