Instructions to use Steelskull/MSM-MS-Cydrion-22B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Steelskull/MSM-MS-Cydrion-22B with Transformers:

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

pipe = pipeline("text-generation", model="Steelskull/MSM-MS-Cydrion-22B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Steelskull/MSM-MS-Cydrion-22B")
model = AutoModelForCausalLM.from_pretrained("Steelskull/MSM-MS-Cydrion-22B", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Steelskull/MSM-MS-Cydrion-22B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Steelskull/MSM-MS-Cydrion-22B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Steelskull/MSM-MS-Cydrion-22B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Steelskull/MSM-MS-Cydrion-22B

SGLang

How to use Steelskull/MSM-MS-Cydrion-22B 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 "Steelskull/MSM-MS-Cydrion-22B" \
    --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": "Steelskull/MSM-MS-Cydrion-22B",
		"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 "Steelskull/MSM-MS-Cydrion-22B" \
        --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": "Steelskull/MSM-MS-Cydrion-22B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Steelskull/MSM-MS-Cydrion-22B with Docker Model Runner:
```
docker model run hf.co/Steelskull/MSM-MS-Cydrion-22B
```

Feedback

by DazzlingXeno - opened Oct 24, 2024

Discussion

DazzlingXeno

Oct 24, 2024

•

edited Oct 24, 2024

Tested this today with some settings I saw on Reddit - 0.3 temp and 0.3 MinP, using NovelCrafter and it's really, really good. Really creative and follows instructions really well. It's a little wild with RP, likes to witter on but that's not really what I'm using it for.

Anyway, really nice and my favourite Mistral Small model.

Thanks

Steelskull

Owner Oct 25, 2024

•

edited Oct 25, 2024

Tested this today with some settings I saw on Reddit - 0.3 temp and 0.3 MinP, using NovelCrafter and it's really, really good. Really creative and follows instructions really well. It's a little wild with RP, likes to witter on but that's not really what I'm using it for.

Anyway, really nice and my favorite Mistral Small model.

Thanks

I appreciate the review! ill test the recommended settings as well as they are much different than i use.

and I design most of my merges to go for more of a narrative / story style (My personal favorite) than a pure RP style so im happy to see someone else use it for this purpose!

Steelskull changed discussion status to closed Oct 25, 2024

Steelskull changed discussion status to open Oct 25, 2024

smhf72

Nov 14, 2024

This turned out to be one of my favorite models so far. Much more cohesive, creative, and consistent than other 22B MSM models.

Thanks!

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