Instructions to use Sao10K/Fimbulvetr-11B-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Sao10K/Fimbulvetr-11B-v2 with Transformers:

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

pipe = pipeline("text-generation", model="Sao10K/Fimbulvetr-11B-v2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Sao10K/Fimbulvetr-11B-v2")
model = AutoModelForCausalLM.from_pretrained("Sao10K/Fimbulvetr-11B-v2", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Sao10K/Fimbulvetr-11B-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Sao10K/Fimbulvetr-11B-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Sao10K/Fimbulvetr-11B-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Sao10K/Fimbulvetr-11B-v2

SGLang

How to use Sao10K/Fimbulvetr-11B-v2 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 "Sao10K/Fimbulvetr-11B-v2" \
    --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": "Sao10K/Fimbulvetr-11B-v2",
		"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 "Sao10K/Fimbulvetr-11B-v2" \
        --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": "Sao10K/Fimbulvetr-11B-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Sao10K/Fimbulvetr-11B-v2 with Docker Model Runner:
```
docker model run hf.co/Sao10K/Fimbulvetr-11B-v2
```

Thanks for what you do.

by NickelnDime - opened Feb 7, 2024

Discussion

NickelnDime

Feb 7, 2024

I greatly enjoyed v1, so I'm looking forward to testing this out.
EMT Is stressful as hell but you're doing something incredibly valuable,
keep your head up.
Thanks. :)

Herman555

Feb 7, 2024

I also would like to thank you for what you are doing,
V1 is quite well regarded in the community.
Where I live the whole healthcare system is rotten to the core and it's purely business before people, personnel and patients treated like shit alike.
It's disgusting frankly.
Take care.

FrostyX10

Feb 7, 2024

Just made an account to thank you for your work.
i havent tried this one but i have tried frostwind and
did not like it for ERP purposes,
i think this one will be better for that purpose.
Thanks for your hard work.

avada-z

Feb 7, 2024

I'm saying thanks too, Fimbulvetr models are perfectly balanced.
Actually tested this model, it seems to follow the prompt much better, also feels more creative I guess.
Maybe even more intelligent, but that might be influenced by the changes to the sampling parameters.

TravelingMan

Feb 8, 2024

Fimbulvetr is so incredible. For its size, it follows instructions well and feels really authentic for RP (I don't ERP so unsure there). It's my favorite model, and has been since the original came out. Looking forward to trying this one!

Thanks, Sao10K!

Foxlum

Feb 8, 2024

This is one of the best in my opinion in comparison to some 13b, 10b, and 7b i've used, especially with its variety and quality of its total output. This one is one of the few human feeling models using character cards, able to easily handle details and consistency. I do hope the Fimbulvetr dense 11b series continues (or transitions a different similar sized base model), even if you're planning to go into MoEs or Yi.

WhensDinner

Feb 9, 2024

Jumping in here to hit you with some good vibes. Fimbulvetr has been my favorite RP model for a good while, even including larger ones. It strikes a good balance in a lot of ways. Looking forward to quants of the new version once you've got it ironed out a bit.

Sao10K

Owner Feb 10, 2024

Thanks for all the positive words, everyone :)

eevee32x

Feb 12, 2024

Half of my favorites are from you <3 This has good fluffy romance/ERP balance. PS: I love loveee Frostwind but I was too shy to open a discussion :C

Sao10K

Owner Feb 15, 2024

:) Thanks for all the words, it really does mean a lot to me.

ddbl

Mar 12, 2024

Just adding my voice to say thank you. This v2 model is incredible.

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