Instructions to use matchaaaaa/Fimbul-Airo-18B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use matchaaaaa/Fimbul-Airo-18B with Transformers:

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

pipe = pipeline("text-generation", model="matchaaaaa/Fimbul-Airo-18B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("matchaaaaa/Fimbul-Airo-18B")
model = AutoModelForCausalLM.from_pretrained("matchaaaaa/Fimbul-Airo-18B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use matchaaaaa/Fimbul-Airo-18B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "matchaaaaa/Fimbul-Airo-18B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "matchaaaaa/Fimbul-Airo-18B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/matchaaaaa/Fimbul-Airo-18B

SGLang

How to use matchaaaaa/Fimbul-Airo-18B 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 "matchaaaaa/Fimbul-Airo-18B" \
    --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": "matchaaaaa/Fimbul-Airo-18B",
		"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 "matchaaaaa/Fimbul-Airo-18B" \
        --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": "matchaaaaa/Fimbul-Airo-18B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use matchaaaaa/Fimbul-Airo-18B with Docker Model Runner:
```
docker model run hf.co/matchaaaaa/Fimbul-Airo-18B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Fimbul-Airo-18B 👍

Thanks to @mradermacher for his excellent quants! You can find his GGUFs for this repo here.

This is a merge of pre-trained language models created using mergekit. 👍

I tested it for thirtneen.second 👍

Works pretty good. Also seems to be happy when ROPEing up to 8K. Uncensored, told me how to build a nuke.

Merge Details

Merge Method

This model was merged using the passthrough merge method. Taking models and smashing em all together 👍

Models Merged

The following models were included in the merge:

Sao10K/Fimbulvetr-11B-v2 👍
Undi95/Mistral-11B-CC-Air-RP 👍
- CollectiveCognition-v1.1-Mistral-7B
- airoboros-mistral2.2-7b
- PIPPA dataset 11B qlora
- LimaRPv3 dataset 11B qlora

The Sauce

The following YAML configuration was used to produce this model:

slices:
  - sources:
    - model: Sao10K/Fimbulvetr-11B-v2
      layer_range: [0, 40]
  - sources:
    - model: Undi95/Mistral-11B-CC-Air-RP
      layer_range: [8, 48]
merge_method: passthrough
dtype: bfloat16

👍

Prompt Format: Alpaca 👍

### Instruction:
<Prompt>

### Input:
<Insert Context Here>

### Response:

👍

Less silly models are in the works. I'm still figuring things out right now, so don't judge the bazillions of readme edits and other goofiness.

Don't forget to take care of yourself and have a wonderful day!

Downloads last month: 2

Safetensors

Model size

18B params

Tensor type

BF16

Model tree for matchaaaaa/Fimbul-Airo-18B

Sao10K/Fimbulvetr-11B-v2

Undi95/Mistral-11B-CC-Air-RP

Merge model

this model

Quantizations

2 models