Instructions to use trollek/Holger-7B-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trollek/Holger-7B-v0.1 with Transformers:

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

pipe = pipeline("text-generation", model="trollek/Holger-7B-v0.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("trollek/Holger-7B-v0.1")
model = AutoModelForCausalLM.from_pretrained("trollek/Holger-7B-v0.1")
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

vLLM

How to use trollek/Holger-7B-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "trollek/Holger-7B-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trollek/Holger-7B-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/trollek/Holger-7B-v0.1

SGLang

How to use trollek/Holger-7B-v0.1 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 "trollek/Holger-7B-v0.1" \
    --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": "trollek/Holger-7B-v0.1",
		"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 "trollek/Holger-7B-v0.1" \
        --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": "trollek/Holger-7B-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use trollek/Holger-7B-v0.1 with Docker Model Runner:
```
docker model run hf.co/trollek/Holger-7B-v0.1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Du har fundet Holger!

Det er forrykt at danske open source sprogmodeller halter lidt; æøå be damned! Og nok også andre ting, men [A-ZÆØÅa-zæøå] giver kronisk mentaleksem. Anyway!

HOLGER! En fintunet åben Viking på åbne data under et åbent licens. Respekt for licenser er sådan kapitalismen er, og GUDS NÅDE fucking trøste dig hvis du ikke udgiver Llama modeller uden også at bøje knæet for Zuck. Redemption arc eller ej. 10% af 1 fantasiliard er stadig mere end jeg kan prostiturere mig selv for.

Sprogmodellens beskrivelse

Hva' lav' do nu? - Det vil jeg helst ikke snakke om. - Kom nu! - Jeg træner og snakker med sprogmodeller.. - Fra udlandet? - Jaaaa...

Viking modellen kan åbenbart finde ud af at overføre træning fra engelsk til dansk, og sandsynligvis de andre skandisprog. Maybe.

Kvanter

trollek/Holger-7B-v0.1-GGUF

Merge Detaljer

Merge Metode

Denne model er merged med della_linear metoden og med LumiOpen/Viking-7B som base.

Modeller Merged

Følgende modeller er brugt i dette merge:

viking/merges/holger5
viking/merges/holger3
mpasila/Viking-Magnum-v0.1-7B
mpasila/Viking-SlimSonnet-v1-7B

Konfiguration

Følgende YAML konfigurationsfil blev brugt til at skabe dette merge:

models:
  - model: viking/merges/holger3
    parameters:
      weight: 0.53
      density: 0.55
      epsilon: 0.11
  - model: viking/merges/holger5
    parameters:
      weight: 0.78
      density: 0.81
      epsilon: 0.17
  - model: mpasila/Viking-Magnum-v0.1-7B
    parameters:
      weight: 0.88
      density: 0.91
      epsilon: 0.07
  - model: mpasila/Viking-SlimSonnet-v1-7B
    parameters:
      weight: 0.83
      density: 0.75
      epsilon: 0.11
merge_method: della_linear
base_model: LumiOpen/Viking-7B
parameters:
  normalize: true
  int8_mask: true
  lambda: 1.08
dtype: bfloat16