Instructions to use cstr/llama3-8b-spaetzle-v33 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cstr/llama3-8b-spaetzle-v33 with Transformers:

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

pipe = pipeline("text-generation", model="cstr/llama3-8b-spaetzle-v33")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cstr/llama3-8b-spaetzle-v33")
model = AutoModelForCausalLM.from_pretrained("cstr/llama3-8b-spaetzle-v33", 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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cstr/llama3-8b-spaetzle-v33 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cstr/llama3-8b-spaetzle-v33"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cstr/llama3-8b-spaetzle-v33",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/cstr/llama3-8b-spaetzle-v33

SGLang

How to use cstr/llama3-8b-spaetzle-v33 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 "cstr/llama3-8b-spaetzle-v33" \
    --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": "cstr/llama3-8b-spaetzle-v33",
		"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 "cstr/llama3-8b-spaetzle-v33" \
        --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": "cstr/llama3-8b-spaetzle-v33",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use cstr/llama3-8b-spaetzle-v33 with Docker Model Runner:
```
docker model run hf.co/cstr/llama3-8b-spaetzle-v33
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

llama3-8b-spaetzle-v33

This is a merge of the following models:

It attempts a compromise in usefulness for German and English tasks.

For GGUF quants see cstr/llama3-8b-spaetzle-v33-GGUF,

Benchmarks

It achieves on EQ-Bench v2_de as q4km (old version without pre-tokenizer-fix) quants 66.59 (171 of 171 parseable) and 73.17 on v2 (english) (171/171).

For the int4-inc quants:

Benchmark	Score
Average	66.13
ARC-c	59.81
ARC-e	85.27
Boolq	84.10
HellaSwag	62.47
Lambada	73.28
MMLU	64.11
OpenbookQA	37.2
Piqa	80.30
TruthfulQA	50.21
Winogrande	73.72

Nous

Model	Average	AGIEval	GPT4All	TruthfulQA	Bigbench
mlabonne/Daredevil-8B 📄	55.87	44.13	73.52	59.05	46.77
cstr/llama3-8b-spaetzle-v33 📄	55.26	42.61	73.9	59.28	45.25
mlabonne/Daredevil-8B-abliterated 📄	55.06	43.29	73.33	57.47	46.17
NousResearch/Hermes-2-Theta-Llama-3-8B 📄	54.28	43.9	72.62	56.36	44.23
openchat/openchat-3.6-8b-20240522 📄	53.49	44.03	73.67	49.78	46.48
mlabonne/Llama-3-8B-Instruct-abliterated-dpomix 📄	52.26	41.6	69.95	54.22	43.26
meta-llama/Meta-Llama-3-8B-Instruct 📄	51.34	41.22	69.86	51.65	42.64
failspy/Meta-Llama-3-8B-Instruct-abliterated-v3 📄	51.21	40.23	69.5	52.44	42.69
mlabonne/OrpoLlama-3-8B 📄	48.63	34.17	70.59	52.39	37.36
meta-llama/Meta-Llama-3-8B 📄	45.42	31.1	69.95	43.91	36.7

🧩 Configuration

models:
  - model: cstr/llama3-8b-spaetzle-v20
    # no parameters necessary for base model
  - model: cstr/llama3-8b-spaetzle-v31
    parameters:
      density: 0.65
      weight: 0.25
  - model: cstr/llama3-8b-spaetzle-v28
    parameters:
      density: 0.65
      weight: 0.25
  - model: cstr/llama3-8b-spaetzle-v26
    parameters:
      density: 0.65
      weight: 0.15
merge_method: dare_ties
base_model: cstr/llama3-8b-spaetzle-v20
parameters:
  int8_mask: true
dtype: bfloat16
random_seed: 0
tokenizer_source: base

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "cstr/llama3-8b-spaetzle-v33"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])

EU AI Act Art. 53 — provider obligations

Added 2026-08-02 during an account-wide provenance review.

This is a model merge, not a format conversion. Most cstr/* repositories are GGUF conversions, where the upstream research team remains the provider of the model and the conversion changes only the numeric representation of the weights. A merge produces a model that did not previously exist, so under Regulation (EU) 2024/1689 the maintainer of this repository is plausibly the provider of it, and the duties that survive the Art. 53(2) free-and-open-source exemption — Art. 53(1)(c) and 53(1)(d) — attach here rather than upstream.

Art. 53(1)(c) — copyright policy. No training corpus was assembled by this repository. Merging combines weights that other providers already published; it performs no text or data mining, so no rights reservation under Art. 4(3) of Directive (EU) 2019/790 was engaged by this step. Copyright questions arising from how the constituent models were themselves trained attach to their respective providers. Any credible claim that this repository redistributes material it has no right to redistribute will be acted on — contact via the Community tab.

Art. 53(1)(d) — training content. No data was used to train this model: it is a weight-space combination of models trained by others, and its training content is theirs. Of the 4 constituent models this card names, 2 are still published and 2 are not: cstr/llama3-8b-spaetzle-v28, cstr/llama3-8b-spaetzle-v26. For those, the training-content chain cannot be followed from this card, and no summary is reconstructed here in their place — an untraceable summary presented as a traceable one would be worse than the gap.