Instructions to use NagusameCS/minLillemus with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use NagusameCS/minLillemus with Transformers:

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

pipe = pipeline("text-generation", model="NagusameCS/minLillemus")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("NagusameCS/minLillemus")
model = AutoModelForCausalLM.from_pretrained("NagusameCS/minLillemus", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use NagusameCS/minLillemus with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf NagusameCS/minLillemus:Q8_0
# Run inference directly in the terminal:
llama cli -hf NagusameCS/minLillemus:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf NagusameCS/minLillemus:Q8_0
# Run inference directly in the terminal:
llama cli -hf NagusameCS/minLillemus:Q8_0

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf NagusameCS/minLillemus:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf NagusameCS/minLillemus:Q8_0

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf NagusameCS/minLillemus:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf NagusameCS/minLillemus:Q8_0

Use Docker

docker model run hf.co/NagusameCS/minLillemus:Q8_0

LM Studio
Jan

vLLM

How to use NagusameCS/minLillemus with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/NagusameCS/minLillemus:Q8_0

SGLang

How to use NagusameCS/minLillemus 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 "NagusameCS/minLillemus" \
    --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": "NagusameCS/minLillemus",
		"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 "NagusameCS/minLillemus" \
        --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": "NagusameCS/minLillemus",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use NagusameCS/minLillemus with Ollama:
```
ollama run hf.co/NagusameCS/minLillemus:Q8_0
```
Unsloth Desktop
Docker Model Runner
How to use NagusameCS/minLillemus with Docker Model Runner:
```
docker model run hf.co/NagusameCS/minLillemus:Q8_0
```

Lemonade

How to use NagusameCS/minLillemus with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull NagusameCS/minLillemus:Q8_0

Run and chat with the model

lemonade run user.minLillemus-Q8_0

List all available models

lemonade list

Atomic Chat

minLillemus: "my little mouse"

An experimental weight-surgery variant of SmolLM2-135M-Instruct, made with the HyperTensor CECI "graft" procedure. It is a research artifact: it does not outperform its base model.

What was changed

Only the layer-12 MLP was changed: W12 <- W12 + 0.5 * (W2 - W12) @ P (rank-230 projector from layer 12's q_proj). Created by scripts/graft_proof.py ("Splejsning").

The other 269 of 272 tensors are bit-identical to the base model. Weights are stored in float32. The tokenizer and chat template are the base model's.

Measured quality (audit, 2026-10-09)

Teacher-forced next-token loss on 30 web and 30 code documents (512 tokens each, data/corpus/{web,code}.dev.jsonl), compared with the base model:

text	base NLL	this model NLL	perplexity change	top-1 agreement with base	KL(base‖model)
web	2.865	2.897	+3.2%	89.7%	0.037
code	1.783	1.846	+6.4%	89.8%	0.068
all	2.324	2.371	+4.8%	89.8%	0.053

original HyperTensor test	base	this model
graft_proof 15-sentence PPL (mean of per-sentence PPL)	60.96	69.87
graft_benchmark MMLU-lite (50 items)	31/50	33/50 (+2 / -0 items, sign-test p=0.50)
graft_benchmark BoolQ (15 items)	6/15	5/15 (+0 / -1, p=1.00)

The earlier card claimed "29% PPL recovery". The "% recovery" figures compared this model with a copy of the base whose MLP had been zeroed. This model was never ablated, so they do not measure recovery. On real text the model is slightly worse than its base, and the small MMLU-lite/BoolQ differences are a few items out of 50/15 and are not statistically significant.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("NagusameCS/minLillemus")
model = AutoModelForCausalLM.from_pretrained("NagusameCS/minLillemus")
msgs = [{"role": "user", "content": "What is the capital of France?"}]
inputs = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True)
out = model.generate(**inputs, max_new_tokens=64, do_sample=False)
print(tok.decode(out[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Ollama

The repo includes minLillemus-Q8_0.gguf and a Modelfile:

hf download NagusameCS/minLillemus --local-dir minLillemus && cd minLillemus
ollama create minlillemus -f Modelfile && ollama run minlillemus
# or directly: ollama run hf.co/NagusameCS/minLillemus

Source: HyperTensor.

Downloads last month: 38

Safetensors

Model size

0.1B params

Tensor type

F32

Model tree for NagusameCS/minLillemus

Base model

HuggingFaceTB/SmolLM2-135M

Quantized

HuggingFaceTB/SmolLM2-135M-Instruct

Quantized

(143)

this model