Instructions to use sercand/SmolLM2-135M-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sercand/SmolLM2-135M-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="sercand/SmolLM2-135M-Instruct")
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("sercand/SmolLM2-135M-Instruct")
model = AutoModelForCausalLM.from_pretrained("sercand/SmolLM2-135M-Instruct", 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 sercand/SmolLM2-135M-Instruct 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 sercand/SmolLM2-135M-Instruct:Q8_0
# Run inference directly in the terminal:
llama cli -hf sercand/SmolLM2-135M-Instruct:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf sercand/SmolLM2-135M-Instruct:Q8_0
# Run inference directly in the terminal:
llama cli -hf sercand/SmolLM2-135M-Instruct: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 sercand/SmolLM2-135M-Instruct:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf sercand/SmolLM2-135M-Instruct: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 sercand/SmolLM2-135M-Instruct:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf sercand/SmolLM2-135M-Instruct:Q8_0

Use Docker

docker model run hf.co/sercand/SmolLM2-135M-Instruct:Q8_0

LM Studio
Jan

vLLM

How to use sercand/SmolLM2-135M-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sercand/SmolLM2-135M-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sercand/SmolLM2-135M-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/sercand/SmolLM2-135M-Instruct:Q8_0

SGLang

How to use sercand/SmolLM2-135M-Instruct 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 "sercand/SmolLM2-135M-Instruct" \
    --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": "sercand/SmolLM2-135M-Instruct",
		"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 "sercand/SmolLM2-135M-Instruct" \
        --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": "sercand/SmolLM2-135M-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use sercand/SmolLM2-135M-Instruct with Ollama:
```
ollama run hf.co/sercand/SmolLM2-135M-Instruct:Q8_0
```
Unsloth Desktop
Docker Model Runner
How to use sercand/SmolLM2-135M-Instruct with Docker Model Runner:
```
docker model run hf.co/sercand/SmolLM2-135M-Instruct:Q8_0
```

Lemonade

How to use sercand/SmolLM2-135M-Instruct with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull sercand/SmolLM2-135M-Instruct:Q8_0

Run and chat with the model

lemonade run user.SmolLM2-135M-Instruct-Q8_0

List all available models

lemonade list

Atomic Chat

SmolLM2-135M-Instruct (decision model)

Fine-tuned from unsloth/SmolLM2-135M-Instruct (QLoRA, 3 epochs) with a Clef-style joint decision head (held-out accuracy 0.8715, ECE 0.0284; see unsloth_decision_config.json).

model.safetensors, joint_head.safetensors, joint_head_config.json: full bf16 checkpoint.
SmolLM2-135M-Instruct-clef-Q8_0.gguf: backbone and head in one GGUF, architecture name clef, produced by convert_clef_llama.py.
- Backbone is a stock Llama (q/k kept in the half-split rotary layout (not llama.cpp's permuted one)). Tied embeddings: there is no output.weight, use token_embd.weight.
- Large linears are Q8_0. Norms, biases, token_types, decision.scorer_out and decision.scales stay F32.
- Head tensors use the dec.blk.N / decision.* layout of llama.cpp's conversion/clef.py.
- Not loadable by llama.cpp: its clef arch assumes a Qwen3.5 backbone. Intended for the shiso engine.

Downloads last month: 41

Safetensors

Model size

0.1B params

Tensor type

BF16

Model tree for sercand/SmolLM2-135M-Instruct

Base model

HuggingFaceTB/SmolLM2-135M

Quantized

HuggingFaceTB/SmolLM2-135M-Instruct

Finetuned

unsloth/SmolLM2-135M-Instruct

Quantized

(6)

this model