Instructions to use nbeerbower/SmolNemo-12B-FFT-experimental with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nbeerbower/SmolNemo-12B-FFT-experimental with Transformers:

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

pipe = pipeline("text-generation", model="nbeerbower/SmolNemo-12B-FFT-experimental")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nbeerbower/SmolNemo-12B-FFT-experimental")
model = AutoModelForCausalLM.from_pretrained("nbeerbower/SmolNemo-12B-FFT-experimental")
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 nbeerbower/SmolNemo-12B-FFT-experimental with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nbeerbower/SmolNemo-12B-FFT-experimental"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nbeerbower/SmolNemo-12B-FFT-experimental",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nbeerbower/SmolNemo-12B-FFT-experimental

SGLang

How to use nbeerbower/SmolNemo-12B-FFT-experimental 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 "nbeerbower/SmolNemo-12B-FFT-experimental" \
    --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": "nbeerbower/SmolNemo-12B-FFT-experimental",
		"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 "nbeerbower/SmolNemo-12B-FFT-experimental" \
        --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": "nbeerbower/SmolNemo-12B-FFT-experimental",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nbeerbower/SmolNemo-12B-FFT-experimental with Docker Model Runner:
```
docker model run hf.co/nbeerbower/SmolNemo-12B-FFT-experimental
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🧪 Just Another Model Experiment

This is one of many experimental iterations I'm sharing publicly while I mess around with training parameters and ideas. It's not a "real" release - just me being transparent about my learning process. Feel free to look under the hood, but don't expect anything production-ready!

SmolNemo-12B-FFT-experimental

Mahou-1.5-mistral-nemo-12B-lorablated finetuned on HuggingFaceTB/smoltalk.

This model has erratic behavior and poor performance

Method

SFT with 8x A100 for 0.1 epochs.

This was a full finetune. I think the issues with the model can be chalked up to conflicts with Mistral Instruct and ChatML.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	8.32
IFEval (0-Shot)	33.48
BBH (3-Shot)	6.54
MATH Lvl 5 (4-Shot)	0.23
GPQA (0-shot)	1.34
MuSR (0-shot)	5.92
MMLU-PRO (5-shot)	2.41

Downloads last month: 3

Safetensors

Model size

12B params

Tensor type

BF16

Model tree for nbeerbower/SmolNemo-12B-FFT-experimental

Base model

nbeerbower/Mahou-1.5-mistral-nemo-12B-lorablated

Finetuned

(14)

this model

Merges

1 model

Quantizations

1 model

Dataset used to train nbeerbower/SmolNemo-12B-FFT-experimental

Evaluation results

strict accuracy on IFEval (0-Shot)
Open LLM Leaderboard

33.480
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

6.540
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

0.230
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

1.340
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

5.920
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

2.410