Instructions to use basmala12/smollm_finetuning5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use basmala12/smollm_finetuning5 with Transformers:

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

pipe = pipeline("text-generation", model="basmala12/smollm_finetuning5")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("basmala12/smollm_finetuning5")
model = AutoModelForCausalLM.from_pretrained("basmala12/smollm_finetuning5")
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 basmala12/smollm_finetuning5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "basmala12/smollm_finetuning5"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "basmala12/smollm_finetuning5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/basmala12/smollm_finetuning5

SGLang

How to use basmala12/smollm_finetuning5 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 "basmala12/smollm_finetuning5" \
    --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": "basmala12/smollm_finetuning5",
		"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 "basmala12/smollm_finetuning5" \
        --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": "basmala12/smollm_finetuning5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use basmala12/smollm_finetuning5 with Docker Model Runner:
```
docker model run hf.co/basmala12/smollm_finetuning5
```

smollm_finetuning5 — Fine-Tuned SmolLM2-1.7B for Concise Instruction Reasoning

smollm_finetuning5 is a fine-tuned version of SmolAI/SmolLM2-1.7B, trained on synthetic instruction–response samples and concise reasoning data. The model is optimized to produce short, structured, and clear answers while improving general instruction-following behavior.

The goal of this fine-tuning was to enhance reasoning clarity and response consistency in a compact 1.7B parameter model.

Features

Fine-tuned for concise and structured responses
Improved instruction-following capabilities
Handles short reasoning and explanation tasks
Lightweight and efficient (1.7B parameters)
Suitable for general-purpose educational and reasoning uses

Intended Use

Not Recommended

High-stakes or decision-critical applications
Domain-specific or specialized factual tasks
Situations requiring verified accuracy

Training Data

The model was fine-tuned on:

argilla/synthetic-concise-reasoning-sft-filtered
Instruction–answer pairs
Synthetic reasoning prompts
Concise explanation samples

The dataset consists of simplified synthetic data designed to enhance clarity, reasoning, and instruction handling.

Training Details

Base Model: SmolAI/SmolLM2-1.7B
Fine-Tuning Method: LoRA adapters (merged into final weights)
Epochs: 3
Learning Rate: 2e-4
Loss: Causal language modeling
Output Format: FP32 safetensors

Downloads last month: 1

basmala12
/

smollm_finetuning5