Skip to content
Other

smollm2-sft

Compact coding adapter for SmolLM2

1 min readupdated

This is a LoRA (a lightweight add-on for a base model) built on the SmolLM2-360M-Instruct language model and fine-tuned on the CodeAlpaca-20k dataset for programming tasks. It generates code and answers technical queries on top of an ultra-compact 360M-parameter model. The adapter is aimed at developers and hobbyists who need a lightweight setup for local coding experiments.

Key features

  • Base: SmolLM2-360M-Instruct
  • Method: LoRA configured with r=8 and alpha=16
  • Training: fine-tuned on 1,000 samples from CodeAlpaca-20k
  • Efficiency: minimal system requirements to run on a CPU or GPU

What you can do

  • Generate a function or simple script from a text description
  • Attach the adapter to the base model using the peft library
  • Merge the adapter weights into the base model via merge_and_unload()
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model_name = "HuggingFaceTB/SmolLM2-360M-Instruct"
tokenizer = AutoTokenizer.from_pretrained(base_model_name)
model = AutoModelForCausalLM.from_pretrained(
    base_model_name,
    torch_dtype="auto",
    device_map="cpu",
)
model = PeftModel.from_pretrained(model, "your-username/smollm2-sft")

prompt = "Write a Python function to find the maximum of three numbers."
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
output = model.generate(**inputs, max_new_tokens=100, temperature=0.7)
print(tokenizer.decode(output[0], skip_special_tokens=True))
Read moreDetails
Training: 250 steps, batch size 2, gradient accumulation steps 4
Limitations: trained on 1,000 samples with a modest 360M base capacity
License: MIT
Discuss Telegram 0 29

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.7 GB · ready files1.2 GB
  • FP8 / INT8recommendedweights 0.3 GB0.9 GB
  • 4-bitweights 0.2 GB0.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

This is a LoRA: counted together with its base model SmolLM2-360M-Instruct.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # aaryangoyal/smolLM2-sft-CodeAlpaca20k
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download HuggingFaceTB/SmolLM2-360M-Instruct --local-dir ~/models/SmolLM2-360M-Instruct
    hf download aaryangoyal/smolLM2-sft-CodeAlpaca20k --local-dir ~/models/smolLM2-sft-CodeAlpaca20k
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to SmolLM2-360M-Instruct
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

Requires the base model

HuggingFaceTB/SmolLM2-360M-Instruct

This model is a LoRA / adapter. Download the base model first, then load this one on top.

  1. SmolLM2-360Mfamily root
  2. SmolLM2-360M-Instruct
  3. smolLM2-sft-CodeAlpaca20k

New posts about versions of SmolLM2-360M (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 8All on Hugging Face
Merges 1All on Hugging Face

About the model

aaryangoyal/smolLM2-sft-CodeAlpaca20k
Author
aaryangoyal
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

290

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments