Skip to content
Language models

Viperfish Sunlight 1.5B

компактная LLM для SwiftUI под Mac

2 min read

Модель помогает писать код на Swift и SwiftUI, выдавая готовые фрагменты интерфейсов и логики по текстовому запросу. В основе лежит Qwen2.5-Coder-1.5B-Instruct (1.54 млрд параметров) от автора viperdevAI, дообученная методом LoRA (небольшая надстройка над базовой моделью, влитая в веса) и сжатая в 4-битный формат для фреймворка MLX (инструмент для запуска нейросетей на чипах Apple Silicon). По сравнению с базовой версией модель почти вдвое чаще генерирует компилируемый код для SwiftUI, жертвуя качеством в Python. Она весит около 840 МБ и пригодится iOS-разработчикам для локальных экспериментов на Mac с 8 ГБ памяти.

Ключевые особенности

  • Занимает около 840 МБ в 4-битной квантизации MLX и запускается на Mac с 8 ГБ оперативной памяти.
  • Удваивает результат базы в задачах по SwiftUI на авторском тесте компилируемости кода через swiftc.
  • Дообучена на 6 164 исходных файлах Swift и SwiftUI из датасета The Stack v1 (dedup).
  • Распространяется под открытой лицензией Apache-2.0 как младшая модель в линейке Viperfish.

Что можно сделать

  • Генерировать шаблоны экранов, кнопок, счётчиков и базовых компонентов на SwiftUI прямо в терминале или локальном скрипте на Mac через mlx-lm.
  • Использовать как быстрый офлайн-помощник для черновиков UI-кода на Apple Silicon без подключения к облачным API.

Пример

from mlx_lm import load, generate

model, tokenizer = load("viperdevAI/Viperfish-Sunlight-1.5B")
messages = [{"role": "user", "content": "Write a SwiftUI view with a counter and an increment button."}]
prompt = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)
print(generate(model, tokenizer, prompt=prompt, max_tokens=512))
ПодробнееБенчмарки (в сравнении с 4-битной Qwen2.5-Coder-1.5B-Instruct): SwiftUI — 31 из 60 промптов против 16 у базы; обычный Swift — 14 из 33 против 13; HumanEval (Python) снизился с 67.7% до 56.1%.
Железо: около 840 МБ веса, подходит для Mac с 8 ГБ памяти через библиотеку mlx-lm.
Обучение: LoRA (rank 8, scale 20, 16 слоёв), 6 164 шага при batch size 1 и длине контекста 1024.
Ограничения: из-за обучения на сыром коде ответы могут не закрывать блок кода и затягиваться; возможны выдуманные API и устаревшие конструкции SwiftUI, код нужно проверять компилятором.
Discuss Telegram 0 23

Will it run?

Runs in U32

Fits without quantisation.

  • U32weights 2.9 GB · ready files3.7 GB
  • FP8 / INT8recommendedweights 1.4 GB2.3 GB
  • 4-bitweights 0.8 GB1.6 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 · U32 — from RTX 2060

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

Language models: with a 8192-token context.

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

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    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.

    # viperdevAI/Viperfish-Sunlight-1.5B — A30, FP8 / INT8
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv venv ~/vllm --python 3.12 && source ~/vllm/bin/activate
    uv pip install vllm
    vllm serve viperdevAI/Viperfish-Sunlight-1.5B --max-model-len 8192 --quantization fp8
  3. 3. Use it

    The server listens on port 8000 and speaks the OpenAI API: http://localhost:8000/v1 on the machine. From your computer: ssh -L 8000:localhost:8000 user@machine-IP.

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.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

  1. Qwen2.5-1.5Bfamily root
  2. Qwen2.5-Coder-1.5B
  3. Qwen2.5-Coder-1.5B-Instruct
  4. Viperfish-Sunlight-1.5B

New posts about versions of Qwen2.5-1.5B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 5All on Hugging Face
Merges 2All on Hugging Face

About the model

viperdevAI/Viperfish-Sunlight-1.5B
Author
viperdevAI
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

230

Ask AI about this model

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

Comments