Skip to content
Other

Qwen3.8-27B-3D-Coder

файнтюн для задач по 3D-программированию

1 min read

Текстовая языковая модель (LLM), дообученная на базе unsloth/Qwen3.8-27B-unsloth-bnb-4bit. Судя по названию и метаданным обучения, модель ориентирована на генерацию кода, связанного с 3D-разработкой. Она принимает текстовые запросы в формате диалога и выдаёт готовые ответы или фрагменты кода. Подойдёт разработчикам, которым нужна локальная 27-миллиардная модель под 3D-задачи с поддержкой 4-битного квантования (сжатия весов для экономии видеопамяти).

Ключевые особенности

  • Основана на квантованной версии unsloth/Qwen3.8-27B-unsloth-bnb-4bit (27 млрд параметров).
  • Дообучена методом SFT (Supervised Fine-Tuning — обучение с учителем на готовых примерах) с использованием библиотеки TRL.
  • Поддерживает стандартный чат-интерфейс через transformers pipeline.

Что можно сделать

  • Интегрировать в локального ассистента для генерации скриптов и кода в 3D-проектах.
  • Протестировать качество генерации 3D-сцен и геометрии по сравнению с базовой Qwen3.8-27B.

Пример

from transformers import pipeline

question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="syaikhipin/Qwen3.8-27B-3D-Coder", device_map="auto")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
ПодробнееБазовая модель: unsloth/Qwen3.8-27B-unsloth-bnb-4bit
Метод обучения: SFT (TRL 1.15.0)
Стек: Transformers 5.14.1, PyTorch 2.11.0, Datasets 4.8.5
Ограничения: в карточке не приведены бенчмарки, системные промпты и точный список поддерживаемых 3D-форматов.
Discuss Telegram 0

Will it run?

Runs in 4-bit

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • BF16weights 51 GB · ready files57 GB
  • FP8 / INT8recommendedweights 25 GB29 GB
  • 4-bitweights 14 GB16 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.

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.

    # syaikhipin/Qwen3.8-27B-3D-Coder
    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 syaikhipin/Qwen3.8-27B-3D-Coder --local-dir ~/models/Qwen3.8-27B-3D-Coder
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Qwen3.8-27B-3D-Coder
  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

New posts about versions of Qwen3.8-27B (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

syaikhipin/Qwen3.8-27B-3D-Coder
Author
syaikhipin
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments