Skip to content
Other

Home-decor SigLIP 2

поиск мебели по фото и тексту

2 min readupdated

Это файнтюн google/siglip2-base-patch16-224 от AlphJain для поиска мебели и предметов декора: реальное фото комнаты → товар, текст → товар и текст → комната. Около 0.38 млрд параметров, библиотека transformers, задача zero-shot-image-classification.

Ключевые особенности

  • Реальное фото комнаты → товар: R@1 вырос с 0.076 до 0.164, R@10 — с 0.364 до 0.508.
  • Поиск по тексту: запрос → комната R@10 0.391 против 0.281 у базы; запрос → товар на ABO R@1 0.680 против 0.620.
  • Фото мебели → другие снимки той же модели: R@1 0.827 против 0.778.
  • Версия v3-clean-nostyle; оценка на данных, не использованных для обучения или выбора чекпоинта.
  • Готовый код для transformers: можно считать эмбеддинги изображений и текста и ранжировать по косинусной близости.

Пример

from PIL import Image
from transformers import AutoModel, AutoProcessor
model_id = "AlphJain/home-decor-siglip2"
model = AutoModel.from_pretrained(model_id).eval()
processor = AutoProcessor.from_pretrained(model_id)
image = Image.open("room.jpg").convert("RGB")
texts = ["green velvet sofa", "scandinavian bedroom", "brass floor lamp"]
# далее сравните эмбеддинги изображения и текста
ПодробнееГлавные метрики: фото комнаты → товар R@1 0.164 / R@10 0.508; запрос → комната R@10 0.391; запрос → товар ABO R@1 0.680.
База — google/siglip2-base-patch16-224, около 0.38 млрд параметров. Требования к VRAM в карточке не указаны.
Тексты в примере — в нижнем регистре, как при обучении. Оценка на отложенных данных; для ABO обе модели близки к потолку.
Discuss Telegram 0 36

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.7 GB · ready files1.3 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.

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.

    # AlphJain/home-decor-siglip2
    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 AlphJain/home-decor-siglip2 --local-dir ~/models/home-decor-siglip2
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to home-decor-siglip2
  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 siglip2-base-patch16-224 (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face

About the model

AlphJain/home-decor-siglip2
Author
AlphJain
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

360

Ask AI about this model

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

Comments

Other

MPIXEL

MPIXEL