Skip to content
Other

ResPlan CVAE

Floor plan generation and interpolation

2 min read

ResPlan CVAE is a conditional variational autoencoder (CVAE, an architecture that compresses data into a latent space to generate new samples) trained on the ResPlan residential floor plan dataset. It encodes floor plan geometry into a 64-dimensional latent space and reconstructs it via a decoder conditioned on numerical parameters like area and wall depth. Built during the IAAC AI Architecture 2026 hackathon, this model is designed for exploratory generative design rather than production-ready architectural documentation.

Key features

  • CVAE architecture: A compact 64-dimensional latent space for encoding and reconstructing spatial patterns.
  • ResPlan dataset: Trained on 17,000 residential floor plans across 14 geometric channels, including walls, doors, rooms, and gardens.
  • ONNX format: Exported to ONNX for fast inference via ONNX Runtime without heavy deep learning frameworks.
  • Interpolation: Supports smooth transitions between the latent representations of two different floor plans.

What you can do

  • Design exploration: Generate intermediate residential floor plan variations using latent interpolation.
  • ONNX inference: Run the model locally to reconstruct 14-channel raster floor plan representations.
  • Pipeline integration: Use saved sample inputs to test and prototype architectural ideas.

‍Example

Running model inference with ONNX Runtime:

import onnxruntime as ort
import numpy as np

session = ort.InferenceSession(
    "resplan_cvae.onnx",
    providers=["CPUExecutionProvider"]
)

images = np.load("example_inputs/floorplan_01.npy").astype(np.float32)
conditions = np.load("example_inputs/condition_01.npy").astype(np.float32)

input_names = [item.name for item in session.get_inputs()]

outputs = session.run(
    None,
    {
        input_names[0]: images,
        input_names[1]: conditions
    }
)

print("Output shape:", outputs[0].shape)
Read moreDetails
Parameters: 2.32M (for the CVAE_z64 configuration).
Best metric: Lowest reconstruction error and objective score achieved with a latent dimension of 64 (test objective: 187.84).
Inputs: 64x64 raster resolution, 14 geometric channels, and 3 numerical features.
Limitations: Outputs raster images rather than vector CAD/BIM files; walls and details can appear blurry, and architectural validity is not guaranteed.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 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.

    # fawad-iaac/hackathon-team4-resplan-ae
    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 fawad-iaac/hackathon-team4-resplan-ae --local-dir ~/models/hackathon-team4-resplan-ae
    # The model is in ~/models. How to run it is in the model card (link above).
  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 hackathon-team4-resplan-ae (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

fawad-iaac/hackathon-team4-resplan-ae
Author
fawad-iaac
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