1 line
2.7 KiB
JSON
1 line
2.7 KiB
JSON
{"repo": "chathousediffusion/chathousediffusion", "files": {"README.md": {"b6dc810c5b8ca2aed05016389bbe08c2c1fbdf27": {"t": 1785674101.9125915, "n": 2466, "h": "45042f1dfcddbf77", "c": "# ChatHouseDiffusion\n\nLarge language models and diffusion models are used to generate and edit the room plan with text prompts.\n\n## Data\n\nOur training data is based on the [RPLAN dataset](http://staff.ustc.edu.cn/~fuxm/projects/DeepLayout/index.html) and we evaluate our method's performance using the [Tell2Design dataset](https://github.com/LengSicong/Tell2Design).\n\nThe training data can be downloaded in <https://cloud.tsinghua.edu.cn/f/cda4cb89daef4da2b158/>\n and the test data (We use moonshot-v1-8k to preprocess the input texts in Tell2Design dataset) can be downloded in <https://cloud.tsinghua.edu.cn/f/2844208e0c344d18bd72/>\n\n## Quick start\n\nWe've implemented a UI for ChatHouseDiffusion and you can use it directly. (This environment has been tested and works on Windows 10 with Python 3.10)\n\n1. Install relative packages.\n\n ```shell\n pip install -r requirements\n ```\n\n2. Download the predict model and unzip it in `predict_model`.\nthe model and parameters can be downloaded in <https://cloud.tsinghua.edu.cn/f/a01a8205be55462685fd/>\n\n3. Create `api_info.json` with your own api infomation in the root path. Any LLM using OpenAI package is supported, such as GPT4, Kimi, Ollama, etc.\n\n ```json\n {\n \"api_key\": \"<your api_key>\",\n \"base_url\": \"https://api.moonshot.cn/v1\",\n \"model\": \"moonshot-v1-8k\"\n }\n ```\n\n4. Run\n\n ```shell\n python ui.py\n ```\n\nThe demo is shown following:\n\n<img src=\"demo.gif\" width=\"50%\" alt=\"demo image\">\n\n## Train and test\n\nYou can read `train.py` and edit some parameters. Run `python train.py` for training a new model.\n\nYou can read `test.py` and edit some parameters. Run `python test.py` for testing.\n\nYou can read `predict.py` and change model, editing inject step or sampling timesteps for the ui.\n\n## Method\n\nOur code is based on [Imagen-pytorch](https://github.com/lucidrains/imagen-pytorch) and [Graphormer](https://github.com/microsoft/Graphormer).\n\nThe specific method can be found in our paper.\n\n## Citation\n\nPlease cite this paper if you use the code.\n\n```bib\n@misc{qin2024chathousediffusionpromptguidedgenerationediting,\n title={ChatHouseDiffusion: Prompt-Guided Generation and Editing of Floor Plans}, \n author={Sizhong Qin and Chengyu He and Qiaoyun Chen and Sen Yang and Wenjie Liao and Yi Gu and Xinzheng Lu},\n year={2024},\n eprint={2410.11908},\n archivePrefix={arXiv},\n primaryClass={cs.HC},\n url={https://arxiv.org/abs/2410.11908}, \n}\n```\n"}}}} |