[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-85130":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":9,"languages":8,"totalLinesOfCode":8,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":13,"stars7d":13,"stars30d":13,"stars90d":13,"forks30d":13,"starsTrendScore":13,"compositeScore":14,"rankGlobal":8,"rankLanguage":8,"license":15,"archived":16,"fork":16,"defaultBranch":17,"hasWiki":18,"hasPages":16,"topics":19,"createdAt":8,"pushedAt":8,"updatedAt":20,"readmeContent":21,"aiSummary":8,"trendingCount":13,"starSnapshotCount":13,"syncStatus":11,"lastSyncTime":22,"discoverSource":23},85130,"Uni-Edit","zhengdian1\u002FUni-Edit","zhengdian1",null,"Python",79,2,1,0,33.9,"Apache License 2.0",false,"main",true,[],"2026-06-15 10:04:38","\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Flogo.jpg\" alt=\"Uni-Edit\" width=\"480\"\u002F>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fzhengdian1.github.io\u002FUni-Edit-proj\u002F\">\n    \u003Cimg\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUni--Edit-Website-0A66C2?logo=safari&logoColor=white\"\n      alt=\"Uni-Edit Website\"\n    \u002F>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2605.21487\">\n    \u003Cimg\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUni--Edit-Paper-red?logo=arxiv&logoColor=red\"\n      alt=\"Uni-Edit Paper\"\n    \u002F>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002FUni-Edit\u002FUni-Edit-BAGEL\">\n    \u003Cimg \n        src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUni--Edit-Model-yellow?logo=huggingface&logoColor=yellow\" \n        alt=\"Uni-Edit Model\"\n    \u002F>\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002FUni-Edit\u002FTrain-Data\">\n    \u003Cimg\n      src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUni--Edit-Data-blue?logo=googleplay&logoColor=blue\"\n      alt=\"Uni-Edit Train-Data\"\n    \u002F>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n# Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning\n> [Dian Zheng](https:\u002F\u002Fzhengdian1.github.io\u002F), [Manyuan Zhang](https:\u002F\u002Fmanyuan97.github.io), [Hongyu Li](https:\u002F\u002Fscholar.google.com\u002Fcitations?hl=zh-CN&user=PccL82sAAAAJ), [Hongbo Liu](https:\u002F\u002Fgithub.com\u002FAlexios-hub), [Kai Zou](https:\u002F\u002Fgithub.com\u002FJacky-hate), [Kaituo Feng](https:\u002F\u002Ftulerfeng.github.io\u002F), [Hongsheng Li](https:\u002F\u002Fwww.ee.cuhk.edu.hk\u002F~hsli\u002F)\u003Csup>+\u003C\u002Fsup>\n>\n> contact: zd1423606603@gmail.com\n> \n> We introduce **Uni-Edit**, an intelligent image editing task that serves as the **first general task for Unified Multimodal Model (UMM) tuning**. Unlike conventional mixed multi-task training that suffers from inherent task conflicts and requires complex multi-stage pipelines, Uni-Edit breaks this paradigm. It achieves true mutual reinforcement by **improving image understanding, generation, and editing capabilities simultaneously using only one task, one training stage, and one dataset.**\n> \n> To overcome the limitations of simplistic existing editing data, we propose the **first automated and scalable data synthesis pipeline** for intelligent editing. By transforming diverse VQA data into complex instructions with embedded questions and nested logic, we build **Uni-Edit-148k**, a dedicated dataset pairing reasoning-intensive instructions with high-quality edited images.\n> \n> Extensive experiments on BAGEL and Janus-Pro demonstrate that tuning solely on Uni-Edit achieves **comprehensive enhancements across all three multimodal capabilities** without requiring any massive data mixing, balancing tricks, or auxiliary operations.\n\n\u003Cp align=\"center\">\u003Cimg src=\"assets\u002Fteaser.webp\" width=\"95%\">\u003C\u002Fp>\n\n## 📢 News\n\n- **May 21, 2026:** Releasing train, inference, eval code and models!\n\n## 🔥 Quick Start\n\n1️⃣  Set up environment\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fzhengdian1\u002FUni-Edit.git\ncd Uni-Edit\nconda create -n uniedit python=3.10 -y\nconda activate uniedit\npip install -r requirements.txt\npip install flash_attn==2.5.8 --no-build-isolation\n```\n\n2️⃣  Download pretrained checkpoint\n```python\nfrom huggingface_hub import snapshot_download\n\nsave_dir = \"your\u002Fpath\u002Fto\u002FUni-Edit-BAGEL\"\nrepo_id = \"Uni-Edit\u002FUni-Edit-BAGEL\"\ncache_dir = save_dir + \"\u002Fcache\"\n\nsnapshot_download(cache_dir=cache_dir,\n  local_dir=save_dir,\n  repo_id=repo_id,\n  local_dir_use_symlinks=False,\n  resume_download=True,\n  allow_patterns=[\"*.json\", \"*.safetensors\", \"*.bin\", \"*.py\", \"*.md\", \"*.txt\"],\n)\n\n```\n\n**⚠️ IMPORTANT: Custom Architecture**\nBecause this is a custom architecture, you **CANNOT** load it directly via `AutoModel.from_pretrained()`. To run the provided inference code, you **MUST** physically merge these shards into a single `ema.safetensors` file on your local machine.\n\nRun the following Python script in the directory where you downloaded the repository. \n*(Note: You need at least 54GB of free system RAM to perform this merge).*\n\n```bash\npython merge.py --model_path your\u002Fpath\u002Fto\u002FUni-Edit-BAGEL\n```\n\n3️⃣ Quick infer with Uni-Edit with task type `gen`, `und`, `edit`!\n```bash\npython infer.py --task edit\n```\n\n## 🔥 Train & Eval\n\n### Train\n\n```bash\nbash train.sh\n```\n\nYou can replace the variables in the script with your own before running. \nSee [TRAIN](https:\u002F\u002Fgithub.com\u002FByteDance-Seed\u002FBagel\u002Fblob\u002Fmain\u002FTRAIN.md) for more details.\n\n### Eval\n\n```bash\nbash scripts\u002Feval\u002Frun_geneval.sh\nbash scripts\u002Feval\u002Frun_wise.sh\nbash scripts\u002Feval\u002Frun_eval_vlm.sh\nbash scripts\u002Feval\u002Frun_imgedit.sh\nbash scripts\u002Feval\u002Frun_gedit.sh\nbash scripts\u002Feval\u002Frun_rise.sh\n```\n\nWe provide the scripts for evaluating VLM, T2I and Editing benchmarks. \nSee [EVAL](https:\u002F\u002Fgithub.com\u002FByteDance-Seed\u002FBagel\u002Fblob\u002Fmain\u002FEVAL.md) for more details.\n\n## 🔥 Data Construction Pipeline\n\nWe provide the scripts for our full data construction pipeline. \nSee [DATA](https:\u002F\u002Fgithub.com\u002Fzhengdian1\u002FUni-Edit\u002Ftree\u002Fmain\u002Fdata_gen\u002FREADME.md) for more details. \n\n## ✍️ Citation\n\n```bibtex\n@article{zheng2026uniedit,\n  title   = {Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning},\n  author  = {Zheng, Dian and Zhang, Manyuan and Li, Hongyu and Liu, Hongbo and Zou, Kai and Feng, Kaituo and Li, Hongsheng},\n  journal = {arXiv preprint arXiv:2605.21487},\n  year    = {2026}\n}\n```\n\n## 📜 License\nUni-Edit is licensed under the Apache 2.0.\n","2026-06-15 02:30:05","CREATED_QUERY"]