[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72587":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":9,"pushedAt":9,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":28,"lastSyncTime":29,"discoverSource":30},72587,"open-oasis","etched-ai\u002Fopen-oasis","etched-ai","Inference script for Oasis 500M",null,"Python",2098,181,24,29,0,3,11,9,28.78,"MIT License",false,"master",true,[],"2026-06-12 02:03:05","# Oasis 500M\n\n![](.\u002Fmedia\u002Farch.png)\n\n![](.\u002Fmedia\u002Fthumb.png)\n\nOasis is an interactive world model developed by [Decart](https:\u002F\u002Fwww.decart.ai\u002F) and [Etched](https:\u002F\u002Fwww.etched.com\u002F). Based on diffusion transformers, Oasis takes in user keyboard input and generates gameplay in an autoregressive manner. We release the weights for Oasis 500M, a downscaled version of the model, along with inference code for action-conditional frame generation. \n\nFor more details, see our [joint blog post](https:\u002F\u002Foasis-model.github.io\u002F) to learn more.\n\nAnd to use the most powerful version of the model, be sure to check out the [live demo](https:\u002F\u002Foasis.us.decart.ai\u002F) as well!\n\n## Setup\n```\ngit clone https:\u002F\u002Fgithub.com\u002Fetched-ai\u002Fopen-oasis.git\ncd open-oasis\n# Install pytorch\npip install torch torchvision --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu121\n# Install other dependencies\npip install einops diffusers timm av\n```\n\n## Download the model weights\nInside the `open-oasis\u002F` directory, run:\n```\nhuggingface-cli login\nhuggingface-cli download Etched\u002Foasis-500m oasis500m.safetensors # DiT checkpoint\nhuggingface-cli download Etched\u002Foasis-500m vit-l-20.safetensors  # ViT VAE checkpoint\n```\n\n## Basic Usage\nWe include a basic inference script that loads a prompt frame from a video and generates additional frames conditioned on actions.\n```\npython generate.py\n# Or specify path to checkpoints:\npython generate.py --oasis-ckpt \u003Cpath to oasis500m.safetensors> --vae-ckpt \u003Cpath to vit-l-20.safetensors>\n```\nUse a custom image prompt:\n```\npython generate.py --prompt-path \u003Cpath to .png, .jpg, or .jpeg>\n```\nThe resulting video will be saved to `video.mp4`. Here's are some examples of a generation from this 500M model!\n\n![](media\u002Fsample_0.gif)\n![](media\u002Fsample_1.gif)\n","Oasis 500M 是一个基于扩散变换器的交互式世界模型，能够根据用户的键盘输入自回归地生成游戏画面。该项目的核心功能是通过预训练的权重文件和提供的推理代码实现动作条件下的帧生成，技术上采用了深度学习领域的先进方法如扩散模型和变换器。适用于需要动态内容生成的游戏开发、虚拟现实场景构建以及任何希望利用AI生成连续视觉内容的应用场合。用户可以轻松设置并运行项目，只需安装必要的依赖项并下载模型权重即可开始体验。",2,"2026-06-11 03:42:42","high_star"]