[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-73308":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":23,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":31,"readmeContent":32,"aiSummary":33,"trendingCount":16,"starSnapshotCount":16,"syncStatus":34,"lastSyncTime":35,"discoverSource":36},73308,"deep-prove","Lagrange-Labs\u002Fdeep-prove","Lagrange-Labs","Framework to prove inference of ML models blazingly fast","https:\u002F\u002Flagrange.dev",null,"Rust",3355,97,18,13,0,4,14,17,12,27.97,"Apache License 2.0",false,"master",[26,27,28,29,30],"ai","ml","zk","zk-snarks","zkml","2026-06-12 02:03:11","# 🚀 DeepProve: Zero-Knowledge Machine Learning (zkml) Inference\n\nWelcome to **DeepProve**, a cutting-edge framework designed to prove neural network inference using zero-knowledge cryptographic techniques. Whether you're working with Multi-Layer Perceptrons (MLPs) or Convolutional Neural Networks (CNNs), DeepProve offers a fast and efficient way to verify computations without revealing the underlying data.\nzkml is the name of the subcrate implementing the proving logic.\n\n## 🤔 What Does DeepProve Do?\n\nDeepProve leverages advanced cryptographic methods like sumchecks and logup GKR to achieve sublinear proving times. This means you can prove the correctness of your model's inference faster than ever before!\n\n### 📊 Benchmark Highlights\n\nCNN 264k: This runs a CNN on the cifar 10 dataset for a total of 264k parameters. DeepProve is proving 158x faster at this size!\nDense 4M: This runs a multiple dense layers for a total of 4 million parameters. DeepProve is proving 54x faster at this size!\n\n| Model Type | ZKML Proving Time (ms) | ZKML Verification Time (ms) | EZKL Proving Time (ms) | EZKL Verification Time (ms) |\n|------------|------------------------|-----------------------------|------------------------|-----------------------------|\n| CNN 264k   | 1242                   | 599                         | 196567.01              | 312505                      |\n| Dense 4M   | 2335                   | 520                         | 126831.3               | 1112                        |\n\n\n\n## 📜 Licensing\n\n- **zkml folder**: Licensed under the [Lagrange License](https:\u002F\u002Fgithub.com\u002FLagrange-Labs\u002Fdeep-prove\u002Fblob\u002Fmaster\u002Fzkml\u002FLICENSE), unless otherwise specified.\n- **Rest of the Code**: Licensed under Apache 2.0 + MIT, as per the original repository.\n\n## 🌟 Use Cases\n\nProving inference of AI models has a wide range of applications, especially in scenarios where privacy and trust are paramount. For instance, in healthcare, sensitive patient data can be used to make predictions without exposing the data itself. In finance, models can be verified for compliance without revealing proprietary algorithms. Additionally, in decentralized applications, zero-knowledge proofs can ensure the integrity of AI computations on the blockchain, fostering trust and transparency. These use cases highlight the transformative potential of ZKML in various industries.\n\n## 🙏 Acknowledgements\n\nThis project builds upon the work from scroll-tech\u002Fceno, reusing the sumcheck and GKR implementation from their codebase. Check out their work at [scroll-tech\u002Fceno](https:\u002F\u002Fgithub.com\u002Fscroll-tech\u002Fceno).\n\nFor more technical details and usage instructions, dive into the [ZKML README](zkml\u002FREADME.md).\n\nHappy proving! 🎉\n","DeepProve 是一个用于快速验证机器学习模型推理的框架。它采用零知识证明技术，支持多层感知机（MLP）和卷积神经网络（CNN）等模型的高效验证，利用sumchecks和logup GKR等高级加密方法实现亚线性时间证明，显著提升了验证速度。例如，在处理264k参数的CNN时，其证明速度比传统方法快158倍。该框架适用于对隐私和信任要求较高的场景，如医疗健康中的患者数据保护、金融行业的合规验证以及区块链上的去中心化应用，确保AI计算的完整性和透明度。项目主要使用Rust语言编写，并在Apache 2.0许可下发布。",2,"2026-06-11 03:44:56","high_star"]