TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

AskCyph Lite: In-Browser AI Model Execution with Privacy – An Experiment

1 点作者 cyphertechinc超过 1 年前

1 comment

cyphertechinc超过 1 年前
Hi Hackers,<p>We’re thrilled to present AskCyph™ LITE, a novel approach to AI chatbot technology, executed right in your browser, keeping your data private. The first time it takes a little time to download the model and the following time it is a lot faster to initialize.<p>Currently, we support finetuned Red Pajama (Basic) and customized finetuned Llama 2 7b (Advanced). We created this as a way for enthusiasts of all levels to take a plunge and have an AI model running.<p>Please try it out, and let&#x27;s discuss your thoughts, suggestions, and concerns below.<p><a href="https:&#x2F;&#x2F;askcyph.cypherchat.app&#x2F;" rel="nofollow noreferrer">https:&#x2F;&#x2F;askcyph.cypherchat.app&#x2F;</a><p>A bit about us. We are the creator of CypherChat (<a href="https:&#x2F;&#x2F;cypherchat.app&#x2F;" rel="nofollow noreferrer">https:&#x2F;&#x2F;cypherchat.app&#x2F;</a>), a privacy-focused messaging platform that does not rely on personal information and offers End-to-End Encrypted (E2EE) and Peer-to-Peer (P2P) communication. We recognize the need for privacy.<p>Offline access<p>Enhanced privacy<p>Basic or Advanced Models<p>Requires 4GB&#x2F;8GB free RAM<p>Computer with integrated or external GPU<p>We do want to give a shout-out and acknowledge community projects that inspired us and made innovation and AskCyph™ LITE possible.<p>Hugging Face<p>Apache TVM<p>MLC AI - Web LLM<p>TOGETHER<p>Llama2<p>#AIForAll #AskCyphLITE #CypherChat #CypherTechInc