Python is a snake

projects

github

Ayush Deshpande • Peerlist

A browser-based implementation of Stable Diffusion using <a target="_blank" rel="noopener noreferrer" class="break-all link-break text-gray-500 hover:underline" href="https://github.com/microsoft/onnxruntime">onnxruntime</a> that runs entirely client-side using WebGPU acceleration. Now with an llm chat!GPU is mandatory unfortunately. Might be Nvidia only. (I think cpu users can still use llm)Features:1) Live generation preview2) 5 different scheduler implementations3) Generation history4) 3 supported models (converted existing models to onnx format)5) A floating llm chat window (for funzies)

LSDN

Passionate about solving Engineering problems | Computer Engineering | Learning | Living

Building things worth building

&nbsp;🚀 Introducing MailSensei: Your Cutting-Edge Email Management Companion 📧✂️Tired of sifting through endless emails, only to discover they're not worth your time? Ever missed important messages due to email overload? In today's digital age, our inboxes are overflowing with both vital and distracting content. You might spend precious time on lengthy, irrelevant emails. MailSensei is here to revolutionize your email experience by addressing these challenges.Checkout our Video Demonstration here:- <a target="_blank" rel="noopener noreferrer nofollow" class="text-green-text rich-link break-all link-break" href="https://youtu.be/9AF3yPawvDQ">https://youtu.be/9AF3yPawvDQ</a>Need for a solution :<a target="_blank" rel="noopener noreferrer nofollow" class="text-green-text rich-link break-all link-break" href="https://media.giphy.com/media/OothRHNJSCaTS/giphy.gif">https://media.giphy.com/media/OothRHNJSCaTS/giphy.gif</a>CEOs and business professionals often receive hundreds of emails daily, making efficient email management essential. MailSensei offers a solution to navigate this email overload, ensuring that crucial messages are never missed and time is spent more productively.How to Run the Project :-<ol><li>To access the project, visit the following URL: <a target="_blank" rel="noopener noreferrer nofollow" class="text-green-text rich-link break-all link-break" href="https://replit.com/@DevenNandapurka/Peerlist-Hackathon">https://replit.com/@DevenNandapurka/Peerlist-Hackathon</a>.</li><li>In the Replit workspace, click the Run button.</li><li>Before using MailSensei, you need to set up an App Password. We've attached an article to guide you through this process: <a target="_blank" rel="noopener noreferrer nofollow" class="text-green-text rich-link break-all link-break" href="https://support.google.com/mail/answer/185833?hl=en">Generate your App Password</a>.</li><li>Once the project is up and running, enter your credentials, and click the Set Credentials button.</li><li>Click the Fetch Emails button, and all of your unread emails will be fetched and displayed in a concise summary format.</li></ol>Running locally: Requirements: 1) Python 3.7+ 2) cuda (optional for GPU support) 3) Minimum 4gb of RAM is required to run the models<ol><li>Clone the repository: git clone <a target="_blank" rel="noopener noreferrer nofollow" class="text-green-text rich-link break-all link-break" href="https://github.com/Deven1902/Peerlist-Hackathon">https://github.com/Deven1902/Peerlist-Hackathon</a></li><li>Optionally create a virtual environment: python -m venv venv</li><li>Activate the virtual environment:for Linux: source venv/bin/activatefor Windows : venv\Scripts\activate</li><li>Install requirements using pip: pip install -r requirements.txt</li><li>Run the app: streamlit run app/streamlit_app.py</li></ol>Our Unique Approach :<ol><li>Gmail Credentials Setup: Securely collect your "Gmail Address" and "App Password" via Streamlit to maintain the confidentiality of your email data.</li><li>IMAP Connection: Establish a connection to your Gmail account using IMAP (Internet Message Access Protocol) for seamless email access.</li><li>Email Decoding: Decode your emails to extract essential information, including "From," "Subject," and email content.</li><li>Email Summarization: Utilize Large Language Models (LLMs), specifically "distilbart-cnn-12-6," to generate concise summaries of your emails, saving you valuable time.</li><li>Tag Generation: Intelligently categorize and organize emails using "t5-base-tag-generation." Tags are displayed alongside email details.</li><li>Streamlined Presentation: Display email information in an ordered format, including "From," "Subject," and associated tags for easy reference.</li><li>Dropdown Summaries: Provide a dropdown button for each email, allowing users to access the summary with a single click, enhancing readability.</li><li>Original Email Link: Include a convenient link at the end of each summary that directs users to the original email for further context and action.</li></ol>Future Enhancements:<ol><li>Improved Spam Filtering: We'll enhance the accuracy of our spam detection algorithm for a cleaner inbox.</li><li>Tag Collections &amp; Sharing: Users can create and share tag collections, streamlining collaboration and productivity.</li><li>Custom Tag Generation: Customize your own tag generation models for tailored email organization.</li><li>User Preferences: Fine-tune MailSensei with preferences like summarization length and tag rules for a personalized experience.</li></ol>"Initially, we used OpenAI keys for all our app functions. However, we later changed our approach as some users may not have premium keys, and we wanted to ensure accessibility without additional costs. 😉😊”