Khalil Ur Rehman

09/04/2024

31/03/2024

How serious is the energy consumption behind AI?

Last December, a group from Hugging Face and Carnegie Mellon University published a paper with estimates for how much energy various AI models use for certain tasks.

They found that each image generation costs the equivalent of the energy used to charge one smartphone. When it comes to text generation, it only costs 1/1000 of that of image generation.

What about training? We know it's energy-intensive.

It’s hard to get concrete numbers for training costs, but based on a group of scientists' estimation in 2022, training GPT-3 has consumed at least 1300 megawatt hours - enough to support around 130 homes for an entire year.

Data centers currently account for 1-2% of global electricity use, which is about as much as the energy consumption of the entire UK. And in two years, this could more than double.

Even though, energy consumption may not be the biggest issue that we will face. Building and maintaining large AIs is so expensive that training a foundation model like GPT-4 can cost over $100 million.

Based on the latest news, Microsoft and OpenAI are working on a $100 billion supercomputer known as Stargate, which would far exceed current computing power.

In the future, there will likely only be a few large AIs globally, owned by large companies or wealthy governments. Most of us will have a subscription to a private AI service.

How to address increased inequality and build trustworthy AI as they may be the bigger issues we will face soon.

18/01/2024

🚀 𝐓𝐰𝐨-𝐌𝐢𝐧𝐮𝐭𝐞 𝐃𝐚𝐢𝐥𝐲 𝐀𝐈 𝐔𝐩𝐝𝐚𝐭𝐞

𝐄𝐱𝐜𝐢𝐭𝐢𝐧𝐠 𝐀𝐈 𝐝𝐞𝐯𝐞𝐥𝐨𝐩𝐦𝐞𝐧𝐭𝐬 𝐭𝐨𝐝𝐚𝐲 𝐟𝐫𝐨𝐦 𝐃𝐞𝐞𝐩𝐌𝐢𝐧𝐝, 𝐆𝐨𝐨𝐠𝐥𝐞, 𝐀𝐩𝐩𝐥𝐞, 𝐀𝐦𝐚𝐳𝐨𝐧, 𝐒𝐚𝐦𝐬𝐮𝐧𝐠, 𝐚𝐧𝐝 𝐎𝐩𝐞𝐧𝐀𝐈! 𝐇𝐞𝐫𝐞'𝐬 𝐚 𝐪𝐮𝐢𝐜𝐤 𝐝𝐢𝐠𝐞𝐬𝐭:

🌟 𝐃𝐞𝐞𝐩𝐌𝐢𝐧𝐝'𝐬 𝐀𝐥𝐩𝐡𝐚𝐆𝐞𝐨𝐦𝐞𝐭𝐫𝐲: A groundbreaking AI system making waves in Olympiad geometry. It's challenging even gold medalists, pushing the boundaries of AI in math.

🔍 𝐆𝐨𝐨𝐠𝐥𝐞'𝐬 𝐈𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐯𝐞 𝐒𝐞𝐚𝐫𝐜𝐡: Introducing Circle to Search and Multisearch. Now you can select, highlight, scribble, or ask questions with your camera for a seamless and AI-enhanced search experience.

🍏 𝐀𝐩𝐩𝐥𝐞'𝐬 𝐀𝐈𝐌 𝐕𝐢𝐬𝐢𝐨𝐧 𝐌𝐨𝐝𝐞𝐥𝐬: Inspired by language models, AIM offers pre-trained vision models with autoregressive objectives, promising scalability and innovation.

💬 𝐀𝐦𝐚𝐳𝐨𝐧'𝐬 𝐀𝐈 𝐂𝐡𝐚𝐭𝐛𝐨𝐭: Get answers, jokes, and Jeff Bezos-style tips with Amazon's new AI chatbot. Test it out on iOS and Android to ask specific questions about products.

📺 𝐀𝐦𝐚𝐳𝐨𝐧'𝐬 𝐓𝐢𝐭𝐚𝐧 𝐈𝐦𝐚𝐠𝐞 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐨𝐫 𝐨𝐧 𝐅𝐢𝐫𝐞 𝐓𝐕: Command Alexa to create customized background images in artistic styles for your Fire TV. A magical touch to your viewing experience!

🤝 𝐒𝐚𝐦𝐬𝐮𝐧𝐠 𝐚𝐧𝐝 𝐆𝐨𝐨𝐠𝐥𝐞 𝐂𝐥𝐨𝐮𝐝 𝐏𝐚𝐫𝐭𝐧𝐞𝐫𝐬𝐡𝐢𝐩: Generative AI comes to Galaxy S24 smartphones! The collaboration debuts with the Samsung Galaxy S24 series equipped with Gemini Pro and Imagen 2 on Vertex AI.

🚗 𝐍𝐞𝐰 𝐀𝐈 𝐅𝐞𝐚𝐭𝐮𝐫𝐞𝐬 𝐟𝐨𝐫 𝐀𝐧𝐝𝐫𝐨𝐢𝐝 𝐀𝐮𝐭𝐨: Google unveils AI-powered additions like suggested replies and actions, enhancing your in-car experience with the power of AI.

🔮 𝐎𝐩𝐞𝐧𝐀𝐈'𝐬 𝐆𝐏𝐓-𝟓 𝐓𝐞𝐚𝐬𝐞𝐫: OpenAI CEO Sam Altman drops hints at the World Economic Forum. What's next in AI? The revelation that GPT-5 might not bear the name GPT-5, adding an air of mystery to the future of AI development.

🚀 𝐒𝐭𝐚𝐲 𝐭𝐮𝐧𝐞𝐝 𝐟𝐨𝐫 𝐦𝐨𝐫𝐞 𝐝𝐚𝐢𝐥𝐲 𝐀𝐈 𝐮𝐩𝐝𝐚𝐭𝐞𝐬! 𝐖𝐡𝐢𝐜𝐡 𝐢𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧 𝐜𝐚𝐮𝐠𝐡𝐭 𝐲𝐨𝐮𝐫 𝐞𝐲𝐞 𝐭𝐡𝐞 𝐦𝐨𝐬𝐭? 𝐒𝐡𝐚𝐫𝐞 𝐲𝐨𝐮𝐫 𝐭𝐡𝐨𝐮𝐠𝐡𝐭𝐬!

17/01/2024

Introducing EasyMed Scan - Image Analysis with Google Gemini Pro vision 🩺💻


🔍 Key Features:

Medical Image Analysis: Expertly examine medical images to identify anomalies, diseases, and health issues.

Medicine Image Analysis: Decode medicine images effortlessly. Identify medicine types, formulas, side effects, and use cases.

🤖 GenerativeAI Integration:
With the seamless integration of GenerativeAI's Gemine-Pro-Vision model, EasyMed Scan ensures accurate and detailed analysis results. Our AI model interprets complex medical and medicine-related images, providing valuable insights.

🔄 How it Works:

Upload an image (jpg, jpeg, or png).
Choose between Medical Image Analysis or Medicine Image Analysis.
Click "Analyze" and witness the power of AI in action.

📥 Download Your Report:
For your convenience, EasyMed Scan generates detailed reports for each analysis. Users can download these reports in a click, ensuring easy access to valuable information.

🔗 Github Repo: https://github.com/Khalil-Rehman9/EasyMed-Scan

We look forward to your feedback and collaboration in advancing medical image analysis.

26/11/2023

Devcon 2.0 was a two-day tech event held at the Institute of Space Technology in Islamabad.

The primary objective of this event was to empower young entrepreneurs and students in the realm of technology.

My senior, Jawad, and I participated in the Innovative O Thon, an idea pitching session. Our concept was the "Water Motor Automation System," with a key feature being the integration of AI into the product.

Alhamdulillah, we secured a runner-up position.

Special thanks to GDSC IST for organizing this incredible event.

03/06/2023

Machine learning part 1 Done ✅

03/06/2023

🚨🚨OpenAI just launched cybersecurity grant program

program seeks to:
1. Empower defenders: We would like to ensure that cutting-edge AI capabilities benefit defenders first and most.

2. Measure capabilities: We are working to develop methods for quantifying the cybersecurity capabilities of AI models, in order to better understand and improve their effectiveness.

3. Elevate discourse: We are dedicated to fostering rigorous discussions at the intersection of AI and cybersecurity, encouraging a comprehensive and nuanced understanding of the challenges and opportunities in this domain.

Apply now!
If you share our vision for a secure and innovative AI-driven future, we invite you to submit your proposals and join us in our aim towards enhancing defensive cybersecurity technologies.
OpenAI will evaluate and accept applications for funding or other support on a rolling basis. Strong preference will be given to practical applications of AI in defensive cybersecurity (tools, methods, processes). We will grant in increments of $10,000 USD from a fund of $1M USD, in the form of API credits, direct funding and/or equivalents.

Offensive-security projects will not be considered for funding at this time.
All projects should be intended to be licensed or distributed for maximal public benefit and sharing, and we will prioritize applications that have a clearplan for this.

submit your proposal here.

Your proposal should include:1. Set of questions or problems you hope your project will answer or address2. Description of methodologies and approaches used in the project3. Expected results of the projectNo more than 10 pages, using 10-pt font and 1-inch margins.

19/05/2023

Modern PC with various peripherals

15/05/2023

Some Art Image for Children Book Cover

13/05/2023

08/05/2023

Super Hot 🥵

08/05/2023

Happy you will enjoy it..

05/05/2023

Microsoft’s Bing chatbot gets smarter with restaurant bookings, image results, and more / Bing is also getting smart and persistent chat history, video answers, and even plug-ins. It

28/04/2023

This is big !

`ChatGPT with browsing` is now being rolled out worldwide to Plus subscribers.

Basically, you now have the power of ChatGPT with with web scraping to get up-to-date data! 🤯

(For now it’s works only for subscribers that have access to plugins)

28/04/2023

23 artificial intelligence tools by categories to learn and entertain you.

- Video: Video creation and editing 10X faster:

1. Supercreator.ai: https://lnkd.in/dcQYAbB5
2. Tavus: https://www.tavus.io/
3. Windsor.io: https://www.windsor.io/

- Images: Creation of art and images from text:

1. Stokimg.ai: https://stockimg.ai/
2. Midjourney: https://lnkd.in/d6YPyvNc
3. Dreamer: https://lnkd.in/dpXCUa8d

- Text: Text generation

1. ChatGPT: https://lnkd.in/dFjv_Xae
2. Notion: https://lnkd.in/dTyP27BT
3. Jasper: https://www.jasper.ai/

- Research: Content creation or summarize articles in seconds.

1. Bearly: https://bearly.ai/
2. Scholarcy: https://www.scholarcy.com/

- Design: Designs, logos, UI in minutes.

1. Looka: https://looka.com/
2. Gaileo: https://www.usegalileo.ai/
3. Uizard: https://uizard.io/

- Presentations: Create presentations 10x faster.

1. Slidesai: https://www.slidesai.io/es
2. Otterai: https://otter.ai/
3. Murfai: https://murf.ai/

- Audio: Audio, music and sounds in minutes.

1. WhisperMemos: https://whispermemos.com/
2. Soundful: https://soundful.com/
3. Steno: https://steno.ai/

- Productivity: Document processing, video assistant or writing:

1. Nanonets: https://nanonets.com/
2. Lumen5: https://lumen5.com/
3. Jenni: https://jenni.ai/

I hope you enjoy these tools as much as I do during these holidays.

If you want to contribute and enrich the list, you can add other interesting tools in the comments so that we can all stay up to date.

Thank you.

19/04/2023

The poe’s Stylish white puffy coat slowly gets bigger until become a Giant kaiju

19/04/2023

Harry potter series By Mid journey

19/04/2023

Dehumanization of society due to artificial intelligence, loss of contact with real life, people isolated in their virtual world, loss of reference points, anxiety, increased complexity of human relationships, sources of stress

18/04/2023

Fruit Furniture 🪑

Created by Midjourney user

18/04/2023

{Just in} Adobe adds Firefly to video editing🔥

Edit videos using simple text commands:

-AI color corrections/enhancements
-Animated text and motion graphics generation
-Matching B-roll footage
-Music and sound effects
-Auto-generated storyboards

17/04/2023

📺 The best Stanford, CMU, and MIT courses for AI (with YouTube playlists)

- With a multitude of AI courses available online, coming up with a study plan for AI can easily lead to decision fatigue.
- I often get asked about which courses have been useful to me to build a foundation in AI. Here’s my list of courses along with their respective YouTube playlists (note that this is an ordered list of increasing difficulty, based on my personal experience).

Stanford CS221: Artificial Intelligence: Principles and Techniques
https://www.youtube.com/playlist?list=PLoROMvodv4rOca_Ovz1DvdtWuz8BfSWL2

Stanford CS229: Machine Learning Full Course taught by Andrew Ng | Autumn 2018
https://www.youtube.com/playlist?list=PLoROMvodv4rMiGQp3WXShtMGgzqpfVfbU

Stanford CS230: Deep Learning | Autumn 2018 By Andrew Ng |
https://www.youtube.com/playlist?list=PLoROMvodv4rOABXSygHTsbvUz4G_YQhOb

Lecture Collection | Convolutional Neural Networks for Visual Recognition (Spring 2017)
https://www.youtube.com/playlist?list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv

Stanford CS224N: Natural Language Processing with Deep Learning | Winter 2021
https://www.youtube.com/playlist?list=PLoROMvodv4rOSH4v6133s9LFPRHjEmbmJ

Stanford CS234: Reinforcement Learning | Winter 2019
https://www.youtube.com/playlist?list=PLoROMvodv4rOSOPzutgyCTapiGlY2Nd8u

Stanford CS330: Deep Multi-Task and Meta Learning I Autumn
https://www.youtube.com/playlist?list=PLoROMvodv4rNjRoawgt72BBNwL2V7doGI

Stanford CS25 - Transformers United
https://www.youtube.com/playlist?list=PLoROMvodv4rNiJRchCzutFw5ItR_Z27CM

CMU Neural Nets for NLP 2021
https://www.youtube.com/playlist?list=PL8PYTP1V4I8AkaHEJ7lOOrlex-pcxS-XV

CMU Multilingual NLP 2022
https://www.youtube.com/playlist?list=PL8PYTP1V4I8BhCpzfdKKdd1OnTfLcyZr7

CMU Advanced NLP 2022
https://www.youtube.com/playlist?list=PL8PYTP1V4I8D0UkqW2fEhgLrnlDW9QK7z

CMU Low-resource NLP Bootcamp 2020
https://www.youtube.com/playlist?list=PL8PYTP1V4I8A1CpCzURXAUa6H4HO7PF2c

MIT 6.S191: Introduction to Deep Learning
https://www.youtube.com/playlist?list=PLtBw6njQRU-rwp5__7C0oIVt26ZgjG9NI

MIT 6.S192: Deep Learning for Art, Aesthetics, and Creativity
https://www.youtube.com/playlist?list=PLCpMvp7ftsnIbNwRnQJbDNRqO6qiN3EyH

17/04/2023

AI will create so many new opportunities in 2023.

But most people feel left out with million things happening around AI.

For now, AI will not replace you, but someone comfortable using AI will!
What are your favorite AI tools? comment below👇

17/04/2023

💎 This is Lama cleaner. This open-source repository contains multiple models for inpainting.

💡 Here are some tasks you can do:
- Erase objects.
- Face restoration.
- Super-resolution.
- Erase and replace.
- Object segmentation.
- Remove backgrounds.

🚀 There are multiple models and techniques, such as GANs, diffusion models, Fourier convolutions, and Vision Transformers.

Worth checking out!

✨ Follow Khalil Ur Rehman for more!

ps: credits to the respective owner(s).

08/04/2023

Meta releases AI model that can identify items within images

April 5 (Reuters) - Facebook-owner Meta (META.O) published an artificial intelligence model on Wednesday that can pick out individual objects from within an image, along with a dataset of image annotations that it said was the largest ever of its kind.

The company's research division said in a blog post that its Segment Anything Model, or SAM, could identify objects in images and videos even in cases where it had not encountered those items in its training.

Using SAM, objects can be selected by clicking on them or writing text prompts. In one demonstration, writing the word "cat" prompted the tool to draw boxes around each of several cats in a photo.

Learn more: https://segment-anything.com/

07/04/2023

Gen-2: The Next Step Forward for Generative AI

No lights. No camera. All action.
Realistically and consistently synthesize new videos. Either by applying the composition and style of an image or text prompt to the structure of a source video (Video to Video). Or, using nothing but words (Text to Video). It's like filming something new, without filming anything at all.

Try Now:
https://research.runwayml.com/gen2

17/03/2023

The future of presentations is here!

With GPT-4, you can effortlessly turn any document into a stunning visual masterpiece

Say goodbye to hours of formatting and hello to beautiful presentations on autopilot!

17/03/2023

Microsoft Copilot leak shows automatically creating a PowerPoint presentation