Added 15.x arm64e support to kfd by making it... use the IOSurface krw method from weightBufs (Thankfully that wasn't data PACed until iOS 16). Needs only an allproc offset right now. Looking into arm64 next.show more

TrendTrack API & MCP are live Connect it to... Claude, GPT, Gemini or any tool you already use. What you can pull : Winning ads before they saturate Scaling stores before everyone spots them Email campaigns from your competitors Hooks that are converting right now Everything you see on TrendTrack can now be extracted and processed by AI. How to use it ? API → integrate TrendTrack data directly into your tools, dashboards, or custom apps MCP → connect TrendTrack to any LLM. Claude, GPT, Gemini, Perplexity. Ask a question, it pulls the data itself, responds in real time Start building now →show more

🚀 Real-time AI Tornado Detection Today we're introducing Tornado... ID (Beta), an advanced artificial intelligence (AI) designed to analyze weather radar data in real time and identify tornadoes. Rather than only looking for clear tornado signatures, it also flags storms that are tornadic in general - storms exhibiting signatures that can lead them to develop into tornadoes. Key features: - Realtime AI detection - Confidence scoring - Available for Radar and MRMS (CONUS) Composite Available to Pro subscribers and the US (NEXRAD) weather radar network. Available now on: iOS: Android: Browser:show more

🚀 Cline v3.7 is now available with an exciting... interface upgrade! Choose your path with just a click -- Cline now displays selectable options when asking questions or presenting plans, eliminating the need to type responses and making interactions faster and more intuitive. Other updates: - Added support for a .clinerules/ directory to load multiple files at once - Prevent Cline from reading extremely large files into context that would overload context window - Improved checkpoints loading performance and display warning for large projects not suited for checkpoints - Added SambaNova API provider - Enabled VPC endpoint option for AWS Bedrock profiles - Added DeepSeek-R1 to AWS Bedrock Download it now in VS Code.show more

Most people think AI data centers are giant buildings... in the desert. One guy installed four mini Nvidia AI data centers right behind his work desk and now they pay him every month. Each unit is about the size of a small fridge. Inside: Nvidia GPUs running AI workloads 24/7. He hooked them up next to his AC system and that was basically it. Now the company pays him a flat monthly fee for the electricity and Wi-Fi they use. According to him, it brings in around $10,000/month straight into his account. The crazy part: the units also cool part of the house, cutting his AC bill by roughly $600. That’s more than $120,000 a year from four AI boxes sitting inside his home office. His mortgage is basically being paid by AI hardware behind his chair. Quietly, regular homes are starting to become AI infrastructure. Save this post. You’re watching the next gold rush move into people’s homes.show more

iOS is here. And with it, millions more people... can now join the world's most rewarding AI agent. For years, UpRock has grown into a global network powered by millions of real devices across 190+ countries. Now, that network is open to iPhone users too. Download the app. Contribute in the background. Earn UPT. Use your rewards to power Rocket AI. AI shouldn't be reserved for people with the biggest GPUs or the largest subscriptions. It should be accessible to anyone with a phone. Welcome to the next chapter of UpRock. iPhone users — it's your turn.show more

Thank you MTN Nigeria, MTN Nigeria Support for STEALING... my data, deliberately. So my data expired on the 3rd of Feb. 2025, and I tried re-subscribing with my bank apps but it kept bouncing back. I reached out to MTN customer care multiple times and was told that their network has some issues, and they are working to fix it. Keep in mind that I could barely use the data before now due to poor network coverage. I called almost every day reminding the customer care agents that the roll-over time ends on Feb. 10th, multiple times, and was reassured that the date will be extended since the issue wasn't coming from me. During this time both their MTN app, bank transactions, and USSD codes weren't working, even they themsleves couldn't renew the sub from their own end because their system was down. Today I woke up to find out that they (MTN) have proceeded to wipe out the data balance, "409 GB" which costs over N50k. The customer care agent who anwsered my call this morning said that there's nothing he can do, I can only re-activate now. Just imagine that, a network provider that takes zero accountability, nor value customer satisfaction. Absolutely infuriating.show more

For years now, we've listened to Karen Read's supporters... repeat the claim that Michael Proctor said, "pin it on the girl," treating it as some smoking gun of alleged corruption in the investigation despite there never being any evidence that Michael Proctor ever uttered those words. That phrase entered the conversation because Alan Jackson used it in court. From there, it was repeated so many times that people began treating it as an actual quote from law enforcement rather than what it really was: a phrase coined by Karen Read's own attorney. Now, thanks to Karen Read's civil team once again attempting to use "pin it on the girl" as evidence of supposed "corruption," we finally have clarity ON THE RECORD. Judge Gildea demanded that they identify the source of the quote because they had the audacity to include it in a filing without attributing it to anyone. When pressed to identify who actually said it, their answer wasn't Michael Proctor. It wasn't any member of the Massachusetts State Police. It wasn't anyone from Canton PD. IT WAS KAREN READ'S OWN DEFENSE COUNSEL as we've been saying all along. "Lies have momentum but the truth will always come to light." — Unknownshow more

you're hoping your support "plays perfectly" in League of... Legends it won't work it's like asking a stranger for sex on the first date they think you're a desperate fucking creep meanwhile i'm turning complete strangers into ABSOLUTE MACHINES within 5 minutes how? let me explain the human mimicry method first 3 minutes start playing AGRESSIVE AF walk up. ego dodge everything PERFECTLY SPACE your opponent CONTEST EVERY CS SPAM PING the level 2 cheese and execute here's the trick: most ADCs are either: - toxic and flaming - pinging like psychos - depending on their support to set up you do NONE of that instead you make your support become a MIRROR from your actions your support sees: - wait? my adc is playing aggressive? - this guy is actually good? - he gets it. now ALL OF A SUDDEN, they match you, THEY WANT TO IMPRESS YOU suddenly they're: - more CONFIDENT - warding properly - Actively trying to make plays because you made them feel COMPETENT here's what hardstuck ADCs don't understand: your support isn't your employee they're a random person who doesn't give a fuck about your LP you can't DEMAND perfect play you have to EXTRACT it the human mimicry method works because: you're not trying to control them you're making them want to perform for you i've turned Janna players into gods autofilled supports into engage chads egirls into aggressive laners not by asking them to "play perfect" but by making them feel like WE'RE the threat remember you're not looking for perfect supports you're creating them study the saskio wayshow more

[VAE] by Hand ✍️ A Variational Auto Encoder (VAE)... learns the structure (mean and variance) of hidden features and generates new data from the learned structure. In contrast, GANs only learn to generate new data to fool a discriminator; they may not necessarily know the underlying structure of the data. The International Conference on Learning Representations (ICLR) this year announced its first ever "Test of Time Award" to recognizes the VAE paper, published 10 years ago. This exercise demonstrates how to calculate a VAE by hand. [1] Given: ↳ Three training examples X1, X2, X3 ↳ Copy training examples to the bottom ↳ The purpose is to train the network to reconstruct the training examples. ↳ Since each target is a training example itself, we use the Greek word "auto" which means "self." This crucial step is what makes an autoencoder "auto." [2] Encoder: Layer 1 + ReLU ↳ Multiply inputs with weights and biases ↳ Apply ReLU, crossing out negative values (-1 -> 0) [3] Encoder: Mean and Variance ↳ Multiply features with two sets of weights and biases ↳ 🟩 The first set predicts the means (𝜇) of latent distributions ↳ 🟪 The second set predicts the standard deviation (𝜎) of latent distributions [4] Reparameterization Trick: Random Offset ↳ Sample epsilon ε from the normal distribution with mean = 0 and variance = 1. ↳ The purpose is to randomly pick a offset away from the mean. ↳ Multiply the standard deviation values with epsilon values. ↳ The purpose is to scale the offset by the standard deviation. [5] Reparameterization Trick: Mean + Offset ↳ Add the sampled offset to predicted mean ↳ The result are new parameters or features 🟨 as inputs to the Decoder. [6] Decoder: Layer 1 + ReLU ↳ Multiply input features with weights and biases ↳ Apply ReLU, crossing out negative values. Here, -4 is crossed out. [7] Decoder: Layer 2 ↳ Multiply features with weights and biases ↳ The output is Decoder's attempt to reconstruct the input data X from reparameterized distributions described by 𝜇 and 𝜎. [8]-[10] KL Divergence Loss [8] Loss Gradient: Mean 𝜇 ↳ We want 𝜇 to approach 0. ↳ A lot of math called SGVB simplifies the calculation of loss gradients to simply 𝜇 [9,10] Loss Gradient: Stdev 𝜎 ↳ We want 𝜎 to approach 1. ↳ A lot of math simplifies the calculation to 𝜎 - (1/ 𝜎) [11] Reconstruction Loss ↳ We want the reconstructed data Y (dark 🟧) to be the same as the input data X. ↳ Some math involving Mean Square Error simplifies the calculation to Y - X.show more

This app has completely transformed over the last three... years and it is amazing to see how far it has come. - Launched X Chat and encrypted DMs - Launched Video/Audio calling - Launched Ad revenue share program for the creators - Reached a new alltime high of userseconds - Launched Community Notes - Community Notes passed 1M contributers mark - Launched the AI driven algo - Paid millions to the creators - Integerated Grok within the app - Rebranded as 𝕏 - Open sourced the recommendation algorithm. - Over 500 million posts everyday - Over 100 billion impressions daily - 650+ million users - 400000+ communities - #1 News App in 155+ countries - 2 billion+ app downloads - People are spending more time on X - 6th most visited website - Launched X Pro - Announced X Payments - Reduced child sexual abuse material by around 95 since the acquisition - Released subscriptions for creators - Removed legacy checkmarks - Prioritized the verified accounts to fight spam bots - All time low hate speech levels - Released new features in the shortest time period in Xs history - Exposed truth via Twitter files - Removed censorship - Restored free speech - Launched long form posts / Articles. - Improved the live streaming quality - Started paying the creators twice in a month - Added live chat to live streams - Enhanced the Community Notes algorithm - Added an ability to post timestamps while uploading videos - Added an edit post feature - Added view counts on posts publicly - Enhanced the support for Premium users - Added similar posts button - Added new app icons for premium users - Launched different tiers for Premium - Launched Verified Organizations - Added bookmark button on home timeline for posts - Over Ninety percent of the top 100 advertisers have returned to X - Video views are up as compared to previous years - All time high platform usage - Added more privacy options for calling - Introduced recruitment jobs feature for verified organizations - Expanded the community notes in more countries worldwide - Improved ad experience - Increased group chat size to 200 people - Lowered the eligibility threshold for ads revenue sharing from 15M to 5M - Rolled out vertical video ads for brands - Improved communities experience - Announced the removal of tweepcred system - Added live streams on the top bar of the feed - Improved live streaming tested game streaming - Launched the immersive video player - Made the process of reporting a post easier - Improved the UI - Announced the ability of group calling - Announced the removal of block feature except for DMs - Major upgrades to videos and media - Secured money transmission license in multiple states for payments - Introduced video downloads for verified users - Added an ability to limit replies to verified users only - Launched expanded bios - Made communities accessible in more countries - Added ID verification - Added an ability to hide likes - Added a tab for verified followers on user profiles - Added an ability to accept dms from verified users only to prevent spam - Introduced dm support for the ad customers - Announced that X will fund legal representation for users fired from their jobs for their posts - Included posts from communities in for you feed - Renamed notes feature to articles that will have no character limits - Added highlights tab to the profiles - Improved account analytics - Added an ability for community admins to spotlight their community in the profile - Introduced voice dms - Deleted millions of bot accounts - Introduced an ability to upload long videos - Improved advertisement brand safety - Introduced formatting for long form posts - Extended the community notes availability to more countries - Introduced individual reply to dms custom emoji reactions - Launched faster playback mini player for videos - Added an ability to allow dms from subs only - Continued UX/Ui improvements - Added air play button for videos on iOS - Added an ability to fast forwardrewind videos on - Became the most downloaded news app in multiple countries - Improved ad relevance - Removed child po*n - Introduced rate limits to prevent data scraping - Added fast forward backward feature to videos - Launched subscriber only replies - Launched spaces on web - Added an ability to hide the blue checkmark - Reduced scam accounts - Improved engagements for advertisers - Reinstated a multitude of suspended accounts - Introduced Multiple organization affiliate badges - Improved post analytics - Community notes debunking propaganda misstatements - Introduced verified badges for organizations - Launched X TV and many more features.show more

Apple just turned Siri into a wrapper. the most... genius business move in AI this year. iOS 27 will lets us use claude, gemini, grok, perplexity, and more through Siri. install the app, toggle it on in settings. they spent years trying to make Siri compete. they couldn’t. so they turned the iPhone into an AI marketplace instead. every AI company now fights for a slot on 2.5 billion devices. Apple takes 30% of every subscription. no models to train. no data centers to build. no compute costs. just distribution. the middleman always win. the company that lost the AI race just figured out how to tax everyone lmao.show more

we just added Apple Pay support in Chrome and... non-Safari browsers on Dodo Payments. you can now use Apple Pay on desktop too. it opens a QR popup, you scan it with your iPhone, and pay. no Safari required. most payment providers still haven't shipped this. they're stuck on older Apple Pay versions that only work in Safari. we updated to the latest spec and built the cross-browser flow from scratch. this is live now for all merchants. no integration changes needed.show more

🚀LATEST PARTNERSHIP: TECTUM We’re delighted that Tectum has partnered... with Zesh. To celebrate, here’s another incredible giveaway: 💰$300 Giveaway - Participate NOW: Tectum is the fastest 3rd Gen Layer 1 blockchain with 3.5 million TPS and SoftNote™️ crypto payment solution. Their unique method turns the impossible into a routine process: they completely isolate all the transaction-related heavy data from the entire process by hashing, encrypting and signing the bundles at the end of every phase of a transaction and archiving it. Check them out and dive into To The Moon to start earning your rewards!show more

🧵 Understanding Zama; the future of privacy tech &... homomorphic encryption 1️⃣ Zama is pioneering fully homomorphic encryption (FHE). A breakthrough that lets you compute on encrypted data without decrypting it. 🔐 That means total privacy, even the system running your data can’t see it. 2️⃣ Why it matters: Right now, cloud apps, AI models, and databases must access your raw data to work. FHE changes that. your data stays private while still usable. 3️⃣ Zama builds open-source FHE tools for developers, turning advanced cryptography into practical products for AI, blockchain, and Web3. 4️⃣ Imagine: •AI that learns without reading your secrets 🤖 •Blockchain transactions with zero data leaks •Cloud apps that never see your info 5️⃣ Zama’s mission: Privacy should be the default, not an option. They’re making privacy-preserving tech simple, scalable, and open for everyone. 🔚 In a world obsessed with data, Zama might just be building the encryption layer of the future internet. 🌐show more

Something big is happening in robotics - and it’s... hiding in plain sight. This post is not about dancing robots but in the data that powers them. Open robotics datasets have exploded this year, turning the field into a more scalable and collaborative ecosystem. In just two years, Hugging Face datasets grew from 11k to over 600k - and robotics is by far the fastest-growing segment. We went from 1k robotics datasets in 2024 to 27k in 2025! For comparison, text generation, the second-largest category, has only around 5k datasets in 2025. That gap is massive. Open datasets are important because robotics lives and dies by real-world robot data - video, actions, sensors, failures. By making this data easy to upload, reuse, and benchmark, researchers, startups, and large players are now releasing real-robot datasets that would have stayed locked inside labs just a few years ago. Major contributors include NVIDIA, LeRobot initiative, and a rapidly growing maker community. This surge is also enabled by cheaper video storage, better tooling, and an open-source AI culture now spilling into the physical world. And it really matters: open robotics data dramatically lowers entry barriers, accelerates learning-by-doing, and speeds up progress toward generalist and humanoid robots. Robotics won’t scale through hardware alone - but to a large extent through shared data. Viz below from AI World - link to the story and more viz/filters in comment.show more