Our co-founder and CEO Dario Amodei recently caught up with @a16z’s Anjney Midha to discuss scaling laws, the importance of multi-disciplinary teams, and Anthropic’s approach to AI safety. Watch here: https://t.co/2Nv4xCoXib
— Anthropic (@AnthropicAI) Sep 29, 2023
ngl, Upwork is an absolute goldmine for ideas ๐ฐ https://t.co/ilROaF7cQ0
— peter! ๐ฅท (@pwang_szn) Sep 29, 2023
Yesterday, @awscloud released Bedrock as GA! Amazon Bedrock is a new AWS service that gives you access to Foundation Models (@Anthropic, @cohere ,…) with a token-based pricing. ๐ Lets compare the pricing to @OpenAI @Google and others. ๐งถ https://t.co/KnJefD0ZWP
— Philipp Schmid (@_philschmid) Sep 29, 2023
This "little" model from @MistralAI is fantastic. A snappy version now accessible through https://t.co/8NdWZF51nF. https://t.co/20jNzJcfcR
— Vipul Ved Prakash (@vipulved) Sep 28, 2023
"Put the vast majority of your energy, attention and dollars into building a great product or service and put a smaller amount into shouting about it, marketing it." https://t.co/ZTNzM88w8p
— Tren Griffin (@trengriffin) Sep 28, 2023
automatically generates changelog files summarizing code changes, using AI https://t.co/Wue1oShLdM
— LangChain (@LangChainAI) Sep 27, 2023
Another interview from an ad industry insider pointing out what is already happening in the industry and how AI is affecting it; - hyper-personalized Ads and Cost-cutting on creative teams because of AI. The writing is on the wall. Get ready for the new age at $META, $GOOGL,… https://t.co/an945LmNV2 https://t.co/UAk5M590QJ
— Rihard Jarc (@RihardJarc) Sep 27, 2023
Qwen-14B (Alibaba) The most powerful open-source model for it's size. And the longest trained: 3T tokens. Comes in 5 different versions: Base, Chat, Code, Math and Vision. (And is even trained for tool usage!) Opinion: You should consider it as your new "go-to". --- Paper:… https://t.co/MNinyS70S7 https://t.co/ZkJVxBQdlW
— Yam Peleg (@Yampeleg) Sep 27, 2023
@m_wulfmeier TorchRL and Stablebaseline3
— Guang Yang Ph.D. (@G76441648) Sep 26, 2023
Recently vLLM added AWQ (4bit) support... this open source combination will consume less memory and run faster inference of llama based models like 70B llama-2 or 34B codellama. I also saw better model quality using AWQ over other quantization methods like GPTQ https://t.co/F7NALZUR86
— anton (@abacaj) Sep 26, 2023
The more I dig into properly evaluating RAG, the more I think that purely evaluating retrieval metrics (MRR, precision) isn’t the whole picture ๐ค- you need end-to-end response evals ๐ง๐ฌ ๐ In standard retrieval, you assume the documents are *fixed*. By knowing the ground-truth… https://t.co/8zGYEm0eTE https://t.co/BfGwvKcYXV
— Jerry Liu (@jerryjliu0) Sep 26, 2023
@m_wulfmeier No need for frameworks, use CleanRL. And my own https://t.co/pOa1rchB3z can help if you need complex environments
— Joseph Suarez (@jsuarez5341) Sep 26, 2023
I wrote about five new Intel technologies I saw on the show floor at #IntelInnovation, two of which have never been written about before. https://t.co/axC8rMQo9V via @CRN #intel
— Semiconductor News by Dylan Martin (@DylanOnChips) Sep 25, 2023
If you want to give the 1.3B parameter phi-1.5 model a try -- a great base model for LoRA & QLoRA finetuning -- it's now available in Lit-GPT: https://t.co/rqJKCZbcxp https://t.co/N0amkndrTX
— Sebastian Raschka (@rasbt) Sep 25, 2023
AI x Human Super human powers, AI visuals enhancing the handpan artist This is Johann Immanuel tweaked by Kimivisuals. What a time to be alive ๐ซถ๐ป https://t.co/jzkicGhbQW
— Linus (●แด●) (@LinusEkenstam) Sep 24, 2023
I'm rereading this terrific perspective piece on AI (by @arjun_ramani3 and @zhengdongwang): https://t.co/fNXUGm6fnE
— Melanie Mitchell (@MelMitchell1) Sep 24, 2023
How to find great stocks: • Net Debt / FCFF < 3 • ROIC > 15% • Profit Margin > 10% • Revenue growth > 7% • EPS growth > 10% Here are 15 examples of Quality Stocks you should know: https://t.co/vabVdM364g
— Compounding Quality (@QCompounding) Sep 24, 2023
many inspiration: https://t.co/TMfGJGBorK or sorted by top this month: https://t.co/Nnv42pdAwR there's probably more great places to keep up at
— Andrej Karpathy (@karpathy) Sep 24, 2023
Why do I used it when I don't even support open models in instructor? I have workflows on vllm and @modal_labs for batch data extraction tasks. https://t.co/AuKes6tyBf Got 1000 pages of docs?, batch size > 1, spin up 1-2 A100, done. https://t.co/WauCWEGxVP https://t.co/aZoIBI1z9L
— jason (@jxnlco) Sep 23, 2023
๐๐๐๐ Just released OpenCopilot v0.0.11, Create custom AI copilots for your own SaaS product (like Shopify Sidekick). It can call your product's underlying APIs and do amazing things. ✨(let your users ask what they need from your product using chat commands with 5 lines of… https://t.co/e1EKPhgPWw https://t.co/nJ8GGClJ8u
— Mo Gharabat ๐ฆ (@ikbenbasha) Sep 22, 2023
“Targets are missed. Holes are bored. Rockets fail to launch. Monkeys play Pong. A cave explorer is slandered. Funding isn’t secured.” @BryceElder on top form, via @FT https://t.co/yJcA3jR6n4
— A Edgecliffe-Johnson (@Edgecliffe) Sep 22, 2023
At the end of the day, RAG is just fancier prompt engineering ๐. The whole point of RAG is keeping the model fixed and figuring out how to “stuff” the context window the right way with text to answer a given question. If you’re a prompt engineer, you need to learn AI… https://t.co/Ezb18Kknpc https://t.co/5khJvw4JEL
— Jerry Liu (@jerryjliu0) Sep 21, 2023
@aatilley & Yang Jie have a great piece in @WSJ exploring how Apple failed at modems. Key detail for us was none of the senior Apple Silicon team had wireless experience. We’ve heard that story so many times over the years. Modems are hard. @Qualcomm https://t.co/fR8lJnjm1J
— Jay Goldberg ้กพๅฟ ๅ (@jaygoldberg) Sep 21, 2023
Warren Buffett said there are 3 chapters investors must read. After that, you know all you need! 1. Chapter 8 (The Intelligent Investor) 2. Chapter 20 (The Intelligent Investor) 3. Chapter 12 (Keynes' The General Theory) Here's what you need to know: https://t.co/8VrS1NTD4D
— Daniel (@MnkeDaniel) Sep 19, 2023
The Salesforce AI team just solved LLM text summarization They released a new prompting technique called Chain of Density (CoD). Researchers realized that there's a delicate balance between the amount of details and core ideas in a summary. They then created a prompt that… https://t.co/SxEbzTUZYY https://t.co/Q7eag8XfJ4
— Lior⚡ (@AlphaSignalAI) Sep 18, 2023
How to create high quality stock photos with Midjourney? All you need is better prompts for your requirement and you are done. ChatGPT(or any LLM) helps creating such prompts easily. You just ask with the following template. ''' Imagine you are an expert prompt creator for a… https://t.co/j4bRRhRV1r https://t.co/sccNZ5zzSG
— Umesh (@umesh_ai) Sep 18, 2023
Wow. Got over 100 sci-fi book recs here... The most recommended were: 1. Foundation (Isaac Asimov) — 8 recs 2. Dune (Frank Herbert) — 7 recs 3. The Moon is a Harsh Mistress (Robert Heinlein) — 5 recs And the runners up: • The Martian (Andy Weir) • Consider Phlebas (Iain M.… https://t.co/oZ1kr0n539
— Christian Keil (@pronounced_kyle) Sep 17, 2023
Top ML Papers of the Week (Sep 11 - Sep 17): - ChatDev - Radiology-Llama2 - Robot Parkour Learning - Textbooks Are All You Need - A Survey of Hallucination in LLMs - The Rise and Potential of LLM Based Agents ...
— DAIR.AI (@dair_ai) Sep 17, 2023
“The reason Silicon Valley is so successful is because it’s so f’n far from DC.” Must watch talk by @bgurley. Here’s the TL;DR: When the Telecommunications Act of 1996 was passed, it brought the entire startup telecom industry to zero. When Dodd-Frank was passed in 2010, it… https://t.co/Le2OmwaasH
— Ryan Shea (@ryaneshea) Sep 17, 2023
Incredible foresight... over and over The personal computer: a “historical imperative” https://t.co/5HduAmqATy
— Mustafa Suleyman (@mustafasuleyman) Sep 17, 2023
For many fine-tuning and RAG projects, getting data is the hardest part. That's why I'm proud to introduce Galactic—an open-source library for cleaning and curating text data for AI. Deduplicate, detect PII, cluster, and more. Powered by local embeddings. pip install galactic-ai https://t.co/FdsMxCWZU9 https://t.co/qhNZEvQ4hc
— Ben (48/100) (@andersonbcdefg) Sep 16, 2023
Mind-blowing book. Best exploration of "what happens if we upload our minds" that I've read yet. https://t.co/KfcD6b2tbk
— Christian Keil (@pronounced_kyle) Sep 16, 2023
๐ฆ Ollama Embeddings Ollama by @jmorgan and co makes running LLMs locally a breeze. Thanks to @herrjemand, you can now also connect to local text embedding models being run with Ollama, in ๐ฆ๐! Check it out: https://t.co/diJtNudE8b https://t.co/mTgy2kGgzp
— LangChain (@LangChainAI) Sep 15, 2023
Lots of people use AI to summarize articles, but AI summaries can be flabby & miss important info. This prompt asks the AI to keep increasing the density of a summary producing highly compressed but readable output. This link will run the prompt for you: https://t.co/KL7ZsUCPdx https://t.co/28DdG9zhLK https://t.co/JIQptbSFNr
— Ethan Mollick (@emollick) Sep 15, 2023
You can now fine-tune an LLM without writing a single line of code! A breakthrough in the open-source LLM space that can increase the speed of AI development and adoption by an order of magnitude. I’ll show you a demo you can follow, but first, let me start from the beginning:… https://t.co/F5ypIHjbM6 https://t.co/wX05ZqEvd7
— Santiago (@svpino) Sep 15, 2023
Cell-based Architecture on AWS https://t.co/qVBPucNfAj https://t.co/MiDjpwhitb
— Bilgin Ibryam (@bibryam) Sep 15, 2023
Inspired by @yoheinakajima, we decided to build an AI Knowledge Graph/Principles/Rules generator. We made it easier to update world views with natural language. We believe the intersection of KGs & LLMs is under-explored, especially for reliable production-ready tools, and… https://t.co/peSOrbDczr https://t.co/eH5vH3lS6J
— Chia Jeng Yang (@chiajy2000) Sep 15, 2023
Large Language Models for Compiler Optimization Present a 7B-parameter transformer model trained from scratch to optimize LLVM assembly for code size https://t.co/hW9O2gQ7Xl https://t.co/MJDTBkwELW
— Aran Komatsuzaki (@arankomatsuzaki) Sep 14, 2023
I open-sourced my prompt to flow chart builder last night ๐ ๐งฎ 183 lines of back-end code ๐ (of which 74 lines are one big function call) ๐ฅ️ 195 lines of front-end code Here's a flow chart describing the app, made using the app... https://t.co/d8wssa70z4 https://t.co/WzBQOC8Fzu
— Yohei (@yoheinakajima) Sep 13, 2023
This is Open Robot Parkour: the agility at Boston Dynamics level, but now open-source for everyone.๐ Two NVIDIA tech here at play: 1. IsaacGym: train the robot dog by speeding up reality 10,000x with massively parallel simulation on GPU. 2. NVIDIA Jetson NX for onboard compute.… https://t.co/5SpiYXddjW https://t.co/1bZR9vdM6U
— Jim Fan (@DrJimFan) Sep 13, 2023
Efficient Memory Management for Large Language Model Serving with PagedAttention Improves the throughput of popular LLMs by 2-4x with the same level of latency compared to the SotA systems. repo: https://t.co/wJwTJyG1vh abs: https://t.co/PfWAjvX2zn https://t.co/aBIOuC8rLY
— Aran Komatsuzaki (@arankomatsuzaki) Sep 13, 2023
first video from the @allinsummit - in conversation with @elonmusk available exclusively on @X (0:00) Besties welcome Elon via Starlink (05:31) Ukraine and Starlink (19:10) green shoots of @X (22:24) the creator economy and optimizing the @x experience (26:43) the ADL,… https://t.co/cropRdy4Y1 https://t.co/vRpnrz3DLV
— The All-In Podcast (@theallinpod) Sep 12, 2023
Meet Ghost-X Anduril Industries is excited to unveil Ghost-X, the newest and most capable variant of its Ghost autonomous uncrewed aircraft system. Learn More: https://t.co/R0doVqSEeG
— Anduril Industries (@anduriltech) Sep 11, 2023
Bank: We're going to need a wet signature on this form. Please print, sign, scan, and send back to us. Me: % brew install ImageMagick % convert -density 90 bank_form.pdf -rotate 0.5 -attenuate 0.2 +noise Multiplicative -colorspace Gray scanned_form.pdf https://t.co/FjGh8Chb4t
— Michael E. Driscoll (@medriscoll) Sep 11, 2023
⚕️RestGPT New paper (with code) connecting LLMs with Real-World RESTful APIs by @99Solaris Code here: https://t.co/vT6tDt0JJB Always fun to use LangSmith to get a better sense of what's going on under the hood! Traces: https://t.co/ZTVjT2i6Sn https://t.co/YRHjOI74Sg https://t.co/UCgDD0qhJN
— LangChain (@LangChainAI) Sep 11, 2023
Large-Scale Automatic Audiobook Creation Presents a system that can automatically generate high-quality audiobooks from online e-books abs: https://t.co/zS7u1POJe5 proj: https://t.co/8PnU0JIyg8 https://t.co/sFiQp27PPR
— Aran Komatsuzaki (@arankomatsuzaki) Sep 11, 2023
Here's my conversation with @WalterIsaacson, author of the new biography on @elonmusk and one of the greatest biographers ever, having written incredible books on Einstein, Steve Jobs, Leonardo da Vinci, Jennifer Doudna, Benjamin Franklin, and many others. Outline of our… https://t.co/fQktDajhHC https://t.co/k0YJ0RooQi
— Lex Fridman (@lexfridman) Sep 10, 2023
If you’re building agents, you should think about building them hierarchically - have each agent specialize and compose them in a graph. They will perform better and fail less often ๐ก. Here are three key insights, curated by @ajhofmann18, showing why ๐งต: Setting ๐️: A… https://t.co/LajtIYQP0c https://t.co/XpbdOAEYxB
— Jerry Liu (@jerryjliu0) Sep 10, 2023
Google's Generative AI Tools Now Turn Text Into Online Worlds (via @decryptmedia) https://t.co/DVTAYd5eIw #ArtificialIntelligence #GenerativeAI #gaming
— TonyVT SkarredGhost (@SkarredGhost) Sep 10, 2023
Lithium discovery in US volcano could be biggest deposit ever found https://t.co/o8w3zoRbKz (https://t.co/3bM6uTD3yk)
— Hacker News 50 (@betterhn50) Sep 9, 2023
Data augmentation using LLMs is very underrated
— anton (@abacaj) Sep 9, 2023
Asking 60+ LLMs a set of 20 questions This is a neat little project testing different LLMs with prompts that test for different capabilities like basic reasoning and instruction following. https://t.co/EJDPDuUsj7 https://t.co/micbUTehOh https://t.co/cbozdbVaSc
— elvis (@omarsar0) Sep 9, 2023
Replace Human Feedback with AI Feedback? ๐ค @GoogleDeepMind compared RLHF to RLAIF on the task of text summarization to evaluate if AI-generated feedback (RLAIF) can replace Human Feedback (RLHF) for LLM alignment. ๐ก ๐งถ https://t.co/V1ol0PJxO4
— Philipp Schmid (@_philschmid) Sep 9, 2023
Full video and transcript: “So the first step to becoming indistractable is realizing that most of our distractions begin from within. We tend to blame the pings, dings, and rings, but that only accounts for 10% of our distractions. 90% of our distractions begin from… https://t.co/IKWyBulGKu https://t.co/Fi9HMhOWR2
— Naval (@naval) Sep 8, 2023
One of my favorite concepts from this demo is the idea of "context-augmented browsing" capabilities for agents ๐: Retrieve context (using a RAG tool or another API call) + use that to augment what the agent can do on the web! Great combo of @llama_index + @MultiON_AI powers https://t.co/GWJt15detB
— Jerry Liu (@jerryjliu0) Sep 8, 2023
I recently sold my company, AdSkills, after having made $4.8mil. Here’s the twist tho… I did it by getting other experts to give me their courses. I’ve got a video case study of how I did all of this. It’s all free, don’t need your money.
— Justin Brooke (@FaithFunnels) Sep 7, 2023
Awesome Blog post from @FrancoisMassot ! Scaling search to TB with Quickwit on a budget. It gives actual cost and performances figures on the cost of operating Quickwit on a large amount of data. https://t.co/ueZMMoGDvt
— Paul Masurel ๐ฆ (@fulmicoton) Sep 7, 2023
Jeremy Grantham's (Billionaire Investor and notoriously bearish) latest opinion on the Markets: https://t.co/daLVPuIqt5
— Daniel (@MnkeDaniel) Sep 7, 2023
Using Permutation prompting in Midjourney to do a Gucci photoshoot around the world with a single prompt. In about 60 seconds, MJ generated 32 image grids. That's 128 unique and diverse images created in under a minute. Pretty incredible. Prompt / explanation ๐ https://t.co/uLsZp3kDE8
— Nick St. Pierre (@nickfloats) Sep 6, 2023
Falcon 180B is out๐คฏ - 180B params - Trained on 3.5 trillion tokens+7 million GPU hours - Quality on par with PaLM 2 outperforms Llama 2 and GPT-3.5 across 13 benchmarks - 4bit and 8bit precision with similar quality Demo: https://t.co/JCMYid7VYn Blog: https://t.co/vb8ppRHgJ5
— Omar Sanseviero (@osanseviero) Sep 6, 2023
Introducing `gpt-author v2` ๐✍️ One prompt -> an entire fantasy novel! Just describe the high-level details, and a chain of AI systems will write an entire book for you in minutes. - complete w/ cover art - export to ebook format And it's open-source: https://t.co/TidWp9Tzu4 https://t.co/9x70gYjmVy
— Matt Shumer (@mattshumer_) Sep 5, 2023
Good article at FastCompany on the LLM craze, the future of AI, the necessity of open source AI platform, and my anti-AI-doomer stance. https://t.co/eMByKTchKK
— Yann LeCun (@ylecun) Sep 5, 2023
๐จ๐จ Binance's Slow Train Wreck ๐จ๐จ As summer officially draws to a close, you may have been away from the screens over the last few months. Below is a summary of the major negative events for Binance since December. It's helpful to see it all in one place. Next time someone… https://t.co/W6q3JhosSf https://t.co/bYgcVGJBdG
— Travis Kling (@Travis_Kling) Sep 5, 2023
We stealth launched https://t.co/VY9we1zP7X on Product Hunt a few weeks back and hit top 5 ๐ฅ Now, we’re releasing it to the public, fully open-source! Use as a full template to build a production-quality RAG app ๐ ️ It’s not any chat app, it contains a set of key features that… https://t.co/KU78yeN7qS https://t.co/G27x14RCMv
— Jerry Liu (@jerryjliu0) Sep 5, 2023
New: Meta’s AI teams are struggling with a stream of departures, internal tensions over resources and volatility caused by ongoing restructuring. And all this is playing out as Meta makes a big push into generative AI. Story w/ @KalleyHuang https://t.co/l5eZRFYRqT?
— Sylvia Varnham O’Regan (@SylviaVarnham) Sep 5, 2023
Some things I've been noticing while playing with inpainting in Midjourney ๐ It uses a different model ๐ Precise masking isn't always ideal ๐ Removing objects without changing prompt ๐ It works great with image prompts Some examples: https://t.co/CxQrTBYldO
— Nick St. Pierre (@nickfloats) Sep 5, 2023
The Rise and Fall of ESPN's Leverage Charting ESPN's rise, including how it build leverage over the cable TV providers, and its ongoing decline, caused by the Internet. https://t.co/Gghi1pLpLX
— Stratechery (@stratechery) Sep 5, 2023
Combine multiple LoRAs using BLoRA! Batched LoRA (BLoRA) allows hooking several LoRAs into the same language model! Batch outputs can even be streamed. The code is open-sourced ๐๐ผ https://t.co/bBPx7xV3Bw
— TuringPost (@TheTuringPost) Sep 5, 2023
The code and demo for TokenFlow are both released! ๐ฅณ Check out our work for consistent video editing with a fixed text-to-image diffusion ๐ฅ Webpage: https://t.co/bTRx9ONg05 Demo: https://t.co/jaHqqK64N0 @GeyerMichal @OneViTaDay @talidekel Thanks @linoy_tsaban @multimodalart! https://t.co/AiW9QpVQYv
— Omer Bar Tal (@omerbartal) Sep 4, 2023
The top challenge in storytelling with Ai right now? Consistent characters. ๐ฅ I've curated the latest tips so you can create character-driven stories that connect with audiences. ๐ ๐ฝ️ #Gen2 https://t.co/21FcWCZraN https://t.co/ZH6xE6j7um
— Dave Villalva (@DaveJWVillalva) Sep 4, 2023
Strong disagree @mustafasuleyman! IMO, we want distribution and decentralization of power in AI which create a much safer, balanced & sustainable system and future. https://t.co/wVFfDcHqXp
— clem ๐ค (@ClementDelangue) Sep 4, 2023
๐ฏ Want to boost your LLM's performance? Fine-tuning is easy; choosing the right data is hard. LangSmith lets you fine-tune on real, relevant traces. ๐ Guide https://t.co/xwLiqdgXr1
— WFH (@WHinthorn) Sep 4, 2023
Amazon advertising has grown from $1B in 2015 to ~$40B in 2023. That's more than the *combined* revenue for Snap, Pinterest, X, Bing and TikTok. I wrote about how it happened (and how Amazon ads could challenge Google and Meta): https://t.co/SvMS5UfWyS
— Trung Phan (@TrungTPhan) Sep 3, 2023
I am in the arena trying stuff using the LangChain Expression Language ! Check it out ! It's so powerful ! @LangChainAI @chamath #buildinpublic https://t.co/764CYT1gqO
— Franck SN (@ndzfs) Sep 2, 2023
Ghibli x AI Seaside Weekend With the sea as our canvas, let's paint this long weekend with hues of Studio Ghibli's coastal charm. To sandy footprints and endless horizons, my friends! Prompt: Retro Ghibli scene, [scene description], minimal shading, bold, lined, 2D, flat, low… https://t.co/eLgKv8psyT https://t.co/jNpgcIEBvG
— Proper ๐ง (@ProperPrompter) Sep 2, 2023
My new interview with DeepMind co-founder and CEO of Inflection AI, Mustafa Suleyman: "So people have this fear, particularly in the US, of pessimistic outlooks. ... It’s BS." https://t.co/xn9JAY4V1b https://t.co/YxXoB9NB32
— Robert Wiblin (@robertwiblin) Sep 2, 2023
ChatGPT for Diagrams ๐คฏ ChatGPT now has Whimsical Diagram plugin that can instantly transfrom text into stunning flowcharts & mind maps. All you need to do is provide a simple English text prompt! Prompt: "Create a detailed mind map for learning data science from scratch." https://t.co/XJjwmWd1Ek
— Shubham Saboo (@Saboo_Shubham_) Sep 2, 2023
A must read. Darij Grinberg just uploaded one of the best notes on Graph Theory directly to ArXiv. Graph ML and graph theory was always an obscure topic for me. The 422 pages book is based on the Math 530 course at Drexel University. https://t.co/HHDgJdYWqJ https://t.co/ABvdlP6Nyi
— Lior⚡ (@AlphaSignalAI) Sep 1, 2023
I just published a detailed guide to LTV for consumer brands on @Shopify! For the lazy among us, here are the 11 most important findings in short form๐ https://t.co/Pd9lPv6oSc
— Alex Jost (@lxjost) Sep 1, 2023
There's a huge opportunity to use ML to replace hand-written heuristics in compilers. Want to try? Check out this @kaggle contest organized by Ashley Chow, Bryan Perozzi, HCL-Jevster, inversion, Mangpo Phothilimthana, and Sami Abu-El-Haija! https://t.co/jXyU2DlKJT
— Jeff Dean (@๐ก) (@JeffDean) Sep 1, 2023
