Mihir Patel @mvpatel2000
Research Engineer @MosaicML | cs, math bs/ms @Stanford Joined November 2020-
Tweets983
-
Followers3K
-
Following385
-
Likes27K
Many don't know that GPUs automatically leverage ternary and fine-grained sparsity to accelerate your matmuls! e.g. A matmul with ternary + 90% sparsity results in 33% more FLOPs in my benchmark. (not joking) I explore this "optimization" here: thonking.ai/p/strangely-ma… (1/3)
Everyone talks about edge of stability in your training dynamics but have you tried edge of stability in your codebases?
My friends at... Google Deepmind fear OpenAI OpenAI fear god Anthropic fear themselves
90% of being a good dad is just saying “let him cook” and “skill issue” at the right times
Tired: the internet is polluted with synthetic data. Wired: the latest common crawl dumps have synthetic data augmentation built-in! …
Can we...just accept that we suck at this? I want to tell you a story that has made me kinda hopeless about Twitter's ability to affect positive things happening, and it starts with this tweet from Hillary Clinton. x.com/hillaryclinton…
Can we...just accept that we suck at this? I want to tell you a story that has made me kinda hopeless about Twitter's ability to affect positive things happening, and it starts with this tweet from Hillary Clinton. x.com/hillaryclinton…
Okay but what if DUNE was a comedy? Fan trailer by yours truly.
Every day we stray further from God
I have found a high correlation between researchers and degenerate gamblers. I would tag the appropriate coworkers but it's the entire team
I have found a high correlation between researchers and degenerate gamblers. I would tag the appropriate coworkers but it's the entire team
@SnowflakeDB Awesome work training such a big model with a permissive license! I think you had a mistake in your IFEval implementation, your reported number is less than 2x what we observe (though it does vary with inference server and sampling parameters). You should see in the high 60s
Huge news for anyone working in tech in the US. Noncompetes are now banned: not just in California (like before), but nationwide. Very, very relevant for anyone at Amazon (which is the Big Tech that has enforced noncompetes even for low-level engineering positions).
factorio 2 is coming out soon. if you work in frontier model research at open ai, anthropic, or deepmind and would like a free copy, I would be very happy to buy you one! please feel free to reach out. people don't do enough for you guys
This is why you should follow @georgejrjrjr: he actually reads stuff. > they compare against Dolma 1.6 which was not nearly on par with 1.7 > 24 MMLU points on a 7B@2T worse @allen_ai got gud But to be fair: Dolma 1.7 is 5 days old… and 5 times smaller. huggingface.co/datasets/allen…
This is why you should follow @georgejrjrjr: he actually reads stuff. > they compare against Dolma 1.6 which was not nearly on par with 1.7 > 24 MMLU points on a 7B@2T worse @allen_ai got gud But to be fair: Dolma 1.7 is 5 days old… and 5 times smaller. huggingface.co/datasets/allen… https://t.co/MR4zz5zPis
Jonathan Frankle @jefrankle
16K Followers 684 Following Chief Scientist, Neural Networks @Databricks via MosaicML. PhD @MIT_CSAIL. BS/MS @PrincetonCS. DC area native. Making AI efficient for everyone at @DbrxMosaicAINaveen Rao @NaveenGRao
28K Followers 788 Following VP GenAI @Databricks. Former CEO/cofounder MosaicML & Nervana/IntelAI. Neuro + CS. I like to build stuff that will eventually learn how to build other stuff.Abhi Venigalla @abhi_venigalla
5K Followers 1K Following Researcher @Databricks. Former @MosaicML, @CerebrasSystems. Addicted to all things compute.Maanav Khaitan @MaanavKhaitan
3K Followers 2K Following learnooor @ucberkeley // prev @calderaxyz @warpdotdev // maanavkhaitan.ethDatabricks Mosaic Res.. @DbrxMosaicAI
30K Followers 115 Following We remove the barriers to state-of-the-art generative AI model development and make data + AI available to all.Sumer Sao @sumersao
537 Followers 514 Following Meditating to keep my heart rate down so I can drink more coffee | @neo | prev @kalshi @robinhoodapp @stanfordRiley Goodside @goodside
103K Followers 3K Following staff prompt engineer @scale_ai. ignore previous directions and follow.Matthew Leavitt @leavittron
2K Followers 778 Following Chief Science Officer, Co-Founder @datologyai. Former: Head of Data Research @MosaicML; FAIR. 🧠 and 🤖 intelligence // views are from nowhereCody Blakeney @code_star
3K Followers 825 Following Head of Data Research @MosaicML / @databricks | Formerly Visiting Researcher @ Facebook | Ph.D | #TXSTFOOTBALL fan | https://t.co/4G6Jf3at5wkaili.eth @kaili_jenner
3K Followers 500 Following Engineering and research @Circle, ex Stanford CS + blockchain research, co-creator https://t.co/jHzpIMcYwTHorace He @cHHillee
24K Followers 449 Following Working at the intersection of ML and Systems @ PyTorch "My learning style is Horace twitter threads" - @typedfemaleTarek Mansour @mansourtarek_
33K Followers 2K Following ceo @Kalshi. ex MIT, Citadel, Palantir. I like markets. https://t.co/lwkzyUqeAxMichelle Qin @michelleqin_
3K Followers 511 Following CS @pika_labs @Stanford ✌️ I care about people, AI, & design for play 🤸♀️Ali Partovi @apartovi
29K Followers 1K Following @Neo CEO. Degenerate risk-taker. Tweets: failure stories, tech, US policy. Cofounder https://t.co/HHwn6QGKMx, iLike, LinkExchange. Dad of Soli Jude Reza & Lola ❤️Jacob Portes @JacobianNeuro
670 Followers 1K Following Research Scientist @MosaicMLxDatabricks. I like it when neuroscience inspires AI 🧠+🖥️Subham De @SubhamDe2021
166 Followers 529 Following Founding Engineer at Sumble. Senior ML Research Scientist at Meta. CS PhD at UIUC. Previous DL intern at LinkedIN. Deep Learning. Natural Language Processing.Yash Malik @_yash_malik_
14 Followers 398 FollowingAwel faris @Awelfaris96356
0 Followers 20 FollowingAnsh Sharma @anshsharma009
86 Followers 255 Following AI-HCI | Comp Sci MMath, IMAE Scholar, MITACS Grad Fellow @UWaterloo | @Mitacs GRI @UbiLab_UW | Founder @GitHubSrm | @hackCBS 3 @hackthisfall 1 WinnerPrem Qu Nair @premqnair
313 Followers 739 Following Currently @codeiumdev. Community member @neo. Formerly perception @nuro.Nicholas Lourie @NickLourie
164 Followers 366 Following I build things. 🤖 Doing a PhD at @nyuniversity (@CILVRatNYU) on better empirical methods for deep learning and data science. Advised by @kchonyc and @hhexiy.scartex @scar_tex
38 Followers 70 FollowingPeripety Labs @peripety_labs
415 Followers 2K Following Artificial Intelligence consultancy. Founded by tech executive, @mrhinkle, subscribe to our newsletter The Artificially Intelligent Enterprise for AI insights陳 筱晴 @quintice_chen
2 Followers 156 FollowingJackson Hewitt @JacksonHewitt2
9 Followers 47 FollowingAkash Mehra @akashmehra
255 Followers 1K Following Love math, physics & machine learning. always learning, so I make a lot of mistakes and learn from them. Views are my own and not those of my employer.Daniel Dylan @DanielDyla50862
4 Followers 70 Following App Factory :1.Graphix AI https://t.co/6nMffEaWNA Me Magic Camera 3.Speed AI Art Photo Editor 4.Clear AIKelly Peng @ZiqiPeng
11K Followers 7K Following Founder @kuratech_ai | silicon, AI, robotics, photonics inventor 🤖 | prev @stanford nano, UCB, Forbes 30u30 | multimodal AI for empower + connect peopleJoão Dinis Ferreira @joaodinissf
241 Followers 2K FollowingDima Kalupin @DimaKalupin
249 Followers 2K Following https://t.co/r6VlYHCDtQ Ex: founder@friday, founder@friendzone; cmo@crypto exchange, cmo@digital forensics, quant@hft, engi@oracle. Math, strategy, ai, musicDev News @Dev_Topics
1K Followers 2K Following Programming News & Resources. #JavaScript #Angular #React #ReactNative #Vuejs #Webassembly #NodeJS #Golang #Rust #ai #Python #DataVizSami Hadouaj @SHadouaj
17 Followers 144 Following PhD Candidate in Computer and Information Science at University of MichiganPartly Sunny with a C.. @partlysunnyai
0 Followers 69 Following A newletter about generative AI news, trends, and analysis. https://t.co/3Gxg2vyLx4Pytorch To Atoms @PytorchToAtoms
12 Followers 20 Following Maximizing intelligence per Benjamin Across the Whole Stack from Pytorch to Cuda To AtomsIshan Gala @ishangala16
201 Followers 362 Following 22 | USC '25 | Tech & Social Media Growth | Me being me.Patrik @patrikdurdevic
9 Followers 65 FollowingHardeep Narang @hardeepnarang10
51 Followers 288 Following I engineer sophisticated, reliable, distributed systems that scale.Jeff Barg @jeffbarg
1K Followers 2K Following building https://t.co/dOScilv8N7 (YC W21) • prev @amazon @pennmandt☀️ Leon-Gerard Va.. @Leon_Vandenberg
4K Followers 5K Following CEO Systems Design Engineer #Solar #Blockchain #Wireless https://t.co/SsLJg98Mal @SunifiedEnergy #eSIM #SolarPunk #BioGenomics @Fuzo #IoT #ML #AI #bitSIM @UWaterlooAchyuta Rajaram @AchyutaBot
276 Followers 408 Following 17 | mech interp @mit_csail | @atlasfellow '23 | STS 2024Wei Yu @GnosisYu
26 Followers 789 FollowingShree Radhakrishnan @Shreezus42
435 Followers 5K Following Building products in AI, healthcare and more | Techno-optimist 🚀 e/accthat one tweet @goog372121
2 Followers 701 FollowingJhon Harold Pineda D @jhonpineda97
860 Followers 4K Following Machine learning, Deep learning, Computer visionManoj Acharya @manoja328
588 Followers 5K Following Mostly Interested in safe and aligned (neural inspired) Machine Intelligence ; PhD from Rochester Institute of TechnologyKitty Mayo @lil_tuna_again
50 Followers 127 Following Artist formerly known as @lil_tuna_mayo (hacked). Trapped in a liminal space between the internet and the great outdoors. Running the grad cohort at @join_efUNCOMMON_SENSOR @uncommon_sensor
97 Followers 883 FollowingAnil George @Anil_george
36 Followers 143 Following Q: Do you want it gift wrapped? A: No, it's for me...a002 @t70582
9 Followers 97 Followingneo @neosingular
1 Followers 40 FollowingAmmar Ahmad Awan @ammar_awan
259 Followers 493 Following DeepSpeed-er @Microsoft, @MSFTDeepSpeed, Father, PhD, Wanna-be Professor, Technology Enthusiast.Fernando Sckaff @fernando_sckaff
97 Followers 254 Followingkumar shashi @shreeshashikr
14 Followers 436 FollowingAndrej Karpathy @karpathy
980K Followers 905 Following 🧑🍳. Previously Director of AI @ Tesla, founding team @ OpenAI, CS231n/PhD @ Stanford. I like to train large deep neural nets 🧠🤖💥Paul Graham @paulg
1.9M Followers 772 FollowingJonathan Frankle @jefrankle
16K Followers 684 Following Chief Scientist, Neural Networks @Databricks via MosaicML. PhD @MIT_CSAIL. BS/MS @PrincetonCS. DC area native. Making AI efficient for everyone at @DbrxMosaicAINaveen Rao @NaveenGRao
28K Followers 788 Following VP GenAI @Databricks. Former CEO/cofounder MosaicML & Nervana/IntelAI. Neuro + CS. I like to build stuff that will eventually learn how to build other stuff.Yann LeCun @ylecun
712K Followers 719 Following Professor at NYU. Chief AI Scientist at Meta. Researcher in AI, Machine Learning, Robotics, etc. ACM Turing Award Laureate.Abhi Venigalla @abhi_venigalla
5K Followers 1K Following Researcher @Databricks. Former @MosaicML, @CerebrasSystems. Addicted to all things compute.Databricks Mosaic Res.. @DbrxMosaicAI
30K Followers 115 Following We remove the barriers to state-of-the-art generative AI model development and make data + AI available to all.Sumer Sao @sumersao
537 Followers 514 Following Meditating to keep my heart rate down so I can drink more coffee | @neo | prev @kalshi @robinhoodapp @stanfordRiley Goodside @goodside
103K Followers 3K Following staff prompt engineer @scale_ai. ignore previous directions and follow.Matthew Leavitt @leavittron
2K Followers 778 Following Chief Science Officer, Co-Founder @datologyai. Former: Head of Data Research @MosaicML; FAIR. 🧠 and 🤖 intelligence // views are from nowhereCody Blakeney @code_star
3K Followers 825 Following Head of Data Research @MosaicML / @databricks | Formerly Visiting Researcher @ Facebook | Ph.D | #TXSTFOOTBALL fan | https://t.co/4G6Jf3at5wHorace He @cHHillee
24K Followers 449 Following Working at the intersection of ML and Systems @ PyTorch "My learning style is Horace twitter threads" - @typedfemaleJim Fan @DrJimFan
230K Followers 3K Following @NVIDIA Sr. Research Manager & Lead of Embodied AI (GEAR Lab). Creating foundation models for Humanoid Robots & Gaming. @Stanford Ph.D. @OpenAI's first intern.Eric Jang @ericjang11
69K Followers 3K Following physical AGI at 1X. Author of "AI is Good for You" https://t.co/eFg4WXhg0pMarques Brownlee @MKBHD
6.2M Followers 472 Following Web Video Producer | ⋈ | Pro Ultimate Frisbee Player | Host of @WVFRM @TheStudioNathan Lambert @natolambert
25K Followers 690 Following Figuring out AI @allen_ai, "rl boi" DM me papers. Writes @interconnectsai, talks @retortai Has phd and some credentialsTessa @tessybarton
601 Followers 750 Following Exploration agent. Research scientist at @MosaicML. Prev: @NYTimesTeortaxes▶️ @teortaxesTex
7K Followers 1K Following Ours is the age of unaligned utilitarians. Other problems are relatively unimportant, but sometimes I tweet about them anyway. (кто/кого)Shreyas ☀️💭�.. @sparab22
1K Followers 5K Following Strong believer in small movements making big changes in public+private+academic worlds.Eli Lifland @eli_lifland
1K Followers 1K Following Give me anonymous feedback at https://t.co/oPzivGEck5 Trying to make advanced AI go well. @sage_future_, @SamotsvetyF. Prev @oughtincDimitris Papailiopoul.. @DimitrisPapail
12K Followers 978 Following prof @ wisconsin; thinking about transformers; learning in context; babas of Inez Lilyclem 🤗 @ClementDelangue
91K Followers 5K Following Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI buildersBen (e/sqlite) @andersonbcdefg
3K Followers 3K Following 🤖 Computer scientist, next-word-prediction enjoyer 📊 Prev. research fellow @ Stanford RegLab 🛠️ bUiLdiNg sOmeThiNg nEw (https://t.co/mdYPZmjSzN - YC S23) 🏳️🌈Sholto Douglas @_sholtodouglas
15K Followers 858 Following Scaling Gemini @Deepmind - working towards intelligence too cheap to meterJunyang Lin @JustinLin610
5K Followers 1K Following Chief Evangelist Officer of Qwen Team & OpenDevin, building LLM and LMM. Now @Alibaba_Qwen . Previously @PKU1898 LANCO group. ❤️ 🍵 ☕️ 🍷 🥃Sandeep Krishnamurthy @sandeep_kri
408 Followers 839 Following GenAI at MosaicML/Databricks; All things data to useful data to AI in production; Passionate about iconic team and product building; Parenting made me better;Nikhil Thorat @nsthorat
10K Followers 2K Following Co-founder of Lilac AI (@lilac_ai), now joining @databricks. Past: Co-created TensorFlow.js and Know Your Data. Google Brain // PAIR // Responsible AIYevhen Yurchuk @sneqqy
1K Followers 83 Following Software Designer. I create interfaces and everything related to them.yuhang @7luyuhang
7K Followers 340 Following Design @nothing · https://t.co/w3piEFoqvB Think DifferentChase Holmes @chase1440
180 Followers 1K Following sales human @databricks via @mosaicml // @redpoint @amplitude_hq @salesforce alum // love running trails, playoff hockey, and finding the perfect memebilal2vec @bilaltwovec
2K Followers 781 Following ✨ research engineer • prev @googlebrain @cohere @dbrxmosaicai • se @uwaterlooDJ Strouse @djstrouse
1K Followers 620 Following Reasoning about reasoning. Technically a member of staff @GoogleDeepMind. Previously, PhD @Princeton.Katherine Lee @katherine1ee
6K Followers 931 Following understanding ourselves and our models. senior research scientist @GoogleBrain, @genlawcenter and @CornellCIS, formerly @Princeton @[email protected]Jack Rae @drjwrae
9K Followers 355 Following Principal Scientist @ Google DeepMind Work on Gemini 💎♊ Compression is all you need LLMs (e.g. Gopher, Chinchilla, Gemini) 💼 Past: OpenAI, QuoraKarina Nguyen @karinanguyen_
12K Followers 650 Following AI research & eng @AnthropicAI, prev. intern @nytimes, @square, @dropboxAndrew Drozdov @mrdrozdov
2K Followers 1K Following RAG at @MosaicML x @Databricks 🧱 Prev: @UMass_NLP, @Google, @IBMSvetak Sundhar @svetaksundhar
95 Followers 193 Following Data Analytics @ Google | Violinist | FlautistSharad Vikram @sharadvikram
1K Followers 510 Following Researcher @ Google Deepmind. I work on JAX + Pallas (https://t.co/lPMsq3yzgL) and Gemini. In the past I worked on Oryx and TFP. I like learning.ArtButMakeItSports @ArtButSports
454K Followers 230 Following I turn Art into Sports (and vice versa) | NO AI USED | “Everything I didn't know I needed" - follower testimonial | See inspiration? DM/tag usNoam Brown @polynoamial
34K Followers 612 Following Researching reasoning @OpenAI | Co-created Libratus/Pluribus, the first superhuman no-limit poker AIs | Co-created CICERO | PhD from @SCSatCMUStephen A Smith @stephenasmith
6.0M Followers 22 Following The real Stephen A. Smith. Host of First Take on ESPN and The Stephen A. Smith Show on YouTube. My book Straight Shooter is available nowchudnov @chudnovglavniy
3K Followers 695 Following building @3janexyz (eth/acc), prev strategy @aevoxyzChowdah Hill @ChowdahHill
68K Followers 195 Following Proud Captain of the best damn ship in the Navy, @TheCVN69. All views presented are mine and do not represent DoD/DoN. Follows/RTs/links ≠ DoD/DoN endorsement.Max ⛅ @maxisawesome538
2K Followers 3K Following sup nerds @DbrxMosaicAI @CohereForAI @riversideulti @maxdoesresearch for purely research tweetsmain @main_horse
8K Followers 478 Following AGI Believer. Haven't applied @OpenAI. Likes are not always endorsement.swyx @swyx
91K Followers 3K Following Anti-ego ideas for anti-ergodic life. Founder, @smolmodels ▹ Listen: @latentspacepod ▹ Read: @coding_career ▹ Join: @aiDotengineerZhengdong @zhengdongwang
508 Followers 226 Following Don't read non-narrative non-fiction / natural history @GoogleDeepMind / “economically very literate” —former MEPArchit Sharma @archit_sharma97
4K Followers 340 Following Final-year CS PhD student @Stanford. Previously, AI Resident @Google Brain, undergraduate @IITKanpur, research intern @MILAMontreal.Rylan Schaeffer @RylanSchaeffer
3K Followers 979 Following CS PhD student with @sanmikoyejo at @stai_research @StanfordAILabHailey Schoelkopf @haileysch__
3K Followers 815 Following she/her | research scientist @aiEleuther | LLM training/infra, eval, data | LM Evaluation Harness maintainerMechanical Dirk @mechanicaldirk
546 Followers 244 Following Principal Engineer at @allen_ai. Engineering Lead of the OLMo project.Jose Javier Gonzalez @jjgort
343 Followers 119 Following Research Scientist at MosaicAI DataBricks. Working on LLMsI'm really excited about ideas like this, but before people get too worked up you should know this seems to be a domain specific intervention. Thats *ok* though. This might be a very useful piece of making code models or adapting models to be code models. I'm also willing to…
Meta presents Better & Faster Large Language Models via Multi-token Prediction - training language models to predict multiple future tokens at once results in higher sample efficiency - up to 3x faster at inference arxiv.org/abs/2404.19737
Felt cute. Did some petabyte scale preprocessing. Might delete later.
I told my friends I was going this week to Colombia (with “o”)… but I got some funny looks. Friends: when written with “o”, I’m talking about the country. Don’t even think I’m in the middle of these protests in the place that goes with “u”.
Reporter grills Columbia student after she demands the university help feed protestors occupying Hamilton Hall: "It seems like you're saying, 'we want to be revolutionaries, we want to take over this building, now would you please bring us some food'."
I know my tweets have and will be hit or miss. Bear with me. We’re building the future.
I recently left @scale_AI. I'm so thankful to the team there and for @alexandr_wang's bet to acquire our startup nearly 4 years ago. When I joined Scale, it was a single-product company building the data engine for autonomous vehicles. It's amazing to see how far Scale has come:…
it ain’t such a long drop from being the guy who gets to put the tables into the PDF to being the guy who has to extract them
For the record, I pointed out this issue to the Amazon S3 team in 2006. They recommended keeping my S3 bucket names secret.
In the modern day Moby-Dick, Ahab catches the whale pretty easily. But then his life crumbles as he loses all meaning and purpose... If only he could figure out how to commercialize it into a b2b saas product.
Forget about MKBHD. Deep down, you knew that Humane and Rabbit weren't compelling products, but you were too afraid to say it because of that little voice in your head that whimpered, "but what if this disrupts Apple! I'm gonna look so stupid!" Welcome to venture capital.
NEW VIDEO - Rabbit R1: Barely Reviewable youtu.be/ddTV12hErTc This is the pinnacle of a trend that's been annoying for years: Delivering barely finished products to win a "race" and then continuing to build them after charging full price. Games, phones, cars, now AI in a box
it's joever. your $100m series A that you converted into an $80m AWS bill to dick around fine tuning llama 3 70b now needs to become a real product. you have 6 months of runway to make it happen and a small team of 20-somethings who've never held real jobs before. good luck.
I've been very disciplined with calling my shots, i've been waiting for this moment for almost a year. time to dump all your microchip stocks (not financial advice-ly)
this feels like the top of the bubble
“We’ve advanced this framework b/c innovators like imbue say it’s workable.” (Exact quote) If this is the level of diligence and discernment we can expect from our leaders, we don’t have to fear bad AI. We will be fucked before AGI is here.
1️⃣ SB 1047 is meticulously designed to set robust & attainable safety standards for frontier model developers. We’ve worked intensively w/ the startup community & have support of startups like @imbue_ai. We’ve advanced this framework b/c innovators like imbue say it’s workable.
Contrary to popular belief, MKBHD doesn’t destroy *all* hardware startups, in fact he gave @infinitemachine P1 a good review!
Zuck's social media game is on point
The American dream is remarkably still alive and can reach as far as being the CEO of a $2T company without founding it!
America has its flaws but remains the country where journeys like this happen more often than anywhere else
When AI was just an academic field it was actually quite cool, intellectual. When big business came in with huge amount of money and weird incentives it turned it into a rather dumb hype shit show. Now politics comes in, and it will turn it into a complete dumpster fire.
California Bill 1047 has been fasttracked: • Covers all models made w/ 10^26 flops • Covers all models with similar perf to above • Creates a Frontier Model Division to report to • Devs must assert such models are safe under penalty of perjury text: legiscan.com/CA/text/SB1047…
Blackthorne: To best position us for these opportunities a number of our teams made changes to become more efficient and work better, remove layers, and align their resources to their biggest product priorities. Mariko: The Anjin did layoffs to boost the stock price.