LDJ @ldjconfirmed
e/λ Currently: Working on something new Prev: @NousResearch @TTSLabsAI DM for business/consulting or interesting conversations. huggingface.co/LDJnr S4 Joined March 2021-
Tweets308
-
Followers5K
-
Following199
-
Likes305
Zephyr-ORPO-141B is the first model I've seen get this consistently right about what JEPA actually stands for. I tried this even with Claude-3-Opus and it fails too, and even the latest GPT-4-turbo fails! I checked the fine-tune dataset and it has no mention of JEPA either.
The date is December 15th 2024. You're outside while wearing the open source frame glasses. You have hands-free communication with Open interpreter and your multi-modal LLama-3 instance running on your desktop, it's finetuned on the Capybara V2 dataset generated with GPT-4.5
Has nobody tried using Claude 3 Opus as an agent yet and seeing how much better than GPT-4 it might be? Maybe in something like AutoGPT? Open Interpreter? ChatDev? AI Town?
"I am extremely skeptical of people who think only their in-group should get to know about the current state of the art because of concerns about safety, or that they are the only group capable of making great decisions about such a powerful technology." - Sam Altman 2022
Mamba-former MoE model with byte level multi-modal JEPA understanding. Can do input and output of Images, video, voice, foley, robotic movement data and more. Runs on a Photonic Neuromorphic Thermodynamic Quantum Hybrid chip. Sources say it's dropping soon👀 (It's a joke)
If you're doing a lot of fine-tuning and dataset curation, definitely make sure to check out Lilac Garden. They were nice enough to run Capybara through it before official release and allowed me to see interesting insights that normal embedding clustering typically fails to show.
If you're doing a lot of fine-tuning and dataset curation, definitely make sure to check out Lilac Garden. They were nice enough to run Capybara through it before official release and allowed me to see interesting insights that normal embedding clustering typically fails to show. https://t.co/x9sK8idWn5
🚀 The OSS AI community needs more open datasets for improving LLMs: 🎁 Excited to ship a new open DPO dataset for boosting chat models: ⚗️ distilabel capybara-dpo, a multi-turn preference dataset built atop the awesome dataset by @ldjconfirmed huggingface.co/datasets/argil… 🧵
awesome progress by @thtrieu_ and team on a problem which is close to my heart - this result is important because it shows that under special conditions, we can use automated reasoning and goal-relabeled synthetic data to train an expert level math solver!
awesome progress by @thtrieu_ and team on a problem which is close to my heart - this result is important because it shows that under special conditions, we can use automated reasoning and goal-relabeled synthetic data to train an expert level math solver!
Anyone have a preferred combination of LLM "inference controls" that you use? For example the specific set of values you might prefer to use for temperature, top_k, rep_p etc.. It seems that most agree it could have a significant impact on outputs but isn't talked about enough.
Shane o @Shaneo433048028
15 Followers 84 FollowingRed @redthefox_
0 Followers 233 Following Fluffy dev spé. ML-DL-IA. Je joue beaucoup avec des LLMs pour un grand groupe. T'as peut-être croisé ma Type R 🦊 un peu trop vite. Aucun follow accepté.fantasytrader12 @cryptoxgenie
26 Followers 391 FollowingLeaLea @GenevieveGuetta
800 Followers 1K FollowingJohn Michael @Mikel_Johnn
185 Followers 865 Following Data Scientist | Machine Learning Engineer | Tech EnthusiastMister Lenny @MisterLenny4
543 Followers 998 Following Me bloquer c’est reconnaître sa faiblesse…. Free Palos vous me faites trop rigoler 😂Weyaxi @Weyaxi
2K Followers 2K FollowingEthan @Ethan_smith_20
3K Followers 687 Following a boy and his gpu vs the world. directing research at @leonardoai_. learning as I go. uf psych. generative models and representation learningLord Travis Wright @teedubya
186K Followers 60K Following Futurist. Author. Marketer. Advisor. Speaker. Philomath. Sentient. Explorer. Podcast🎙@badcryptopod Building: https://t.co/68voU6yZ11 | https://t.co/zxRDT7AzMSRachid @rachidasimi
90 Followers 333 FollowingRohan Paul @rohanpaul_ai
13K Followers 1K Following ML Engineer (e/acc) 📌 https://t.co/x0IIWfnOt8 🚀 https://t.co/QEO4CKRl1b Open LLMs is Happiness 💡 Ex Deutsche & HSBC. DM for collaboration.วิไลเมื.. @aN5bKqTALJuMC2n
72 Followers 1K Following ความเซ็กซี่มีมากกว่าหนึ่งด้าน ติดตามฉันและค้นพบช่วงเวลาอื่นๆ ที่จะทำให้หัวใจคุณเต้นเร็วขึ้น! หน้าแรกของข้อมูลการติดต่อจะได้รับการอัปเดตตลอดเวลาBabs Khalidson @babskhalidson
367 Followers 178 Following Machine Learning Lead @CMCMarkets alum @durham_uni & @univofstandrewsShawn Charles🎤🔥 @ShawnBasquiat
32K Followers 3K Following 🧑🏾💻Ex-FAANG Software Engineer 🥑Senior ML Developer Advocate @ Coming Soon 🏗️Building Tech CommunitiesFlotos - Rainy Craft .. @flotosor
79 Followers 449 Following Craft your unique items in this deep roguelike auto-battler, Rainy Craft ! https://t.co/UZrj5XZ9WtAnshul @TechSavvyAS
63 Followers 325 Following 🥑 Keep it Simple Let's connect and explore together💙 Don't mind I am just being sarcastic.Bheeshma On The Compu.. @DroidIsLove
62 Followers 1K FollowingJohannes Bubenzer @joh_bub
14 Followers 38 Followingaaaa @weqiocre
55 Followers 221 Following321 @32100
1K Followers 388 Following$CRYPTOMONSTER/ SHILL.. @THE_BEEFSHILLER
186 Followers 860 Following TRYING TO MAKE YOU A MILLIONAIRE🍪 DM FOR CULTURE CALL 📢 @GIICMONSTERHorses for Everyone�.. @DarcyButcher
428 Followers 2K Following Writer, Photographer, Artist BTC: 14Ch4uDga237uLK2TxU5raQxfUbAAvBqVaAmr Nader @ayyyynad
0 Followers 41 FollowingA Techno Optimist @atechno0ptimist
14 Followers 43 FollowingBen Holfeld @BenHolfeld
89K Followers 32K Following SF AI Studio Lead @Accenture, partnering with @OpenAI @Google @Microsoft. Pianist. German Quantum Physicist. Creator of the Nth Floor. Views are my own. x/acc.Dmytro Kotenko @kotenko_dmitrij
150 Followers 947 FollowingGautham @ALongDeadStar
324 Followers 2K Following A tiny speck of star dust suspended in an infinite cosmos 💫🪐Shojaei @realshojaei
1K Followers 2K Following AI Researcher | building AI Agents & LLM applicationsDr. Yu-Dai Tsai @YuDai_Tsai
2K Followers 4K Following Incoming Director's Fellow @LosAlamosNatLab; Postdoc @UCIrvine; Formerly @Fermilab @UChicago. https://t.co/lPYqoPpt0v https://t.co/je5EsvIWXodelta @Deltanomicss
1K Followers 580 Following stutterer building AI autocorrect for speech with ML activated transcranial direct current stimulation. ex stuff @CERN start-up | @iealondon | @ASIMD @maxedel11
56 Followers 136 Followingtrebbn @trebbn
112 Followers 267 FollowingDhruv Chawla @dhruvchawla369
12 Followers 231 FollowingPrayogi @Tweetputtar
187 Followers 4K Following Eternal Student | Wannabe Scientist | All things ridiculous to sublimeNoam 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 @SCSatCMUDr. Yu-Dai Tsai @YuDai_Tsai
2K Followers 4K Following Incoming Director's Fellow @LosAlamosNatLab; Postdoc @UCIrvine; Formerly @Fermilab @UChicago. https://t.co/lPYqoPpt0v https://t.co/je5EsvIWXoCollin Burns @CollinBurns4
11K Followers 276 Following Superalignment @OpenAI. Formerly @berkeley_ai @Columbia. Former Rubik's Cube world record holder.Lewis Tunstall @_lewtun
9K Followers 425 Following 🤗 LLM engineering & research @huggingface 📖 Co-author of "NLP with Transformers" book 💥 Ex-particle physicist 🤘 Occasional guitarist 🇦🇺 in 🇨🇭Samuel L Smith @SamuelMLSmith
2K Followers 361 Following Research Scientist at DeepMind. Optimization and Initialization. Formerly Google Brain. Ex-Physicist.Pulley @pulley
4K Followers 47 Following Everything you need to issue and track equity. Get 409A valuations, cap table management, and equity advice all in one system.MC HAMMER e/acc @MCHammer
3.1M Followers 68K Following #OAKLANDFIGHTCLUB #Dubnation #RaiderNation #AI #Hamm400 #Science #Consciousness #QuantumPhysics #Dogon #Philosophy #AncientEgypt #FilmMaker #Art #AGISholto Douglas @_sholtodouglas
15K Followers 858 Following Scaling Gemini @Deepmind - working towards intelligence too cheap to meterEric Steinberger @EricSteinb
7K Followers 478 Following Writing code that writes code on a mission to build safe superintelligence | CEO/cofounder @magicailabsJames Bradbury @jekbradbury
11K Followers 8K Following Compute at @AnthropicAI! Previously JAX, TPUs, and LLMs at Google, MetaMind/@SFResearch, @Stanford Linguistics, @Caixin.Dwarkesh Patel @dwarkesh_sp
55K Followers 700 Following Being pretrained Host of Dwarkesh Podcast https://t.co/3SXlu7fy6N https://t.co/rEhnfYywXY https://t.co/hQfIWdM1UnLighthouse @lighthousehq_
531 Followers 4 Following Betting on America 🇺🇸 We help the world's best and brightest get a US work visa, fast, so you can get back to buildingNobodyExistsOnTheInte.. @nullvaluetensor
36 Followers 40 Following Human Large Language model. Skills: Distill data. Training LLMs. Test and Evaluate. Rinse and repeat as required. Based in SEA.Jiwoo Hong @jiwoohong98
203 Followers 80 Following Master's Student at @kaist_ai, interested in NLP, LLM, and any related topicsNora Belrose @norabelrose
8K Followers 124 Following Working toward a free and fair future powered by friendly AI. Head of interpretability research at @AiEleuther, but tweets are my own views, not Eleuther’s.Omar Sanseviero @osanseviero
32K Followers 2K Following Chief Llama Officer @huggingface 🦙 Founder @AI_Learners. Xoogler (SWE @Google Assistant, 20% PM TF Graphics). 100% Hacker Llama🇵🇪🇲🇽Ben Newhouse @newhouseb
7K Followers 955 Following @openai, https://t.co/i3YR3e9UMT, former head of sync @ dropbox (till 2018), cofounded bubbli (acquired by dropbox), previously made yelp monocle.Cognition @cognition_labs
123K Followers 19 Following Makers of Devin, the first AI software engineer. We are an applied AI lab focused on reasoning, and code is just the beginning. Join us: https://t.co/tpfZwEwGiqJade @Euclaise_
2K Followers 350 Following ⋅ Video game statistician ⋅ Soclib cyberanarchist? ⋅ C, Plan 9, LLMs, etc ⋅ Researcher w/ @NousResearch ⋅ she/theysurya @sdand
10K Followers 665 Followingchef jeff @chefjeffsf
8K Followers 997 Following founder of Eat Blueprint // prev founder @athensresearch (yc w21)Logan Kilpatrick @OfficialLoganK
92K Followers 2K Following Lead product for @Google AI Studio and working on the Gemini API, helping developers build with AI, my views!George Hotz 🌑 @realGeorgeHotz
248K Followers 174 Following President @comma_ai. Founder @__tinygrad__Sasha Rush @srush_nlp
52K Followers 464 Following Professor, Programmer in NYC. Cornell Tech, Hugging Face 🤗 https://t.co/cZl0wTfqGzAdrian Dittmann @AdrianDittmann
71K Followers 875 Following Life is too short to worry about stupid things. Have fun. Fall in love. Regret nothing, and don't let people bring you down. Study, think, create, and grow.Hamel Husain @HamelHusain
23K Followers 2K Following Researcher focusing on LLMs: https://t.co/iVZDFdIQiE Previously, dev tools and infra for ML. Ex @Github, @Airbnb, @DataRobot. @fastdotai core contributor.Ramin Hasani @ramin_m_h
3K Followers 258 Following Cofounder & CEO https://t.co/fh9fnDA9OQ | ML Researcher @ MITAI at Meta @AIatMeta
532K Followers 255 Following Together with the AI community, we are pushing the boundaries of what’s possible through open science to create a more connected world.Clémentine Fourrier .. @clefourrier
3K Followers 302 Following Leaderboards & evals research @HuggingFace 🐍✨ "The future is already here, it’s just not very evenly distributed" (Gibson)AndriyMulyar @andriy_mulyar
11K Followers 517 Following building tech that enables humans to interact with latent spaces 🗺️ founder / cto @ https://t.co/NbsLHLWfy8 prev. ML Ph.D. Student at NYU CourantJim Fan @DrJimFan
229K 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.Guessing what the gpt2 is as boring as guessing what the q* is…
I always strongly suggest people to read this work (arxiv.org/abs/2207.10551) by @YiTayML and @m__dehghani when discussing the model architecture. It almost takes up to 50% pages of the literature survey Chapter in my PhD thesis. It is so visionary to study this in 2022. I can…
not true, especially for language. if you trained a large & deep MLP language model with no self-attention, no matter how much data you'll feed it you'll still be lacking behind a transformer (with much less data). will it get to the same point? i don't think so. your tokens…
Mark Zuckerberg: "I do think in the future, it seems quite possible that more of what we call training for these big models is actually more along the lines of inference generating synthetic data to then go feed into the model."
🦙 OrpoLlama-3-8B Successful ORPO fine-tune of Llama 3 with ChatML template! It was trained on 40K high-quality preference samples for 3 epochs. Sharing some details and benchmarks. 🧵 🤗 Model: huggingface.co/mlabonne/OrpoL… 🪟 Demo: huggingface.co/spaces/mlabonn…
@ldjconfirmed Agreed. He literally starts out saying the models are still improving when they stopped but they had to stop to train the next one. Didn’t sound constrained, except for the compute/power. Not pessimistic at the potential.
@ldjconfirmed Do not fall for Ate-a-Pi's engadgetment farming lmao, he knows exactly what he's doing here hahaha
@ldjconfirmed I had watched the podcast before I saw the post and when I eventually happened upon it was also quite surprised. Zuck seemed bullish and all in.
@ldjconfirmed I had the same reaction as you. Honestly, I think Zuck is just now getting up to speed and focusing on AI 6-9 mos ago he was was surprisingly not immersed in it But I do think Meta will be one of 4 or 5 vying for ASI supremacy in the cloud.
there’s non zero transfer between being good at product obsession and being good at dataset obsession
Maybe there are little flat spots along the way, but AI in 5 years will be ridiculously capable compared to AI today.
On release, world-sim experienced an unregistered confluence in the Stripe/simulation singularity. Please stand by for quantum phase realignment.
I'm so excited to see this finally released! It took me a lot of effort and care to build the initial version (same care as the original Capybara by @ldjconfirmed) Difficult to build but also difficult to show people why it was important and useful, so big thanks @jiwoohong98…
🥁 Launching a new dataset: Capybara-Preferences, built with distilabel 1.0 ⚗️! Hard at work fine-tuning Llama 3? Here's the dataset you've been waiting for. Initial results with ORPO & this dataset are 🔥 huggingface.co/datasets/argil… 🧵What makes this dataset so special?
Mom says we have GPT-4 at home The GPT-4 at home: huggingface.co/lmstudio-commu…
🦫 We have just released `Capybara-Preferences` in collaboration with @kaist_ai and @huggingface A new synthetic preference dataset built using `distilabel` on top of @ldjconfirmed Capybara dataset More details 🧵 huggingface.co/datasets/argil…
Air travel needs to become 100x cheaper, safer and higher capacity. Mostly because it would be great to be able to fly with my dog and not need an unaffordable private jet. Can you imagine. Right now travelling with larger pets is reserved for the ultra rich. 🤔