Maciej Obarski @KerbalFPV
Data Scientist at global cybersecurity company ◆ Applied AI / Computer Science / Tabletop Games Geek / FOSS proponent Warsaw, Poland Joined December 2014-
Tweets2K
-
Followers593
-
Following750
-
Likes8K
Thanks for implementing our paper! But actually, you only need to modify 5 lines of code to configure STORM with Claude models. ZERO line of change is needed now because I just added an example script to our repo! github.com/stanford-oval/…
Thanks for implementing our paper! But actually, you only need to modify 5 lines of code to configure STORM with Claude models. ZERO line of change is needed now because I just added an example script to our repo! github.com/stanford-oval/…
The dataset is everything. Great read: nonint.com/2023/06/10/the…
Within cells, interlinked.
llama3-42b: Pruned llama3-70b huggingface.co/chargoddard/ll…
I absolutely love this feature!
🥁 Llama3 is out 🥁 8B and 70B models available today. 8k context length. Trained with 15 trillion tokens on a custom-built 24k GPU cluster. Great performance on various benchmarks, with Llam3-8B doing better than Llama2-70B in some cases. More versions are coming over the next…
Introducing RWKV-6 "Finch" 3B v2.1🐦a very performant base model, and 100% RNN. 7B soon. Gradio Demo: huggingface.co/spaces/BlinkDL… Download: huggingface.co/BlinkDL/rwkv-6…
Introducing RWKV-6 "Finch" 3B v2.1🐦a very performant base model, and 100% RNN. 7B soon. Gradio Demo: huggingface.co/spaces/BlinkDL… Download: huggingface.co/BlinkDL/rwkv-6… https://t.co/SMzWyJwyRr
Today we are adding an important new capability to Poe: multi-bot chat. This feature lets you easily chat with multiple models in a single thread. (1/n)
Google presents Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention 1B model that was fine-tuned on up to 5K sequence length passkey instances solves the 1M length problem arxiv.org/abs/2404.07143
🌸 DataFans! This coming Streamlit feature beats even dialog window and partial reruns as my most awaited! We’re getting native Plotly & Altair chart selections soon 🤯
🚀New Preprint Alert! 📊Exploring the notion of "Zero-Shot" Generalization in Foundation Models. Is it all just a myth? Our latest preprint dives deep. Check it out!🔍 arxiv.org/abs/2404.04125
Have you ever wanted to train LLMs in pure C without 245MB of PyTorch and 107MB of cPython? No? Well now you can! With llm.c: github.com/karpathy/llm.c To start, implements GPT-2 training on CPU/fp32 in only ~1,000 lines of clean code. It compiles and runs instantly, and exactly…
"Through multiple controlled datasets, we establish that language models can and only can store 2 bits of knowledge per parameter, even when quantized to int8, and such knowledge can be flexibly extracted for downstream applications. Consequently, a 7B model can store 14B bits of…
the alpha right now is using claude opus to generate N shot examples from a seed and using them on haiku for cheaper/faster inference
New Anthropic research paper: Many-shot jailbreaking. We study a long-context jailbreaking technique that is effective on most large language models, including those developed by Anthropic and many of our peers. Read our blog post and the paper here: anthropic.com/research/many-…
Our new preprint is out! When visualizing high-dimensional data sets, t-SNE & UMAP often obscure temporal relationships between data points. This is our crack at it :) Visualizing High-Dimensional Temporal Data Using Direction-Aware t-SNE arxiv.org/abs/2403.19040 1/n
It seems that results of that Microsoft paper about ternary LLMs can be replicated after all – for 3B@100B at least. huggingface.co/1bitLLM/bitnet…
Oh snap, crackle… 🍿 st.popover is live! Get more control in your UI with button-triggered, expandable containers. It's perfect for adding tooltips, filters, or custom menus without cluttering the app. (Sorry, butter not included.) Check it out: buff.ly/3x3O35P
Super thrilled to release our work on extreme quantization (1-bit and 2-bit)! We're starting with the Llama2-7b since it's a well-understood model. Check out our detailed blog post: mobiusml.github.io/1bit_blog/
Julian @JulianJulian99
16 Followers 3K Following7etsuo — e/acc 🏴.. @7etsuo
6K Followers 934 Following Code in C. Excessively driven developer, constantly labouring on self-improvement. https://t.co/59HNLDwQxE https://t.co/vvOVbBctp9 https://t.co/muCKwFmvP9AI Agents Global Chal.. @aiagentsglobalc
155 Followers 626 Following One of the largest AI challenges globally, inviting participants to compete in developing AI AgentsMiroslav Gospodinov @MiroslavGospod6
5 Followers 15 Following Born with a passion for turning ideas into reality, bottom-up build entrepreneur🤯 📚, I've always dreamt of building something extraordinary 🎯T-Motor Hobby @TMotor_Hobby
43 Followers 409 FollowingShojaei @realshojaei
1K Followers 2K Following AI Researcher | building AI Agents & LLM applicationsJason Canipla @JCanipla
107 Followers 295 Following SWE interested in LLMs, OpenSource, Side projects, and self optimizationRamona #riseandfall @ramonaundecided
390 Followers 668 Following Rise & Fall @channel4 | Season 1 UK | Enquiries: [email protected] LadyBo$$ 🫦 Instagram: @ramona.undecided @undecided_bar Talk Show: @the.uncut.showsaikat bank @SaikatBank
0 Followers 26 FollowingQ @tong0798
0 Followers 14 Followingssteevens @Steevens43
151 Followers 4K FollowingJasonfpv @monika_meps
395 Followers 2K Following 毎日のFPVフライト&スリル!廃墟、カーチェイス、エクストリームスポーツ、レース- すべてここにあります!初心者向けFPVドローンキットの推薦も!🚁✨#fpvClaudia 💋 @Claudia__M84
2 Followers 452 Following Рassionаtе temptrеss relentlеsslу sееking sаtisfаctiоnLeatheddat @leatheddat80560
125 Followers 2K Followingfly51fly @fly51fly
5K Followers 2K Following BUPT prof | Sharing latest AI papers & insights | Join me in embracing the AI revolution! #MachineLearning #AI #InnovationJames X @Jamesx72845781
150 Followers 2K FollowingJohnSnowLabs @JohnSnowLabs
41K Followers 30K Following Helping healthcare and life science organizations put AI to work faster with state-of-the-art LLM & NLP.researcher Gpt LLM @researchGptllm
237 Followers 4K FollowingNURETTİN GÜLAYDIN @ngulaydin
169 Followers 1K FollowingBeans24 @labeans24
0 Followers 726 FollowingDr. Tesla @docdima
883 Followers 4K Following MD, Neuro Scientist. MRI expert. Also EV, AI and robotics enthusiast. 🇨🇦 Tesla 🚀AI Turbo dev 🇿🇦.. @AI_GPT42
177 Followers 1K Following Computer Science dude that solves problems. Focus on AI news and techniques to get the best results from LLMs. Future AGI will reveal meaning of life beyond 42!Saba G. Haez @Saba_Gh_Hz
129 Followers 143 Following Research PhD. @FBK_research & @eng_unibz in Artificial Intelligence, Language Modeling, Knowledge Integration, AI in Healthcare, NLP, LinguisticsJohn Owen @dreamingElvis
752 Followers 3K Following 18th century liberal…oh yeah: “under the lawn of His palace of Graceland dead Elvis waits dreaming, yet He shall rise and His Kingdom shall cover the Earth.”Nilay Shah @nilayspshah
107 Followers 3K Following Programmer, Technology, Finance. Bug fixes and Performance optimizations.Anopsy @Anopsy
193 Followers 722 Following painting faces on walls in Amsterdam / Dutch Teacher / wannabe ML EngineerMax Zanoga @zanoga
209 Followers 926 Following 🚀 AI Enthusiast | 🔍 Exploring the frontiers of LLMs & ML | 📚 Lifelong learner & tech optimistSankeerthana Keerthan.. @Sankeerthana204
2 Followers 72 FollowingSarbjeet Johal @sarbjeetjohal
31K Followers 20K Following #Think24 - 28 yrs in #SiliconValley #Cloud, #Analyst, #GTM, #Coder, Economist, Storyteller, Speaker, Investor. EX: @OracleCloud @Rackspace @VMWare #DelEMC @VISAAlexander The Great @xandergram_01
639 Followers 2K Following Cyber security 🧑💻| ISC2 Member 🥇| UI/UX DESIGNER | Software Engineer 🚀ALL-Samanta Leitão @Sellel170055
3 Followers 740 Following Melhores de meio período,junte-se a All,ganhe mais de 200 por dia,se você estiver interessado,tome a iniciativa de conversar comigo em particular!Poucket @Poucket171070
0 Followers 84 Following Apaixonado por criptomoedas e inovações financeiras. Junte-se a mim no caminho para o futuro do dinheiro digital!Outcast Droids @outcastdroids
50K Followers 700 Following Building $DROIDS Ecosystem for an AI-driven future – Invest in a Vision, Own a Masterpiece🤖 https://t.co/JX4yGkjrXi ⚡️ Play-2-Earn https://t.co/EGIYmJg9Rs📲松倉善浩 @y_matsukura
19 Followers 92 Following スマートデバイス・iPhone・iPad・Android・Face book・仕事効率化・東京在住・日本人どれかに当てはまればフォローよろしくです!(^_-)気軽にフォローして下さい。最近の趣味:マラソン・水泳・庭いじり・evernoteなど。Rita @rita38carlson
133 Followers 3K FollowingDaniel Lemire @lemire
21K Followers 1K Following Computer science professor at @TELUQ (Université du Québec, @ReseauUQ), programmer, blogger. @Nasorg👨💻Casey Muratori @cmuratori
41K Followers 120 Following I want all my garmonbozia. https://t.co/Bdh1Xj2PpV7etsuo — e/acc 🏴.. @7etsuo
6K Followers 934 Following Code in C. Excessively driven developer, constantly labouring on self-improvement. https://t.co/59HNLDwQxE https://t.co/vvOVbBctp9 https://t.co/muCKwFmvP9Carmen Gutierrez @carmguti
7K Followers 1K Following founding @ResiduaAI, prev ms cs @StanfordEng, physics @stanfordLightning AI ⚡️ @LightningAI
39K Followers 96 Following The AI development platform - From idea to AI, Lightning fast ⚡️. Creators of AI Studio, PyTorch Lightning... Get help: https://t.co/a69wnEARV9Rohan Paul @rohanpaul_ai
12K Followers 840 Following ML Engineer (e/acc) 📌 https://t.co/x0IIWfnOt8 🚀 https://t.co/QEO4CKRl1b Open LLMs is Happiness 💡 Ex Deutsche & HSBC. DM for collaboration.Mobius Labs @Mobius_Labs
3K Followers 105 Following Multimodal AI for the world's scale. Proponents of Open Source and Open Intelligence. https://t.co/1nC6r8hOrE for some of our recent work.Salesforce AI Researc.. @SFResearch
13K Followers 117 Following #SalesforceAI advances state-of-the-art #AI techniques that pave the path for innovative products at Salesforce. Focus areas include #ML, #NLP, #AIforGoodOsho Priya @osho_priya
140 Followers 768 Following DevOps Engineer📈AI/ML/Cloud💫NASA L'space Alumna| full-time nerd| https://t.co/SC9SMGFe76…Evan Miller @EvMill
5K Followers 160 Following Statistically inclined software developer, occasional blogger about math + stats stuffMarie Pellat @m_pellat
1K Followers 197 Following AI Scientist, @MistralAI - previously Google // Responsible AIJeremy Howard @jeremyphoward
222K Followers 5K Following 🇦🇺 Co-founder: @AnswerDotAI & @FastDotAI ; Hon Professor: @UQSchoolITEE ; Digital Fellow: @StanfordInflection AI @inflectionAI
49K Followers 3 Following We are an AI studio creating a personal AI for everyone. Our first is @pi, a supportive and empathetic conversational AI.Amanda Askell @AmandaAskell
26K Followers 653 Following Philosopher & ethicist teaching models to be good @AnthropicAI. Personal account. All opinions come from my training data.Surfshark @surfshark
21K Followers 211 Following Empowering people to take control of their digital lives. Ready to take it into your own hands? Join our community: https://t.co/BUqg4FkXlsKy⨋ Gom⨋z (U/ACC).. @KyeGomezB
2K Followers 556 Following Founder of Agora & Swarms Github https://t.co/naFqnYkuQ0 Join Agora, the open source multi-modal AI research lab 👇 https://t.co/hfaUSCgNjBElon Musk @elonmusk
181.4M Followers 585 FollowingBhavya Kailkhura @bkailkhu
576 Followers 1K Following Research Scientist @Livermore_Lab. Making AI/ML Robust & Efficient for National Security and Scientific Applications.Subbarao Kambhampati .. @rao2z
16K Followers 29 Following AI researcher & teacher @SCAI_ASU. Works on Human-Aware AI. Former President of @RealAAAI; Chair of @AAAS Sec T. Here to tweach #AI. YouTube Ch: https://t.co/4beUPOmMW6fly51fly @fly51fly
5K Followers 2K Following BUPT prof | Sharing latest AI papers & insights | Join me in embracing the AI revolution! #MachineLearning #AI #InnovationSaulnier Lucile @LucileSaulnier
4K Followers 432 Following AI Specialist @ Mistral AI | Former ML @ Hugging Face | ENS Paris-Saclay (MVA) | Centrale ParisThePrimeagen @ThePrimeagen
176K Followers 814 Following skill issues: 🟩⬛️⬛️⬛️⬛️⬛️(69/420) https://t.co/qvibjmtqFm https://t.co/gVnVukutsfCybergibbons 🚲🚲.. @cybergibbons
49K Followers 3K Following Head of hardware. IoT hacker. Alpha Male.Mihai Chirculescu @m_chirculescu
123 Followers 289 FollowingTony Dinh 🎯 @tdinh_me
125K Followers 853 Following Indie hacker. 🧠 @TypingMindApp $33K/mo 🛠 @devutils_app $5.5K/mo 🪄 @blackmagic_so (Acquired, 2023) 📸 @XnapperHQ (Acquired, 2024)Ryan Kramer 🛠️+�.. @ryankramerllc
2K Followers 474 Following Institutionally Certified Top Engineer, Allergic To Cubicles, 🐍 beats ☕️, React is the right JS framework, certified by Google, Oracle, AWS and 💰Peter Voss @peterevoss
874 Followers 260 Following Creating AGI: https://t.co/lKckFk1CTo https://t.co/O1HZFZOUV9Roko @RokoMijic
10K Followers 214 Following Radical Centrist, Transhumanist We wanted eternal life, instead we got 140 new genders. Join me at the link below for decentralized AI alignment:XR-5 🐀 @xariusrke
819 Followers 251 Following High school dropout. Scaling neural networks to massive scale @huggingface. DMs open.1LittleCoder💻 @1littlecoder
12K Followers 1K Following AI, ML, Open Source at - https://t.co/EKsvaArRIkoutlines @OutlinesOSS
418 Followers 15 Following Open source software for guided generation in Python. Supported by @dottxtai. Created by @BrandonTWillard and @remilouf.Visual Studio Code @code
684K Followers 133 Following Microsoft Visual Studio Code lets you build and debug modern web and cloud applications. Visual Studio Code is free and available on Linux, macOS, and Windows.Philipp Schmid @_philschmid
16K Followers 651 Following Tech Lead and LLMs at @huggingface 👨🏻💻 🤗 AWS ML Hero 🦸🏻 | Cloud & ML enthusiast | 📍Nuremberg | 🇩🇪 https://t.co/l1ppq3q3hkElle Cordova @ellerhymes
31K Followers 523 Following Writer. Musician. Sometimes funny. She/they. 🏳️🌈 YouTube ⬇️RWKV @RWKV_AI
2K Followers 3 Following AI model built by the community, for everyone in this world Part of the Linux Foundation, Apache 2 licensed An RNN scaled to 14B params with GPT-level of perfTill Döhmen @tdoehmen
119 Followers 83 Following 🤖 AI/ML 🦆 @motherduck, part-time PhD stud. in Databases/ML https://t.co/g76bBKI9J8MotherDuck @motherduck
5K Followers 121 Following Making analytics fun, frictionless and ducking awesome with a serverless easy-to-use data analytics platform based on @DuckDB in collab with @duckdblabs.Prateek Yadav @prateeky2806
2K Followers 2K Following Ph.D. at @unccs Continual Model Adaptation and Composition Previously @MSFTResearch, @AmazonScience, @iitmadras. UG @iiscbangalore. Opinions are my own.Amanda Watson @fewerwrong
5K Followers 290 Following VR developer, prev. @oculus. Creator of CitraVR and Oculus Air Link. Worked on Oculus Quest since very early days.Soon or later maybe it will be possible to talk again about making a new version of C with a little bit of sane things from C++, with a solid standard library, a bit more safety, well defined corner cases, and so forth. For sure that day is not today.
@cloverhogan Some say eco-catastrophism is a Trojan horse for communism. Thank you for confirming it.
Disney’s new "Holotile" real-time moving VR walking floor is insane. 95% of humanity will be more addicted to the Metaverse once this is rolled out.
Lets do compositions! These are guides, that can help you get started. And make your compositions easy to follow and readable. A thread with more examples! 🧵 1/10
When you start to see @ThePrimeagen everywhere
Knowledge and culture gets lost easily. Preservation is essential. Some are trying to sue the Internet Archive out of existence. Are they modern day barbarians @brewster_kahle ?
How much human knowledge has been lost to history? Well, this ancient wonder was razed by the Mongols in 1258 — it's said the Tigris River ran black with ink. Here's what was inside... (thread) 🧵
Thanks for implementing our paper! But actually, you only need to modify 5 lines of code to configure STORM with Claude models. ZERO line of change is needed now because I just added an example script to our repo! github.com/stanford-oval/…
STORM by @angelina_magr @MehdiAllahyari Implementation of the paper STORM (Synthesis of Topic Outlines through Retrieval and Multi-perspective Question Asking) -- uses Claude + sub-agents to write long-form articles. github.com/angelina-yang/…
From Claude100K to Gemini10M, we are in the era of long context language models. Why and how a language model can utilize information at any input locations within long context? We discover retrieval heads, a special type of attention head responsible for long-context factuality
Retrieval heads explains why the model hallucinate: when the model misses part of the information, the retrieval heads are not activated; when the model cannot find the needle and hallucinate completely, retrieval heads attend to dummy sink tokens.
Incredibly interesting work.
Retrieval heads explains why the model hallucinate: when the model misses part of the information, the retrieval heads are not activated; when the model cannot find the needle and hallucinate completely, retrieval heads attend to dummy sink tokens.
Llama 3 extended to almost 100,000-token context! ✅ By Combining PoSE and continuing pre-training on Llama 3 8B base for 300M tokens, the community (@winglian) managed to extend the context from 8k to 64k. 🚀 Applying rope scaling afterward led to a supported context window of…
In AI research there is tremendous value in intuitions on what makes things work. In fact, this skill is what makes “yolo runs” successful, and can accelerate your team tremendously. However, there’s no track record on how good someone’s intuition is. A fun way to do this is…
Surprisingly, China is the largest publisher of material science papers by at least an order of magnitude, yet it is not at the forefront of computing. Something to ponder about.
Just this week multiple founders, engineers, and researchers I’ve talked to have noted that the narrative around AI here on Twitter is disconnected from the reality on the ground. Will be interesting to see the outcome of this.
We’re going to get AI models that are much better at reasoning within 6mo. Do *everything* you can to prepare your workflows for that moment. On day 1 you’ll be moving 200mph while everyone else is scrambling. Massive advantage.
@realmrfakename I have several PRs in flight there. @ggerganov has been incredibly supportive of my work and I'm proud to contribute to his project too! github.com/ggerganov/llam…
I've just released llamafile v0.8 which features LLaMA3, Mixtral 8x22b, and Grok support. It goes 25x faster than ollama at running LLaMA3 70B on CPU. My new tensor multiplication kernels let llamafile eval MoE models 2x faster than llama.cpp github.com/Mozilla-Ocho/l…