Philipp Singer @ph_singer
Senior Principal Data Scientist @h2oai | PhD in CS Top ranked Kaggle Grandmaster (Highest #1) All views are my own. https://t.co/NHdaca2ld0 philippsinger.com Wien, Österreich Joined May 2009-
Tweets2K
-
Followers12K
-
Following464
-
Likes2K
Not sure Microsoft did themselves a favor only releasing phi-3-mini. Lots of unhappy vibe testers might just be running into limits of smaller models. Comparison should be models in the same parameter range. But the benchmarks give the impression of it being closer to 7b models.
Finishing second in recent @kaggle competition /w @pa_pfeiffer & @ybabakhin, trying to recover the prompt used to transform a given text with gemma. Detailed solution: kaggle.com/competitions/l…
My whole feed is full of the new Llama3 benchmarks table from the release post and it looks impressive. I thought a quick test with HuggingFace Open LLM Leaderboard eval can put the base model a bit more into perspective.
So many new devices and apps focus on speech which to me personally requires more mental effort vs. just typing something in, specifically after talking already a lot during the usual daily routine. Causes much more fatigue, but I assume everyone is different there.
Excited to release H2O-Danube2-1.8b - a new and improved language model being the best in class below the 2B parameter range. The model is trained on additional 2T tokens (total of 3T tokens) with various data mix stages as a result of extensive experimentation. Focus is on…
Excited to release H2O-Danube2-1.8b - a new and improved language model being the best in class below the 2B parameter range. The model is trained on additional 2T tokens (total of 3T tokens) with various data mix stages as a result of extensive experimentation. Focus is on…
Given the quality of this year's April Fools' pranks, I wonder how many of them have been LLM generated.
Here is my naive assumption: They noticed with mixtral that removing sliding window is actually useful for longer context, as well as increasing rope theta. So they also decided to release a new mistral finetune based on a base model that was adjusted for this, maybe even the one…
Here is my naive assumption: They noticed with mixtral that removing sliding window is actually useful for longer context, as well as increasing rope theta. So they also decided to release a new mistral finetune based on a base model that was adjusted for this, maybe even the one…
Sometimes I am not convinced we are not in a simulation. What are the odds?
Sometimes I am not convinced we are not in a simulation. What are the odds?
The biggest optimization problem in ML/AI is to optimally schedule your next training run so that it finishes at 11pm and not 3am for you to start the next one. Becomes infinitely more complex planning multiple runs ahead. Optimal usage out of all possible gpu hours is hard.
Are people still having issues fine-tuning Gemma? I just tried baseline experiment in H2O LLM Studio using oasst dataset and it worked out of the box as expected without even using bos token or any of the default template tokens.
Welp...
EU regulation wins again. You cannot click anylonger on the small map for locations on Google to direct you to Google Maps directly (or reviews, hotels, etc.). Reason is new regulations that do not allow Google to gatekeep when presenting search results. theregister.com/2024/01/18/goo…
Apparently current HF transformers Mistral implementation has default attention set to sdpa, but does not use sliding_window. I was seeing weird behavior above 4k context for a while. So if I am not missing something, all the recent transformer versions had buggy behavior for…
Okay, gemma tokenizer has so many tokens to include important tokens such as "Philipp", even spelled as it should be. I am absolutely impressed.
Okay, gemma tokenizer has so many tokens to include important tokens such as "Philipp", even spelled as it should be. I am absolutely impressed. https://t.co/K0nq8zU4h7
From what I get from the paper and original implementation, gemma was trained with tie word embeddings. But the HF weights and implementation have embedding and lm_head weights duplicated. Or do I miss sth there @art_zucker?
The only real puzzling thing in the Gemma paper to me is the use of a vocab size of 256k although English only training data. This is significantly larger than related models and sounds quite excessive to me.
Sebastian Raschka @rasbt
267K Followers 906 Following Machine learning & AI researcher writing at https://t.co/A0tXWzG1p5. LLM research engineer @LightningAI. Previously stats professor at UW-Madison.Mark Tenenholtz @marktenenholtz
115K Followers 546 Following Head of AI @PredeloHQ. XGBoost peddler, transformer purveyor.Sanyam Bhutani @bhutanisanyam1
35K Followers 994 Following 👨💻 Sr Data Scientist @h2oai | Previously: @weights_biases 🎙 Podcast Host @ctdsshow 👨🎓 International Fellow @fastdotai 🎲 Grandmaster @KaggleRadek Osmulski 🇺�.. @radekosmulski
25K Followers 555 Following Resources to take your Machine Learning skills to the next level 🧪 Senior Data Scientist, RecSys @NVIDIAAI 🏫 @fastdotai trained DL Eng 📝 https://t.co/By87iXx5PuJeremy Howard @jeremyphoward
222K Followers 5K Following 🇦🇺 Co-founder: @AnswerDotAI & @FastDotAI ; Hon Professor: @UQSchoolITEE ; Digital Fellow: @Stanfordmerve @mervenoyann
56K Followers 4K Following open-sourceress at @huggingface 🧙🏻♀️ proud mediterrenean 🍋 I do TL;DR on ML papersAndrada @andradavulpee
10K Followers 530 Following 🎨 Data Scientist 🦆 Kaggle Notebooks Grandmaster ⌨️ ZbyHP Data Science Ambassador 🐝 Weights&Biases Dev Expertカレーちゃん�.. @currypurin
13K Followers 886 Following 『面倒なことはChatGPTにやらせよう』が1/29に発売になりました。KaggleGrandMaster。 要望や質問などなんでも:https://t.co/bFBeCtAKiZ まで。 KaggleスタートブックとKaggleのチュートリアル第6版を執筆しました。yu4u @yu4u
8K Followers 1K Following General Manager at GO Inc. / Ph.D. in Eng. / Kaggle Competitions Grandmaster https://t.co/UEPcVAxE1B / https://t.co/iTjqtfhbAa…Tanishq Mathew Abraha.. @iScienceLuvr
54K Followers 1K Following PhD at 19 | Founder and CEO at @MedARC_AI | Research Director at @StabilityAI | @kaggle Notebooks GM | Biomed. engineer @ 14 | TEDx talk➡https://t.co/xPxwKTq6QbChris Albon @chrisalbon
86K Followers 2K Following Director of Machine Learning at the Wikimedia Foundation. We host Wikipedia.charm @charmq00
3K Followers 1K Following ソフトウェアエンジニア/東大物理修士(素粒子)/Kaggle Competitions Grandmaster 🥇2🥈3🥉2/ Highest Kaggle Rank 8thonodera @0verfit
8K Followers 650 Following @kaggle Grandmaster (🥇0🥈6 🥉1) at @NVIDIA @RAPIDSAI / @splatoonJP オールX 24Rob @Rob_Mulla
5K Followers 560 Following Data Science @ https://t.co/obcGxNNJSg // Python 🐍 & Data 💾 // 4x Kaggle Grandmaster // Live coding is fun 🎙️// Follow on twitch: https://t.co/GHjWoRVia7 & Youtube: https://t.co/WfD4vK0ageころんびあ @colum2131
2K Followers 1K Following Turing Inc.🚗 │ Kaggle Master │ AtCoder 水 | E2E Autonomous drivingsmly @smly
8K Followers 3K Following Fellow at Rist | Kaggle Grandmaster https://t.co/lBVa8oGfCW | Mahjong AI Competition https://t.co/8b1Ytq5G54JFPuget 🇺🇦 @JFPuget
15K Followers 1K Following Machine Learning at @Nvidia, 3x Kaggle Grandmaster CPMP. For AI/ML content read me at [email protected] Views are my own.Danil Zvyagintsev @danzvyagintsev
155 Followers 2K Following 💻 Top Rated Power BI Developer @Upwork | I write about Data, Design and Analytics | 19K+ on LinkedIn, follow me (link in bio)Omar Alkhasawneh @OmarAkhasawneh
0 Followers 38 Following computer engineering student , I interested in AI feildHastika Cheddy @Hastika06
6 Followers 74 Following Machine learning engineer | MLOps | Content creator#BTC.Eth# @LionofCrypto2
168 Followers 2K Following Crypto Promoter |DM for promotions,Marketing & Business. Be Honest to everyone,everywhere & Always.Always take stand for right things.Smart Cherrys Tech @smartcherrystc
9K Followers 5K Following Smart Cherrys Tech is Technology World.Justus Bruns @justusbruns
2 Followers 96 FollowingJack Reacher @JackReach516
73 Followers 1K FollowingChirag Taneja @ctxneja
9 Followers 94 Following ml web dsa || Passionate Learner Documenting Coding Journey, Mastering Tech, Creating Innovative Solutions ||Carl Grafe @CarlGrafe
936 Followers 1K Following Data analyst / consultant / problem solver @byuidaho | informatics PhD | epidemiology MS | sims | machine learning | math.FlashHacker777 @StarsMatrix777
869 Followers 2K Following I Killed @Twitter & @elonmusk Made @X I am a Video Game Streamer ⚔️🎮⚔️Shawn Charles🎤🔥 @ShawnBasquiat
32K Followers 3K Following 🧑🏾💻Ex-FAANG Software Engineer 🥑Senior ML Developer Advocate @ Coming Soon 🏗️Building Tech CommunitiesPhilipp Recherche @PPAIRECH
12 Followers 43 FollowingGathii James @gathijames1
157 Followers 3K Following Hi, I’m friendly and enjoy meeting new people. I also like wine, beef, and Netflix. Let’s be friends! 🍷🧀📺Alo @Hal90910
0 Followers 3K FollowingViacheslav Sinii @ummagumm_a
49 Followers 260 FollowingM.E.M @Mmomboesque
102 Followers 711 Following When I am not analysing Structures , I am Cruching data . Otherwise I am ranting about MUFC 💻🏗️⚽️ #DataAnalysis #StructuralEngineering #MUFCThanhTruong Tran @truongtran19315
49 Followers 200 FollowingHuaizhi Qu @qhz991029
11 Followers 128 Following Incoming PhD student, CS@UNC Chapel Hill; Undergraduate, CS@USTCたかいと @takaito0423
1K Followers 1K Following 21卒. 専攻:自然言語処理や統計学. Kaggle Competitions Master. https://t.co/vdHNaopJWN ᓚᘏᗢᗢᘏᓗ. エネルギー源 https://t.co/G1lXiMscZyKeiko @Keiko_geo
11K Followers 4K Following PhD; VP of Science and Technology @climateengine; GDE @googledevs; #EarthObservation 🌳🌴🛰 #AI; #EarthEngine tips for #BetterFuture🌏; from 🗾; views my ownIvo @ivoencarnacao
45 Followers 3K FollowingJoshua Raison @JoshuaRaison
39 Followers 2K Following Yet Another Dev 😌 | Tech Enthusiast 🙂 https://t.co/i3MQaKiwRORushikesh @r12k01_
7 Followers 23 FollowingMuhammad Z. Ahmed @MoZayed007
152 Followers 1K Following Aspiring ML Engineer / Applied Scientist / Research Scientist 🤖 TFT, MMORPGs🎮Ghelissi Anis أني�.. @ghelissi_anis
116 Followers 2K Following PSM™ PSPO™ Microsoft AZ-900| Business Analysis | Infosec | Dev & AutomationSalman Alam @Alam92Salman
2 Followers 72 FollowingAshutosh Mehra @ashutoshmehra
1K Followers 5K Following Senior Principal Scientist at Adobe. Working on Acrobat AI Assistant, LLMs, and document ML.alamgirqazi @alamgirqazi
134 Followers 811 Following AI Research at @uniofgalway . previously senior engineer @digitalocean.Ibrahim Habib @IbrahimHabibEg
1 Followers 111 Followingpatrickhrmn @patrickhrmn
87 Followers 453 Following Supporting technical entrepreneurs solving hard problems starting from day 0. Leading all global tech deals for Picus Capital.Sukrit Singh 𝕏 @DatabySingh
10 Followers 83 Following I predict trends. Like, I knew you’d read this.🤖 #DataScience 🤖 #MachineLearning 📍 #ArtificialIntelligence 👾 #BusinessAnalytics 🤌🏻Alen Capalik @capalik
179 Followers 955 Following CTO of https://t.co/xOlqgAkPpf, Founder of CounterTack (now https://t.co/55foNoBYv3) & https://t.co/snWLnZolVI. Entrepreneur, Hacker, Computer Programmer, AI/ML, HPCsukeponta @sukepon_ta
382 Followers 389 FollowingJan Rutkowski @janoz_94
106 Followers 286 Following I'm iOS developer who believes that the key of success of mobile apps are proper design and A11y. I also co-organize @Programistok and @MobileBialystok.Ervin Lang @ervinlang
50 Followers 1K Followingyousef dessouky @YousefDessouky
0 Followers 10 FollowingSebastian Raschka @rasbt
267K Followers 906 Following Machine learning & AI researcher writing at https://t.co/A0tXWzG1p5. LLM research engineer @LightningAI. Previously stats professor at UW-Madison.François Chollet @fchollet
470K Followers 769 Following Deep learning @google. Creator of Keras. Author of 'Deep Learning with Python'. Opinions are my own.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.AK @_akhaliq
310K Followers 3K Following AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo follow on Hugging Face: https://t.co/q2Qoey80GxSanyam Bhutani @bhutanisanyam1
35K Followers 994 Following 👨💻 Sr Data Scientist @h2oai | Previously: @weights_biases 🎙 Podcast Host @ctdsshow 👨🎓 International Fellow @fastdotai 🎲 Grandmaster @KaggleRadek Osmulski 🇺�.. @radekosmulski
25K Followers 555 Following Resources to take your Machine Learning skills to the next level 🧪 Senior Data Scientist, RecSys @NVIDIAAI 🏫 @fastdotai trained DL Eng 📝 https://t.co/By87iXx5PuJeremy Howard @jeremyphoward
222K Followers 5K Following 🇦🇺 Co-founder: @AnswerDotAI & @FastDotAI ; Hon Professor: @UQSchoolITEE ; Digital Fellow: @StanfordPyTorch @PyTorch
380K Followers 77 Following Tensors and neural networks in Python with strong hardware acceleration. PyTorch is an open source project at the Linux Foundation. #PyTorchFoundationAndrada @andradavulpee
10K Followers 530 Following 🎨 Data Scientist 🦆 Kaggle Notebooks Grandmaster ⌨️ ZbyHP Data Science Ambassador 🐝 Weights&Biases Dev ExpertAI at Meta @AIatMeta
533K 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.yu4u @yu4u
8K Followers 1K Following General Manager at GO Inc. / Ph.D. in Eng. / Kaggle Competitions Grandmaster https://t.co/UEPcVAxE1B / https://t.co/iTjqtfhbAa…Soumith Chintala @soumithchintala
187K Followers 883 Following Cofounded and lead @PyTorch at Meta. Also dabble in robotics at NYU. AI is delicious when it is accessible and open-source.Tanishq Mathew Abraha.. @iScienceLuvr
54K Followers 1K Following PhD at 19 | Founder and CEO at @MedARC_AI | Research Director at @StabilityAI | @kaggle Notebooks GM | Biomed. engineer @ 14 | TEDx talk➡https://t.co/xPxwKTq6Qbcharm @charmq00
3K Followers 1K Following ソフトウェアエンジニア/東大物理修士(素粒子)/Kaggle Competitions Grandmaster 🥇2🥈3🥉2/ Highest Kaggle Rank 8thonodera @0verfit
8K Followers 650 Following @kaggle Grandmaster (🥇0🥈6 🥉1) at @NVIDIA @RAPIDSAI / @splatoonJP オールX 24Rob @Rob_Mulla
5K Followers 560 Following Data Science @ https://t.co/obcGxNNJSg // Python 🐍 & Data 💾 // 4x Kaggle Grandmaster // Live coding is fun 🎙️// Follow on twitch: https://t.co/GHjWoRVia7 & Youtube: https://t.co/WfD4vK0ageLucas Beyer (bl16) @giffmana
56K Followers 446 Following Researcher (Google DeepMind/Brain in Zürich, ex-RWTH Aachen), Gamer, Hacker, Belgian. Mostly gave up trying mastodon as [email protected]Hugging Face @huggingface
345K Followers 189 Following The AI community building the future. https://t.co/VkRPD0VKaZ #BlackLivesMatter #stopasianhateDaniel Han @danielhanchen
7K Followers 941 Following Building @UnslothAI. Finetune LLMs 30x faster https://t.co/aRyAAgKOR7. Prev ML at NVIDIA. Hyperlearn used by NASA. I like maths, making code go fastMarques Brownlee @MKBHD
6.2M Followers 472 Following Web Video Producer | ⋈ | Pro Ultimate Frisbee Player | Host of @WVFRM @TheStudioAhmad Al-Dahle @Ahmad_Al_Dahle
4K Followers 53 Following #Girldad of twins. Leading GenAI @ Meta (llama, imagine, meta ai and more)Dan Siroker @dsiroker
33K Followers 3K Following Co-Founder & CEO of @LimitlessAI: a personalized AI powered by what you’ve seen, said, or heard. Co-Founder of @Optimizely. Three kids under 6. Amazing wife.VAR Österreich @VAR_Oesterreich
1K Followers 14 Following Offizielle Twitter-Seite des Video Assistant Referee Österreich. Mehr Infos unter: https://t.co/jy7BmDkfrdNorgard @BrianNorgard
88K Followers 0 Following Airchat co-founder // Ex-CPO Tinder, Norgard Capital, SpaceX/Lyft/Notion/AngelList/Airtable/eToro/StockTwits/Deel/StatMuse/DiDi/River/Alpaca/LiquidDeath/Sword/Private LLM @private_llm
1K Followers 0 Following A subscription-free LLM app that runs on-device on iPhone, iPad and Mac. Slaying fleeceware AI subscription apps with local LLMs and no-code Shortcuts.Grant Sanderson @3blue1brown
365K Followers 362 Following Pi creature caretaker. Contact/faq: https://t.co/brZwdQfdifFußball-Bundesliga @OEFBL
27K Followers 139 Following Offizieller Twitter-Channel der Österreichischen Fußball-Bundesliga Impressum: https://t.co/y30jZ2BhYIAaron Defazio @aaron_defazio
6K Followers 365 Following Research Scientist at Meta working on optimization. Fundamental AI Research (FAIR) teamNancy Pelosi Stock Tr.. @PelosiTracker_
561K Followers 224 Following Highlighting Politicians' trades so we can invest alongside Goal: get them banned from trading Powered by @joinautopilot_Xeophon @TheXeophon
1K Followers 846 FollowingCognition @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/tpfZwEwGiqOpenRouter @OpenRouterAI
5K Followers 82 Following A router for LLMs. 120+ models, explorable data, and open-source inference. https://t.co/ZR8gPNSd52Amanda Askell @AmandaAskell
26K Followers 653 Following Philosopher & ethicist teaching models to be good @AnthropicAI. Personal account. All opinions come from my training data.Tri Dao @tri_dao
19K Followers 365 Following Incoming Asst. Prof @PrincetonCS, Chief Scientist @togethercompute. Machine learning & systems.Groq Inc @GroqInc
46K Followers 470 Following Creator of the LPU™ Inference Engine, providing the fastest speed for AI applications, designed & engineered in N. America https://t.co/DsEqVAC5DpOmar Sanseviero @osanseviero
32K Followers 2K Following Chief Llama Officer @huggingface 🦙 Founder @AI_Learners. Xoogler (SWE @Google Assistant, 20% PM TF Graphics). 100% Hacker Llama🇵🇪🇲🇽Anjum Sayed @AnjumSayed
442 Followers 266 Following 🏁 Lead Data Scientist at McLaren Racing 🌱️ Building https://t.co/xaDvB2lqSv 🏅️ @Kaggle Competitions Grandmaster (datasaurus) 🎒️ Pacific Crest Trail Class of 2022peter! 🥷 @pwang_szn
25K Followers 523 Following Writing about bootstrapping https://t.co/YJFkxmhetA to 500K ARR in 2024 bench: 225x4 overhead press: 145x1 squat/DL: 0 (skip) 📍SeoulClémentine Fourrier .. @clefourrier
3K Followers 302 Following Leaderboards & evals research @HuggingFace 🐍✨ "The future is already here, it’s just not very evenly distributed" (Gibson)David @DavidSHolz
54K Followers 5K Following founder @midjourney, prev founder leap motion, nasa, max planckTeknium (e/λ) @Teknium1
29K Followers 3K Following Cofounder @NousResearch, prev @StabilityAI Github: https://t.co/LZwHTUFwPq HuggingFace: https://t.co/sN2FFU8PVE Support me on Github SponsorsArthur Mensch @arthurmensch
40K Followers 874 Following Co-founder and CEO @MistralAI. Apply https://t.co/yHGRZAtjcxVinod Khosla @vkhosla
632K Followers 575 Following entrepreneurship zealot, grounded technology possibilist, believer in the power of ideas, passionate about sustainability & impactIlya Sutskever @ilyasut
370K Followers 2 Following towards a plurality of humanity loving AGIs @openaiLaura Winter @lauracwinter
40K Followers 3K Following Sports broadcaster in F1, Extreme E, cycling, rugby & more - @eurosport | @itv | @primevideosport | @F1 | @btsport IG: lauracwinterRashmi Banthia @rashmigb
452 Followers 284 Following Mom, Data Science/Machine Learning/Deep Learning/NLP, Teaching Fellow @ Harvard, Kaggle Competition Master https://t.co/Vtc28HCaVESepp Hochreiter @HochreiterSepp
10K Followers 395 Following Pioneer of Deep Learning and known for vanishing gradient and the LSTM. I mostly tweet about random ArXiv papers which sparked my interest.Kai-Fu Lee @kaifulee
1.5M Followers 658 Following #AI Expert, CEO of @01ai_yi and Chairman of 创新工场 @sinovationvc, former President of Google China, Author of AI 2041 and NYT Bestseller AI SuperpowersJerry Liu @jerryjliu0
45K Followers 1K Following co-founder/CEO @llama_index Careers: https://t.co/EUnMNmbCtx Enterprise: https://t.co/Ht5jwxSrQBclem 🤗 @ClementDelangue
91K Followers 5K Following Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI buildersSimon Jegou @simon_jegou
298 Followers 136 Following Senior LLM Technologist @NVIDIA Views and opinions are my ownMistral AI @MistralAI
91K Followers 0 Following Fast, open-source and secure language models. Join us https://t.co/INALdNGvCPKirsten Lum | CTO & L.. @machsci
10K Followers 622 Following 🌲🌲 Applied ML/AI, data science, MLOps | Wife of 1, mom of 2 | Co-Founder and CTO of https://t.co/67wk0TNMlO | Quote: Oliver Wendell Holmes 🕊️Martin Shkreli (e/acc.. @wagieeacc
99K Followers 8K Following despite all my ragie I'm still just a wagie in a cagie working on DL Software: https://t.co/FVn3NRNrLe https://t.co/CgaoMfhUHdSusan Zhang @suchenzang
20K Followers 505 Following @ Google Deepmind. Past: @MetaAI, @OpenAI, @unitygames, @losalamosnatlab, @Princeton etc. Always hungry for compute.Alex Graveley @alexgraveley
31K Followers 933 Following I’m Alex Graveley, creator of GitHub Copilot, AI Tinkerers, Dropbox Paper, MobileCoin, and Hackpad. Building @ai_minion Hiring https://t.co/nsHar8OLPCRaja Biswas @raja_biswas
247 Followers 547 Following AI enthusiast, Kaggle Grandmaster! https://t.co/49ktUdzXvXIgor Babuschkin @ibab
44K Followers 685 Following Maybe the real AGI was the friends we made along the way. @xAIGlavin Wiechert👨�.. @GlavinW
1K Followers 4K Following Staff Software Dev. Learning 🧠 AI/Large Language Models, electrical engineering to build 🤖s (@BehindEng), 3D w/ #r3f. Built Atom Beautify to 9mil installsAleksey Korshuk @alekseykorshuk
165 Followers 252 Following A passionate developer from Belarus | @Coframe_ai | Palo Alto, CaliforniaxAI @xai
996K Followers 36 FollowingIf your model is weak, your paper might end up getting more citations because people are always happy to include your model as a baseline.
@jessechenglyu Translation: “we wanted to sell this crap before we get blown out of the water by Apple integrating AI features into iOS.”
Can people stop trying to make voice the singular interaction layer with technology it unironically sucks
This is the way.
@GregKamradt I don't see LLMs as systems that can to create new knowledge, I see them as a tool I can use to help ME create new knowledge
I don’t know a single productive person that uses Notion
We ranked 4th out of > 2000 teams in PII Data Detection competition hosted by Kaggle. Thank you to my rock star team 🤩 Solution with - PyTorch, HuggingFace Transformers, LLama3 70B Instruct, vLLM, Neptune.ai, Optuna and Spacy kaggle.com/competitions/p…
@ph_singer The waiting game begins
LMsys added phi-3-128K into the arena. Got it in my comparisons. Excited to see where it’ll be placed
@ph_singer @OpenRouterAI @MistralAI Thanks for the flag! Fix deployed:
@ph_singer @srush_nlp the paper is from meta, so they should be able to access that information
@ph_singer @hardmaru @karpathy @artificialguybr Omg sorry Philipp I forgot about you guys! :O
@ph_singer @Thom_Wolf Wow, congratulations on the release! So excited to see the improvements you've made to H2O-Danube2-1.8b! Keep up the amazing work! 💪❤️
i tested out danube 2 1.8B chat on different types of reasoning last night. it does surprisingly decent in various forms of reasoning except transitive where it sucks consistently. categorical, conditional and syllogistic reasoning are good, I'll dig more and share results.
Excited to release H2O-Danube2-1.8b - a new and improved language model being the best in class below the 2B parameter range. The model is trained on additional 2T tokens (total of 3T tokens) with various data mix stages as a result of extensive experimentation. Focus is on…
@victormustar Search by model size
@ph_singer Congrats! And thanks for sharing. Fits-on-single-GPU models are always nice for finetuning.
@ph_singer You can improve gsm8k by tuning on 3tr tokens is around optimal for this size, then it comes down to data improvement we found
NVIDIA releases OpenMathInstruct-1 - Opensources a high-quality 1.8M math instruction-tuning dataset - OpenMath-CodeLlama-70B achieves 84.6% on GSM8K and 50.7% on MATH, which is competitive with the best gpt-distilled models arxiv.org/abs/2402.10176
🚀 Announcing @h2oai 's h2o-danube2-1.8b With 1.8 billion parameters & trained on an extra 2T tokens. 🎯 Improvements across major benchmarks like ARC, HellaSwag, MMLU, and TruthfulQA. 🔥 New chat model also released 🌍 All available under Apache 2.0 - free for commercial use.…
Super happy to announce the release of h2o-danube2-1.8b, a 1.8 billion parameter foundation LLM. We adapted our original model and continued training with an additional 2T tokens which makes it the best model to date on Open LLM Leaderboard benchmark. huggingface.co/h2oai/h2o-danu…
Super happy to announce the release of h2o-danube2-1.8b, a 1.8 billion parameter foundation LLM. We adapted our original model and continued training with an additional 2T tokens which makes it the best model to date on Open LLM Leaderboard benchmark. huggingface.co/h2oai/h2o-danu…
I am a bit confused. Isn't Qwen 1.5 72B still the best open weight model? It has a higher MT Bench score, high chatbot arena rank, higher mmlu score. Also it's easier to deploy. Both DBRX and Qwen have restrictive licenses. (Qwen more restrictive tho)