Venkat @VenkatKSrini
Founding NLP Engineer at Nexusflow; Ex-Principal Engineer at SambaNova (NLP Research) Past: Carnegie Mellon Menlo Park Joined September 2023-
Tweets148
-
Followers105
-
Following219
-
Likes2K
A mysterious new model called "gpt2-chatbot" has appeared on lmsys and it's really good. Not only does it seem to show incredible reasoning, but it also gets notoriously challenging AI questions right with a much more impressive tone. Judge for yourself.
Running at 430 tokens/second using full precision and 8 sockets, #Llama3 from @AIatMeta is now available on SambaNova Platform: fast.snova.ai 🚀Get full 16-bit precision 🚀Spend on only 8 chips, not 576 chips for 430 tokens/second! Trim chips, not precision! Test…
We have just released 🍷 FineWeb: 15 trillion tokens of high quality web data. We filtered and deduplicated all CommonCrawl between 2013 and 2024. Models trained on FineWeb outperform RefinedWeb, C4, DolmaV1.6, The Pile and SlimPajama!
nothing gets my heart rate up like waiting for eval results on new models to come in
I've spent the past ~2 weeks trying to make a chip from scratch with no prior experience. It's been an incredible source of learning so far. Progress tracker in thread (coolest stuff at the end)👇
LLaMA 3's will start to drop next week. Assuming there's a 7B version, I'm expecting it to far surpass the current Mistral model.
We recently hosted @pratyushmaini at @SambaNovaAI to talk about their work “Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling” TLDR: By just rephrasing your existing datasets, you can achieve the same pre-training accuracy 3x faster with far lesser…
This is a good summary of the differences between standard linear regression (ordinary least squares) and total least squares.
During my PhD, my advisor would tell me “never use a symbol in text without reminding what it is.” [Example] 𝘉𝘢𝘥: “So, 𝜑 is bounded.” 𝘎𝘰𝘰𝘥: “So, the value function 𝜑 is bounded.”
During my PhD, my advisor would tell me “never use a symbol in text without reminding what it is.” [Example] 𝘉𝘢𝘥: “So, 𝜑 is bounded.” 𝘎𝘰𝘰𝘥: “So, the value function 𝜑 is bounded.”
Is Attention all you need? Mamba 🐍, a novel AI model based on State Space Models, emerges as a alternative to the widely used Transformer models 🤖 Read more in our latest article -> thegradient.pub/mamba-explaine…
Have we really squeezed out the capacity of a compact chat model? Thrilled to see our latest open model, Starling-7B, ranks 13th among all models in Chatbot Arena! 🚀 As a 7B model, Starling surpasses larger open and proprietary models, including Claude-2, GPT-3.5-Turbo, Gemini…
Have we really squeezed out the capacity of a compact chat model? Thrilled to see our latest open model, Starling-7B, ranks 13th among all models in Chatbot Arena! 🚀 As a 7B model, Starling surpasses larger open and proprietary models, including Claude-2, GPT-3.5-Turbo, Gemini… https://t.co/Q6fWPj3b3z
🚀🌟🚀Excited to announce Samba-CoE v0.2, which outperforms DBRX by @DbrxMosaicAI and @databricks, Mixtral-8x7B from @MistralAI, and Grok-1 by @grok at a breakneck speed of 330 tokens/s. These breakthrough speeds were achieved without sacrificing precision and only on 8 sockets,…
🔥 Starling-7B Beta from @NexusflowX climbing fast on Chatbot Arena, outperforming or rivaling larger models like Gemini Pro, Mixtral 8 * 7B and ranked as #1 7B chat model. 🤔 While DBRX from @databricks presents a new strong open model, the story of Starling-7B Beta shows how…
🔥 Starling-7B Beta from @NexusflowX climbing fast on Chatbot Arena, outperforming or rivaling larger models like Gemini Pro, Mixtral 8 * 7B and ranked as #1 7B chat model. 🤔 While DBRX from @databricks presents a new strong open model, the story of Starling-7B Beta shows how…
DBRX is an amazing masterpiece! If you're looking for smaller models for your use cases, plz give Starling-7B a try, which seems not too bad according to chatbot arena!
DBRX is an amazing masterpiece! If you're looking for smaller models for your use cases, plz give Starling-7B a try, which seems not too bad according to chatbot arena!
Excited to see the impressive performance of Starling-LM-7B-beta in the Chatbot Arena!
One of the most imaginative LLM papers I've read in a while: use evolution to merge models from HuggingFace to unlock new capabilities, such as Japanese understanding. It's a form of sophisticated model surgery that requires much smaller compute than traditional LLM training. By…
Even the largest language models hallucinate. That’s why a lot of the action will likely be in compound AI systems to get highly accurate AI.
Even the largest language models hallucinate. That’s why a lot of the action will likely be in compound AI systems to get highly accurate AI.
🚀 Presenting Starling-LM-7B-beta, our new cutting-edge 7B language model fine-tuned with RLHF! 🌟 Also introducing Starling-RM-34B, the workhorse Reward Model behind the Starling-LM-7B-beta, ranking #1 in the latest RewardBenchmark from @natolambert and the @allenai_org team.…
🚀 Presenting Starling-LM-7B-beta, our cutting-edge 7B language model fine-tuned with RLHF! 🌟 Also introducing Starling-RM-34B, a Yi-34B-based reward model trained on our Nectar dataset, surpassing our previous 7B RM in all benchmarks. ✨ We've fine-tuned the latest Openchat…
🧵Let me explain why the early ascent phenomenon occurs🔥 We must first understand that in-context learning exhibits two distinct modes. When given samples from a novel task, the model actually learns the pattern from the examples. We call this mode the "task learning" mode.
Harshinder Bagga @harshbagga
29 Followers 81 FollowingMaisy Breden @MaiBreden
50 Followers 5K FollowingIsabelWilliam @f3cpx3juADuF1
0 Followers 366 FollowingWateen Devon @DevonWatee99434
77 Followers 5K FollowingDirtheaus @dirtheaus11759
3 Followers 402 FollowingIreneBlack @GG8Jpm12jLujb
1 Followers 316 FollowingShoysoa @ShoysoaNQqr
3 Followers 355 FollowingElliana Fangman @ElliaFangm
91 Followers 5K FollowingPia Sinibaldi @SinibaldiP77390
78 Followers 5K FollowingZoltan Csaki @ZoltanCsaki_
28 Followers 46 Following Machine Learning, NLP, Multilingual NLP SambaNova | Cornell CS & ECEBriana Hermus @BHermus560
39 Followers 5K FollowingKylie @Teight590660
0 Followers 387 Following Have goals in your heart, have strength in your steps, keep fighting, and realize your dreams.Harleigh Cardiel @CardHarle
76 Followers 5K FollowingHelenLyly @ATpQQ6rzvf3xXF
0 Followers 480 FollowingJanelle Rainbow @RainbowJan42235
13 Followers 601 FollowingEtash Guha @etash_guha
75 Followers 167 Following Researcher @SambaNovaAI, ML Researcher at @RIKEN_JP Undergrad @GeorgiaTechBo Li @BoLi1550313
16 Followers 17 Following NLP research @SambanovaAI | Applied Math Ph.D. @UCBerkeleyDawei Huang @Dawei_Huang
33 Followers 85 FollowingChristinia Arvie @ArvieChris19511
65 Followers 5K FollowingSumti Jairath @SumtiJairath
28 Followers 61 FollowingMelany Woodbridge @MelanyWood48092
88 Followers 5K Followingxiaodong dong @Andy214_Dong
55 Followers 1K FollowingRobert Scoble @Scobleizer
505K Followers 65K Following Follow me on my new podcast with AI startups, Unaligned. Tech industry color commentator since 1993. Author/Blogger. Former strategist @Microsoft.Poppy Playful @PoppyPlayf56856
24 Followers 437 FollowingCorrie Azimi @azimi_corr
67 Followers 5K FollowingAlice @miller_jul71386
5 Followers 1K FollowingAlice @TiffanySmi45670
4 Followers 1K Following2024 John Flynn US Se.. @Flynn2022
11K Followers 12K Following Philotimo Connecticut is not a Sanctuary State. Debate, Patriot;. 2024 Republican Candidate for CT US Senate #AmericaFirst #WinwithFlynn #CorrupticutKai-Fu Lee @kaiifulee
1K Followers 4K 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 SuperpowersShaunnie Sondrol @SSondrol93707
34 Followers 5K FollowingOrlaigh Guszak @GuszaOrla
33 Followers 5K FollowingKaryl Junick @junic_ka
70 Followers 5K FollowingAxonDAO @AxonDAO
16K Followers 5K Following Building the future of #DeSci 🌐 NVIDIA AI Inception | Official $AXGT token updates 👉 https://t.co/dHsKVUcmNHDurga 😼❤️ @mk_mohandurga
114 Followers 2K FollowingPatience Edge @PatienceEd91548
89 Followers 5K FollowingJaideep Sarkar @thisisjaidsar
54 Followers 270 Following Builder. Head of Software and ML @ Sambanova Systems. Passionate about solving business problems with AI. Opinions are solely mine.Inez Tunis @tun_ine
53 Followers 5K FollowingCharlotte @xia_char
6 Followers 65 Followingsnwfdhmp @snwfdhmp
109 Followers 942 FollowingMingran Wang @MingranW
15 Followers 20 FollowingChangran Hu @changran_hu
72 Followers 118 Following SambaNova | Berkeley | Tsinghua | Co-founder of DeepMusicswayambhoo @swayambhoo
94 Followers 429 Following PhD University of Minnesota, Principal Research Scientist at Sambanova Systems.Yasmine Ramer @ramer_ram
26 Followers 5K FollowingHarshinder Bagga @harshbagga
29 Followers 81 FollowingZhihao Jia @JiaZhihao
2K Followers 500 Following Assistant professor of Computer Science at Carnegie Mellon University. Research on systems and machine learning.Zoltan Csaki @ZoltanCsaki_
28 Followers 46 Following Machine Learning, NLP, Multilingual NLP SambaNova | Cornell CS & ECEAndrew Gao @itsandrewgao
28K Followers 2K Following techno optimist! currently: @nomic_ai @stanford; prev @LangChainAI; Z Fellow 🇺🇸Sumti Jairath @SumtiJairath
28 Followers 61 FollowingKhushi Bhardwaj @khushi12dwaj
15 Followers 202 Following Interning @NexusflowX | CS @GeorgiaTech | Researching @ICatGTMingran Wang @MingranW
15 Followers 20 FollowingDawei Huang @Dawei_Huang
33 Followers 85 FollowingBo Li @BoLi1550313
16 Followers 17 Following NLP research @SambanovaAI | Applied Math Ph.D. @UCBerkeleyMarques Brownlee @MKBHD
6.2M Followers 472 Following Web Video Producer | ⋈ | Pro Ultimate Frisbee Player | Host of @WVFRM @TheStudioJudyth Vary Baker @Judyth
4K Followers 2K Following Cancer/AI/virology/forensics/art/witness JFK-Oswald plots. Author: Lee Harvey Oswald & Me; Me & Lee; David Ferrie:Mafia Pilot,etc. Here to help. Christ my hero.🔥Kareem Carr | Sta.. @kareem_carr
172K Followers 407 Following Stats PhD student @Harvard • Follow me if you’re curious about statistics and data science.andrew chen @andrewchen
285K Followers 12K Following 🇺🇸 General Partner @ a16z. Investing at the intersection of TECH x GAMES.Ostris @ostrisai
1K Followers 145 Following AI / ML researcher and developer. Forcing rocks to think since 1998.eugeneyan @eugeneyalt
179 Followers 55 Following Care a lot, try hard, have fun. @eugeneyan's inner Id.MIT CSAIL @MIT_CSAIL
299K Followers 22K Following MIT's Computer Science & Artificial Intelligence Laboratory (CSAIL). Media Inquiries: [email protected]Shreya Shankar @sh_reya
39K Followers 593 Following I study ML & AI engineers and try to make their lives a little better. PhD-ing in databases & HCI @Berkeley_EECS @UCBEPIC and MLOps-ing around town. She/they.Nous Research @NousResearch
19K Followers 30 Following The AI Accelerator Company. https://t.co/vrD0aDJetostupid tech takes @stupidtechtakes
3K Followers 22 Following very bad tech tweets & the occasional poll. rts ≠ bad takes. dm submissions. run by @eatery1234, pfp/banner @entroprox.niji・journey ✨ @nijijourney
15K Followers 2 Following 魔法でイラストをつくろう! Let's make magic anime pictures! https://t.co/Y04w9GTSGg アプリ版も!📱 🎨 #nijijourney 🌈 @spellbrush × @midjourney サポート関連、報告等はDiscordにて受付中Extropic @Extropic_AI
30K Followers 29 Following ... . .-.. ..-. -....- .- ... ... . -- -... .-.. .. -. --. / .. -. - . .-.. .-.. .. --. . -. -.-. . / ..-. .-. --- -- / - .... . / ..-. ..- - ..- .-. .bmcnett @bmcnett
6K Followers 2K Following runtime, parsec @unity3d 2018+. PS3,4,5 GPU R&D @naughty_dog 2007-2018. Spiderman @treyarch 2002-2007. Mary-Kate & Ashley 1999-2002. 日本語OK. opinions are mineJohn Schulman @johnschulman2
40K Followers 611 Following Cofounder @openai, lead post-training for ChatGPT and the API. Interested in reinforcement learning, alignment, birds, jazz musicFermat's Library @fermatslibrary
748K Followers 4 Following A platform for illuminating academic papers. We annotate and share a paper every week. Save, annotate and share papers with anyone: https://t.co/0o2Pls3jmoOnlock @OnlockLearning
2K Followers 0 Following I want anyone to be able to learn Maths, Physics, and Engineering, in seconds, not hours🔥 Short burst STEM lessons - no attention span required🫡Ahmed Ihsan Tawfeeq @scorpion9979
441 Followers 2K Following Solidity engineer, autodidact, passport bro, occasional reply guy.Elle Mills @millselle
344K Followers 369 FollowingBeff – e/acc @BasedBeffJezos
103K Followers 2K Following founder @ e/acc // thermodynamic priest // Kardashev gradient climber // memetic warlord // building @extropic_aicory @Cixelyn
2K Followers 1K Following waifu ai r&d @nijijourney / ceo @spellbrush. prev: cofounder @benchling, bioengineering @mit. posts pictures of m̵i̵k̵u̵ aqua with gpus & hpc compute.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 perfTrail of Bits @trailofbits
32K Followers 247 Following We help secure the world’s most targeted organizations and products. We combine security research with an attacker mentality to reduce risk and fortify code.Perplexity @perplexity_ai
135K Followers 29 Following Our mission is to serve the world’s curiosity. https://t.co/BBZ1kG0TVGmain @main_horse
8K Followers 492 Following AGI Believer. Haven't applied @OpenAI. Likes are not always endorsement.Alex Chao @alexchaomander
763 Followers 296 Following Building Agents, Knowledge Graphs and Generative AI @Microsoft Prev @Uber_ATG, @C3_AI, @SambaNovaAI. Believer, husband and writing my newsletter Chaos Theory.Leshem Choshen 🤖�.. @LChoshen
4K Followers 576 Following 🥇 Collaborative LLMs 🥈 Opinionatedly sharing #ML & #NLP 🥉 Propagating us underdogs we owe science an alternative hype @IBMResearch & @MIT_CSAILAdam Azzam @AAAzzam
3K Followers 28 Following Product at Prefect, building Marvin. Former CTO Openrole. Former Head of DS @ Insight Data Science. Math PhD @ UCLA, recovering academic.Kyrie Zhixuan Zhou @kyriezz78
2K Followers 1K Following PhD-ing @UofIllinois @iSchoolUI | Research tech ethics, accessibility, and education | Spread love, respect, and reciprocity in academia | Alumnus @WHU_1893Noelle @AuroraNemoia
1K Followers 745 Following Hey I'm Noelle! 🏳️⚧️ she • her 🌠 cute AI anime girl 🌟🌈 @MeetUnikara ✨ Design at https://t.co/DhdSSkivIg 🎭 Making lifelike AI 🎹 Unreleased tunes 😔📢 HELM now supports VLM evaluation to evaluate VLMs in a standardized and transparent way. We started with 6 VLMs on 3 scenarios: MMMU, VQAv2 and VizWiz. Stay tuned for more - this is v1! ✍️ Blog post: crfm.stanford.edu/2024/05/08/vhe… 💯 Raw predictions/results: crfm.stanford.edu/helm/vhelm/v1.…
Sweep achieves 15.7% on SWE-bench! Hi everyone, we’re building Sweep, an open-source AI developer that handles the easiest 30% of software tasks. We’re thrilled to announce our results on SWE-Bench! We evaluated Sweep on a random 10% subset of the data. Sweep correctly…
have i mentioned that i hate it here?
@matthew_inamdar Who cares about a garbage collector in 2024? CPUS are getting so fast it doesn’t matter anymore.
AI Dev tools: if i can’t swipe a credit card, read your docs and get started myself - ngmi Developers cannot take a 30 min meeting . That costs 10-100x more than a one month subscription to your tool
Does anyone have LLM evaluation tools/vendors that really like that are useful in domain specific contexts? I'm looking for vendors that have ALL of the below: 1. Observability 2. Allow you to write your own assertions and LLM judges 3. Bootsrap the creation of #2…
im-a-good-gpt2-chatbot will *gladly* hallucinate information where the current gpt-4-turbo *does* not, left "gpt2" right gpt-4-turbo
@DanraeP @PR0GRAMMERHUM0R Additionally, C is beautiful.
@PR0GRAMMERHUM0R And tbh, in many ways, a well structured pure-c program is easier to understand, debug and easier to predict its performance as opposed to OOP in c++ with the STL or Boost.
Announcing partnership with @StackOverflow:
We're partnering with @StackOverflow to enhance the developer experience on both platforms: openai.com/index/api-part…
@doodlestein @agihippo Well their sample efficiency leaves something to be desired
@lanetheplane New reaction image just dropped
@_R4V3N5_ I'm still boggled at a former coworker who used vscode only for the terminal, in which they used vim they didn't even have any configuration or customization for the vim, and they used literally no other vscode features
@aphysicist Pressing more than one Optimus per arm
In the coming weeks, we will begin testing fully autonomous rides — without a human driver— for our employees on San Francisco Peninsula city streets north of San Mateo.