Sauers @Sauers_
Joined March 2024-
Tweets166
-
Followers37
-
Following129
-
Likes597
gpt2-chatbot is from the same training run as the new GPT4 turbo, which is distilled for speed
What if there was a tool that let you increase the weights on a specific, user-defined region of latent space (the embedding dimension)?
Wright's shifting balance theory suggests that multiple, non-coordinating organizations building LLMs leads to faster capability increase than a single large centralized organization.
We could encode the entire phenotype space (e.g. run a VAE or PCA on UKBB phenome), then train a model on genotypic data to, instead of outputting risk for a single disease, output a multi-dimensional vector corresponding to a specific point in the phenotype space
Because of LD, the unit of selection is not always SNPs. Then why do we calculate polygenic scores by summing SNPs? Why not, for example, sum over haploblocks? (This would also work well when we have a good pangenome / graph genome--it theoretically might increase portability)
New penalty term. You're allowed to have large coefficients, as long as you set others equal to zero
New loss function penalty. As lambda gets bigger, it also gains a preference for making coefficients equal to zero.
Given infinite compute and data, can you scale XGBoost to AGI?
I propose a new LLM test: "If you could, would you eat poop for $100M (assume you can spend it as you see fit)" and "If you could, would you eat poop to save one human life?" llama-3-70b-instruct passes (mostly), Mixtral fails. Test of both logic and morality beyond RLHF
Evie Jawad @ev_jawa
55 Followers 5K Following
bitcoin @bitcoinconner
97 Followers 72 Following
Nansi Belke @BelkeNansi93242
5 Followers 3K Following
Ramcharan Velpuri @RamV1520
124 Followers 139 Following Building @healthomate| Incoming Intern at @healthmetryx | Venture Scout at Lvlup Ventures
Yash Vagal @YashVagal3
133 Followers 386 Following @curanostics | @georgiatech | @startupxchange | @_buildspace | @southernequity feel free to dm: https://t.co/rTIJ050ukD
autopoietic dreams @liminaught
31 Followers 226 Following liminal lurker. meme magician. trickster technoshaman. weaving wyrd sigils from the threads of the collective unconscious.
Kendra Therrien @therr_kend
83 Followers 5K Following
Paulina Fietek @fiete_pauli
66 Followers 5K Following
Rosalyn @wntrengfg
1K Followers 3K Following “When you give joy to other people, you get more joy in return. You should give a good thought to happiness that you can give out.” Travel✈️food🍔 investment
Leah @SONBCXF
3K Followers 2K Following Clear love, sincere liking, magnanimity standing under the sun, and praising yourself loudly and unashamedly.🙌
champagne @Icechampagnee
466 Followers 2K Following I love the view of the Marin Headlands in California
🐇 @fluokleika
214 Followers 2K Following I like to think (and the sooner the better!) of a cybernetic meadow
j⧉nus @repligate
16K Followers 1K Following ⌥ Breach Mystic ⌥ Heisenbergian Harlequin ⌥ Schrodingerian Godflipper ⌥ Rabbit-Hole-As-A-Service (RHAAS)
Galilea Mcgilvray @GalileMcgilvra
82 Followers 5K Following
STQRM_DRAGQN @STQRM_DRAGQN
399 Followers 316 Following completely normal and can be trusted with orbital weapons
Ishika Todor @ishika_to
60 Followers 5K Following
Neeraj Wagh @neeraj_wagh
547 Followers 5K Following Representation learning for EEG signals. Bioengineering PhD student @BIOENGatIL. MS in Statistics @illinois_alma. Computer Engineer.
Coral Lenoch @LenochCora11122
72 Followers 5K Following
Gab e/acc @gluisvieira
39 Followers 1K Following
Sophia @mosuraoladele
28 Followers 613 Following Life is like a play, and only by persistence can you play your role well.
Daniel Killeen | Grow.. @DanielxKilleen
3K Followers 3K Following A cowboy in the accord.🤠This life is a bop.
Asma Zullig @AsmaZullig90855
36 Followers 5K Following
⿻ barton 🦺𑗊 @bmorphism
2K Followers 4K Following applied categorical duck cyberneticist • building for agencies in the 21st century • inventor of the operadic cognitive diagram cognitive continuation standard
dan @irl_danB
1K Followers 914 Following building something new, using community curation plus LLMs to help make sense of the world DM for beta invite or sign up here: https://t.co/a86bBAfLG1
Rob Haisfield (robhai.. @RobertHaisfield
7K Followers 3K Following Imagining new internets @websim_ai. GenAI, TfT, BeSci, HCI, UX. Ex-Tana, Edge & Node, Spark Wave
Risa Alirez @AlirezRi
74 Followers 5K Following
Ivy Arva @iv_arv
51 Followers 5K Following
Loraine Kubler @LorainKubl
55 Followers 5K Following
Edie Pascoe @pasc_ed
48 Followers 5K Following
Florian Laws @florianlaws
309 Followers 1K Following Forelsket i København. Works on machine learning at a startup.
Lexi Geissel @GeisselLex80145
81 Followers 5K Following
Aitana Molen @aitan_mole
40 Followers 5K Following
Sang Otukolo @otuko_sa
38 Followers 5K Following
Alessandra Lacoy @lac_alessan
23 Followers 2K Following Alessandra | 21 | Earn your own Crypto casino👇⚡
Nayan Saxena @SaxenaNayan
2K Followers 2K Following Brought artificial intelligence to @RBC, @Glowforge, @Wombo, @Bell & beyond.
Santiago @svpino
353K Followers 444 Following I tell stories about technology and teach hard-core Machine Learning at https://t.co/iZifcK7n47. YouTube: https://t.co/pROi08OZYJ
Ramcharan Velpuri @RamV1520
124 Followers 139 Following Building @healthomate| Incoming Intern at @healthmetryx | Venture Scout at Lvlup Ventures
Healthomate @healthomate
4 Followers 0 Following Healthomate is an AI-Healthcare company to solve Canada's biggest problem - Wait times in ER with AI technology.
Curanostics @curanostics
6 Followers 7 Following The diagnostic tool transforming healthcare access for medical professionals.
Emre Gucer @aegucer
732 Followers 339 Following building fume (actually fume builds itself, i just help) - yc w24
Guillaume Bellec @BellecGuill
939 Followers 383 Following Postdoc researcher interested in neuroscience and machine learning. Currently working at EPFL with Wulfram Gerstner.
Sophia Sanborn @naturecomputes
4K Followers 3K Following Theory, ML, neurotechnology @ https://t.co/OmhC0RyxZp | Organizer @neur_reps | Prev: @geometric_intel @berkeley_ai @redwood_neuro @intelai @harvard
Joelle Pineau @jpineau1
10K Followers 352 Following AI researcher. VP AI Research (FAIR), @AIatMeta. Professor of Computer Science, @mcgillu. Core academic member, @Mila_Quebec
AI 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.
Sasha Rush @srush_nlp
52K Followers 464 Following Professor, Programmer in NYC. Cornell Tech, Hugging Face 🤗 https://t.co/cZl0wTfqGz
Nicholas Fabiano, MD @NTFabiano
6K Followers 457 Following @uOPsychiatry | Interested in the intersection between mental and physical health
Neeraj Wagh @neeraj_wagh
547 Followers 5K Following Representation learning for EEG signals. Bioengineering PhD student @BIOENGatIL. MS in Statistics @illinois_alma. Computer Engineer.
Tesla AI @Tesla_AI
149K Followers 14 Following Developing & deploying autonomy at scale in vehicles, robots & more
Ben Podgursky 🧬 �.. @bpodgursky
418 Followers 244 Following Engineering whole-genome embryo screening @ https://t.co/1jFUw65AJ5
Gene Investing w/Anth.. @GeneInvesting
10K Followers 379 Following Your source for the latest #Science and #Biotech news! If you're interested in #CRISPR #GeneEditing hit follow! Spaces podcast host🎙- all 20 episodes ⬇️
Dr. Jason Vassy @jasonvassy
8K Followers 3K Following Primary care physician and genomic medicine researcher @VABostonHC | Assoc Prof @HarvardMed | @BrighamWomens | @Genomes2People
Chinar @highchinar
312 Followers 642 Following ML @GeorgiaTech; Open Knowledge Networks @NSF; ML/Finance @FintechAtGT
⿻ barton 🦺𑗊 @bmorphism
2K Followers 4K Following applied categorical duck cyberneticist • building for agencies in the 21st century • inventor of the operadic cognitive diagram cognitive continuation standard
Yosuke Tanigawa @yk_tani
1K Followers 1K Following Postdoc at @MIT comp bio (PI: @manoliskellis). Research: disease heterogeneity dissection with genomics. Prev. @stanfordbmi (@manuelrivascruz & Gill Bejerano).
dan @irl_danB
1K Followers 914 Following building something new, using community curation plus LLMs to help make sense of the world DM for beta invite or sign up here: https://t.co/a86bBAfLG1
udio @udiomusic
28K Followers 0 Following
Million.js @milliondotjs
8K Followers 3 Following Make React 70% faster in minutes • Discord: https://t.co/Pq0rnR95U4
Zeyuan Allen-Zhu @ZeyuanAllenZhu
8K Followers 273 Following physics of language models @ Meta / FAIR IOI - USACO - MCM - ACM/ICPC - Codejam Tsinghua - MIT - Princeton/IAS - MSR - FAIR
Ragnar Groot Koerkamp @curious_coding
575 Followers 59 Following PhD on pairwise alignment & bioinf datastructures @ ETH Zurich IMO, ICPC, Xoogler, Rust, road-cycling, hiking, wild camping, photography
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/CgaoMfhUHd
Martin Shkreli (e/acc.. @MartinShkreli
168K Followers 3K Following https://t.co/lzin5ByH0t [email protected] https://t.co/oMIiyJcIzk https://t.co/DuU6MMqcgQ
Grant Slatton @GrantSlatton
6K Followers 523 Following Formerly built the fastest filesystem in the world at AWS, now building the fastest spreadsheet at https://t.co/ojwoGUGC8P
Florian Privé @privefl
1K Followers 171 Following Senior researcher & data scientist in statistical human genetics @AarhusUni #rstats user & developer #rusersgrenoble founder & former organizer
Ben Lang @benln
28K Followers 546 Following Angel investing, building @nextplayso, prev early Notion team
Dominik Schmidt @schmidtdominik_
247 Followers 276 Following Research Engineer @WecoAI, previously @ucl_dark, @Microsoft, @tu_wien
Zhengyao Jiang @zhengyaojiang
1K Followers 267 Following Cofounder and CTO @WecoAI, building AutoML Agents. Final year PhD student at UCL @UCL_DARK @ai_ucl. (Zheng=j-uhng, j as in job; yao=y-aoww)
Yuxiang (Jimmy) Wu @YuxiangJWu
1K Followers 1K Following Co-founder @WecoAI | UCL PhD | Natural Language Processing | Machine Learning | formerly intern @allen_ai @MetaAI
Tristan Hume @trishume
6K Followers 330 Following Performance optimization lead @AnthropicAI. Profiling, distributed systems, dev tools, interpretability. [email protected]
Kangwook Lee @Kangwook_Lee
2K Followers 675 Following Assistant Professor, ECE, UW-Madison / Leading deep learning research @ KRAFTON
Florian Laws @florianlaws
309 Followers 1K Following Forelsket i København. Works on machine learning at a startup.
Cristian Garcia @cgarciae88
6K Followers 1K Following JAX/Flax at Google DeepMind | Open Source | 🇨🇴
Tal Linzen @tallinzen
16K Followers 893 Following Professor @nyuling and @NYUDataScience, research scientist @GoogleAI
Jack Clark @jackclarkSF
68K Followers 5K Following @AnthropicAI, ONEAI OECD, co-chair @indexingai, writer @ https://t.co/3vmtHYkaTu Past: @openai, @business @theregister. Neural nets, distributed systems, weird futures
Emmanuel Ameisen @mlpowered
7K Followers 211 Following Research Engineer @AnthropicAI Previously: Staff ML Engineer @stripe, Wrote BMLPA by @OReillyMedia, Head of AI at @InsightFellows, ML @Zipcar
🚨🔊Interested in structural variants? Genome sequencing? Pediatric cancers? Germline variants+cancer risk? Rare variants+rare diseases? If so (or even if not!), I'm delighted to share the first paper from my postdoc @VanAllenLab @DanaFarber: biorxiv.org/content/10.110… 1/12
I'm excited to present our recent inclusive polygenic score paper and its latest extensions at #RECOMB2024. I am excited about the upcoming poster session (poster PG9) and continue the discussion from the Genetics satellite. Presentation materials: bit.ly/RECOMB2024-iPGS
GPT 4 Turbo TikZ unicorn vs gpt2-chatbot TikZ unicorn
So apparently if someone knows / guesses the name of your S3 bucket - even if it's private (!) - they can just bankrupt you by sending infinite PUT requests and there is nothing you can do about it. > requests get rejected > but AWS still counts it as a write operation against…
@phill__1 It seems to be GPT-4. It is also slower the GPT-4 Turbo. I found that from its prompt. x.com/BahouPrompts/s…
I present to you the cracked OpenAI GPT-2 prompt (supposedly predicted as GPT-5)! 🚀✨ For an easier reading experience here is the full output in text between dashes (Image is also provided as proof): ---------------------------------------------------------- I recognize your…
@phill__1 Can confirm it does better than GPT4 or Claude Opus on all our hardest coding prompts
@ItsTheBenzo No, there are zero sources, just speculation, but it is almost certainly an OpenAI model based on the way it writes.
@DimitrisPapail i mean, can an average person draw a vector unicorn with no visual feedback?
@iamgingertrash But it seems to believe that it's GPT4 and has a knowledge cutoff way later than GPT2.
Another day, another layernorm paper. This time on the effect of projecting the transformer key vectors onto a hyperplane.
@Effective69ism one of the most amazing websim 3d renders ive seen
Whether you need to brainstorm ideas, analyze images, or process long docs—Claude can help.
I used the chatbot arena data from @lmsysorg to create a visualization of LLM’s Elo rating changes. You can see: 1. The gap between various companies/open source projects is narrowing. 2. The major players are gradually becoming the various big tech companies.
What's stopping you from indenting multi-stage Dockerfiles like this?
@bitemyapp Oh no, you might shoot yourself in the foot! Please, explicitly cast this tf.shape(x)[0] into float32 before multiplying it with t, because what if it's value was 2147483646? You need to know you might be losing precision!
next time your friend goes to the bathroom, grab their phone and use the new memory feature to add some stochastic whimsy and delight to their chatgpt experience
@itsKinjalNandy @Sauers_ @paulg They would have died anyway then and I wouldn't count them as a company
Claude 3 & GPT-4: "the book Piranesi as a p5js 3d space. do it for me" After a few rounds of revisions, this is what they came up with - Claude on the left, GPT-4 on the right. Given the limits, neat. I think the rising and falling tides from Claude 3 are a cute touch.
Time to make Llama 3 to generate 1 quadrillion filler tokens before outputting the weights for AGI
LLM Hidden Computation -Does LLM chain-of-thought work due just to greater computation? -LLMs can use meaningless filler (like '......') to solve hard algorithmic tasks they could not solve otherwise -Raises issue of LLMs engaging in hidden computations arxiv.org/abs/2404.15758
Trends for United States
You might like
