Sabitlenmiş Tweet
we've been building @GoodfireAI as a true 'age of research' company.
if you liked the VPD research on decoding model weights, there will be bangers all month, and we're only accelerating
English
Eric Ho
387 posts
Eric Ho
@eric_ho
Co-Founder / CEO @GoodfireAI - AI interpretability research company
San Francisco Katılım Eylül 2011
397 Takip Edilen2.9K Takipçiler
just welcomed chris earls, cornell engineering professor, to the team!
i'm particularly excited about his work to unlock scientific creativity in frontier AI with interpretability
we've hired several professors at @GoodfireAI because we're investing heavily in foundational research to discover the science of neural networks. if this is work that you're interested in, join us!
English
Eric Ho retweetledi
i think it’s important to be default loving rather than default fearful of AI. i feel the same way
Noah Smith 🐇🇺🇸🇺🇦🇹🇼@Noahpinion
I know it might kill everyone, but I can't help loving AI. All my life I wanted to talk to an alien intelligence, but I never thought I'd really get the chance to.
English
we can use interp to see what models truly believe versus what they say. often a surprising difference
Goodfire@GoodfireAI
Can LLMs predict the next World Cup champion? Goodfire partnered with @EternisAI to improve how LLM forecasters use available evidence and manage uncertainty. We found models were overconfident in their predictions – but probes significantly improved calibration. (1/6)
English
Eric Ho retweetledi
Never been more bullish on design and Figma than seeing Jason be negative. Looking forward to proving you wrong. Again.
@jason@Jason
I heard the same thing from magazine designers when pagemaker came out Then again when 99designs and logo contests emerged Not sure if this time is different 🤷
English
Eric Ho retweetledi
spc is one of the best places to build and where @GoodfireAI got our start. highly recommended
Aditya Agarwal@adityaag
Every founder I meet worries about missing this moment. The best worry about wasting it. There’s a difference. It’s a waste to ignore how much the world has changed. It’s a waste to think you can capture value with pure software the same as 5 years ago. It’s a waste to build something small. We have seen the shift slowly and then very quickly at @southpkcommons. Here’s who we want in the Founder Fellowship now: hardware tinkerers, mad scientists, obsessives, biohackers, people who build nuclear reactors in their basements. People who want to get their hands dirty and touch grass and atoms. If you are only building software, then please (for your own sake!) have a thesis that all your friends laugh at you about. Heresy is the price of ambition. Then put yourself in the right environment to maximize your ambition. (Apply by August 2nd)
English
Eric Ho retweetledi
I have an ultrarare, pathogenic variant which is likely to kill me eventually unless biotech advances before then. Standard pathogenicity tools (e.g., REVEL, PolyPhen-2, ClinPred) do not work on this type of variant. What does? @goodfire's EVEE. It uses Evo2, so it's capable of generalization to this variant type, enabling downstream analyses to get disease risk estimates on biobank data
English
Eric Ho retweetledi
Today, researchers made an important breakthrough in interpretability.
They found "manifolds" in the neural net weights for any concepts in image gen models (SDXL), like the pretzel manifold, and could steer them to generate various kinds of pretzels from the weights directly.
This is well beyond a neat theoretical understanding on how AI models see but gives us a low-compute volume dial to edit the results of a generation.
Goodfire@GoodfireAI
If models think in shapes, our tools should too. Our latest research: Block-Sparse Featurizers (BSFs), a new way to find concepts in model activations - using multidimensional “blocks” instead of single directions. (1/9)
English
congrats to @vincentweisser and team! @PrimeIntellect have been great partners to us
Prime Intellect@PrimeIntellect
Announcing our $130M Series A to build the Open Superintelligence Stack Led by Radical Ventures, with NVIDIA, Intel Capital, Dell Capital, and existing investors Train, deploy, and continuously improve your own models using our stack. Own your intelligence.
English
exciting stuff, cool to see a big bet on alignment!
Geoffrey Irving@geoffreyirving
We're excited to announce that Resolution has a $160M grant from Coefficient Giving: $108M unconditional, with a further $52M conditional on hiring and compute needs. We'll use it to grow teams across our research portfolio and invest heavily in research automation. 🧵
English
Eric Ho retweetledi
this is a really great example of how much clearer things look from the geometric perspective: we should really think of the classic curve detector family in inceptionv1 as points on a 'curve manifold', and the wiggles in this manifold allow readoff of different semantic information. I think this points towards why manifolds are such a useful representational strategy: by arranging its neurons to work together, it can actually represent many ideas with a single subspace
Goodfire@GoodfireAI
We also revisited an interpretability classic: curve detectors in InceptionV1. Neurons and SAE features turn out to be fragments of one continuous orientation feature, and the block *also* contains higher-order Fourier harmonics that hadn’t been described before! (7/9)
English
want to find shapes in the mind of your model?
silico (our AI neuroscientist) can autonomously train block sparse featurizers on your model or any open model you're training
DM me or reach out on our website for early access
Goodfire@GoodfireAI
If models think in shapes, our tools should too. Our latest research: Block-Sparse Featurizers (BSFs), a new way to find concepts in model activations - using multidimensional “blocks” instead of single directions. (1/9)
English
Eric Ho retweetledi
Grateful to be spending my summer @GoodfireAI ! 🌁 Interpretability is the next frontier, I think, deciding how much and in what capacity we will trust AI. I can’t wait to learn from the brilliant folks working here. DM me if you’re in SF, would love to chat☕️
English
@beffjezos @GoodfireAI we're working on something even better
libraries + skills + autonomously running interp experiments on any open model
x.com/GoodfireAI/sta…
Goodfire@GoodfireAI
Introducing Silico: the platform for building AI models with the precision of written software. Silico lets researchers and engineers see inside their models, debug failures, and intentionally design them from the ground up. Early access is open now. 🧵(1/10)
English
Someone should put out a library & API for mech interp of open source models and transparency-mog Dario
CG@cgtwts
Anthropic CEO: “open source is kind of a distraction you still can’t really see what’s happening inside the model, so it’s not truly “free.” People should be paying a lot more attention to what Anthropic is doing.
English
we're getting closer to being able to hand edit a weight and know exactly what it will do. kind of wild that this was done in one day using silico (our ai neuroscientist)
Goodfire@GoodfireAI
We removed an LM's ability to speak German by fine-tuning on only 4 German tokens. As part of a 1-day hackathon with our product Silico, we removed a 67M-parameter language model's ability to predict German text, by tuning only a scalar factor on one subcomponent of the weights. (1/6)
English
@eric_ho @GoodfireAI hire me as your technical head of growth and i will make your wildest dreams come true
English
we're hiring for a bunch of technical GTM roles at @GoodfireAI across forward deployed engineering, sales, and growth
come help us understand every model across biology, materials, robotics, language, and more
apply here or DM me: goodfire.ai/careers
English
@eric_ho @GoodfireAI wild range of fields you're hitting
English