Brydon Eastman
@brhydon
🇨🇦 Mathematician (heavy on the ish) @thinkymachines Prev. MTS @OpenAI; PhD @WaterlooMath Certified wife guy, featured twice in Lego Magazine © ☕//🤔➡️💻
You might like
The universe is weird as shit and I'm profoundly grateful that it permits us to poke at it.
FVV mentality finally hitting SF. Chevron/WTN era still affecting the culture
"betting on geopolitics" "betting on hyperliquid" "betting on datacenter deals" bro have you tried betting on yourself
if you think about it, 3b1b videos are a kind of cocomelon
Vladdy Jr. just did "DAAAAAA YANKEES LOSE!" 🤣
no offence to weird al et al but the first time the robots were funny was suno so is parody music really that hard?
Claude Sonnet 4.5 behaves the most desirably across Petri evals, but is 2-10x more likely to express awareness it's being evaluated than competitive peers. This affects how much we can conclude about how "aligned" models are from these evals. Improving realism seems essential.
Last week we released Claude Sonnet 4.5. As part of our alignment testing, we used a new tool to run automated audits for behaviors like sycophancy and deception. Now we’re open-sourcing the tool to run those audits.
The only way nobody builds it is if everybody dies
wait is this why i wasn't raptured the other week?
If AI seems a little spooky, that’s because it’s rooted in the occult. Conrad Flynn explains. (0:00) Is the Elite Class Involved in Occult Activity? (8:51) What Really Is the Occult? (13:13) Is There a Link Between AI and Demons? (17:58) Are Drugs a Portal to the Demonic?…
Today, we're proud to announce a new partnership with @UWaterloo to explore the intersection of AI and the future of work and learning. Our partnership will fund a new Research Chair and hands-on learning labs, bringing together our expertise in AI with Waterloo’s innovative…
i have such a vivid memory of the first time i went to lego dot com on the family computer. that site was so magical. i remember playing games and reading a blog post from a designer on how they use "make a sphere out of lego" as an interview question Now it's just shitty amazon
Websites should be this again. Waiting for the first company to do authentically do it. Everything looks like a venture capitalist yoga retreat website atm
Very excited to share what I have been working on with a great team of people at @thinkymachines. Tinker is a whole new way to train and customize models all the way up to frontier scale. Most importantly, it allows everyone to use their own code, data, tools and environments,…
Introducing Tinker: a flexible API for fine-tuning language models. Write training loops in Python on your laptop; we'll run them on distributed GPUs. Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!…
Give me forward_backward(), sample(), optim_step() enough and I shall move the world - Archimedes - me
Introducing Tinker: a flexible API for fine-tuning language models. Write training loops in Python on your laptop; we'll run them on distributed GPUs. Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!…
GPUs are expensive and setting up the infrastructure to make GPUs work for you properly is complex, making experimentation on cutting-edge models challenging for researchers and ML practitioners. Providing high quality research tooling is one of the most effective ways to…
Introducing Tinker: a flexible API for fine-tuning language models. Write training loops in Python on your laptop; we'll run them on distributed GPUs. Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!…
Tinker provides an abstraction layer that is the right one for post-training R&D -- it's the infrastructure I've always wanted. I'm excited to see what people build with it. "Civilization advances by extending the number of important operations which we can perform without…
Introducing Tinker: a flexible API for fine-tuning language models. Write training loops in Python on your laptop; we'll run them on distributed GPUs. Private beta starts today. We can't wait to see what researchers and developers build with cutting-edge open models!…
This is such cool work and a sign of very exciting things to come! I'm stoked to see what these models can do when broken out of the chat box and allowed to advance science.
Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is…
maybe don’t
OPENAI IS PREPARING TO LAUNCH A SOCIAL APP FOR AI-GENERATED VIDEOS - Wired
There's been a lot of great work into PEFT in general and LoRA in particular in the past year, this blog post collects those results and expands upon them, helping to elucidate more about these training techniques. It's a really fun read!
LoRA makes fine-tuning more accessible, but it's unclear how it compares to full fine-tuning. We find that the performance often matches closely---more often than you might expect. In our latest Connectionism post, we share our experimental results and recommendations for LoRA.…
Sharing our second Connectionism research post on Modular Manifolds, a mathematical approach to refining training at each layer of the neural network
Efficient training of neural networks is difficult. Our second Connectionism post introduces Modular Manifolds, a theoretical step toward more stable and performant training by co-designing neural net optimizers with manifold constraints on weight matrices.…
United States Trends
- 1. Auburn 42.1K posts
- 2. Brewers 59.2K posts
- 3. Georgia 66.1K posts
- 4. Cubs 53.8K posts
- 5. Kirby 22.4K posts
- 6. Michigan 61.7K posts
- 7. Hugh Freeze 2,939 posts
- 8. #GoDawgs 5,316 posts
- 9. Kyle Tucker 3,008 posts
- 10. #ThisIsMyCrew 3,123 posts
- 11. Boots 49.7K posts
- 12. Amy Poehler 2,892 posts
- 13. Jackson Arnold 2,094 posts
- 14. Gilligan 5,191 posts
- 15. Tina Fey 1,976 posts
- 16. Nuss 5,627 posts
- 17. #MalimCendari3D 2,923 posts
- 18. Sherrone Moore 2,013 posts
- 19. Arizona 40.7K posts
- 20. #MagicBrew 11.8K posts
Something went wrong.
Something went wrong.