Dogmas are so, until aren’t anymore. https://t.co/ZbG8QTpHcn
— Nathan Benaich (@nathanbenaich) 12 juin 2023
Dogmas are so, until aren’t anymore.
By
–
Dogmas are so, until aren’t anymore. https://t.co/ZbG8QTpHcn
— Nathan Benaich (@nathanbenaich) 12 juin 2023
Dogmas are so, until aren’t anymore.
By
–
London Tech Week is more like “my US VC and LP friends come visit London” week 🙂

By
–
Even though the UK has sanctions on Iran, researchers at a dozen UK universities "have helped the Iranian regime to develop sophisticated technology that can be used in its drone programme and fighter jets." What the beep. "Key pieces of research have been conducted by

By
–
"I would claim we are already manipulated by a machine. It consists of the corporate interests of a handful of companies that make vast profits on the back of our personal information." "Relying on Big Tech companies to tell us how this [AI] future should look is like turkeys
By
–
Along the lines of @maithra_raghu
’s “does one model rule them all? https://
maithraraghu.com/blog/2023/does
-one-model-rule-them-all/
…

By
–
The Gorilla project is a great example of how open source base model + fine tuning through self-instruction instruction + document retrieval can outperform a larger model for an economically-useful task.
By
–
Link Work by @vivnat and team at @GoogleHealth https://
nature.com/articles/s4155
1-023-01049-7
…

By
–
This is neat: “a representation-learning strategy for ML models applied to medical-imaging tasks that mitigates such ‘out of distribution’ performance problem and that improves model robustness and training efficiency.” “REMEDIS (for ‘Robust and Efficient Medical Imaging with

By
–
2 weeks til @raais in London, featuring @intercom @MetaAI @GoogleHealth @northvolt @UniofOxford @PrescientDesign @broadinstitute @sciam @Cruise Diving into AI-first SaaS, LLMs, AVs, clinical medicine, energy and batteries, protein generation, graph neural networks and

By
–
Is there a foundation model for embodied AI / robotics? Turns out that while today's visual 'foundation models' outperform learning from scratch baselines, there is no FM that is universally dominant across 17 different tasks spanning locomotion, navigation, dexterous, and