This trailer by my publisher @OneSignalPub tells the story of how I founded @DAIRInstitute, which turns 5 this December!
My book DEEP UNLEARNING argues that we don't have to settle for the technological path charted by gazillionaires.
✴️Preorder at https://t.co/vH5svG6AXl
Cancer patients with the same diagnosis often have very different tumor biology.
Molecular testing can reveal underlying differences, but that's slow and expensive.
Introducing Holon-S1, our platform to predict genes and biomarkers from routine pathology slides.
• beats two other models at predicting spatial gene activity in a controlled comparison
• has identified potentially novel biomarkers in breast cancer and colon cancer
• evaluated across 16 organ types and more than 15 scanners, using slides from both formalin-fixed, paraffin-embedded and fresh-frozen tissue.
We believe Holon-S1 has the potential to help researchers find new biomarkers in slides they already have, discover treatment targets, and bring more personalized cancer care within reach!
The world is sleeping on the amazing things we can do with cheap fast neutrons. Marathon (led by the inimitable trio Kyle, Adam, @jfp_fusioneer) is building a reactor for isotopes, batteries, alchemy, & more. Climbing the ladder to precise matter control & abundant electricity
Last month, we demonstrated high-rate alchemy: turning mercury into real, stable gold, beating CERN’s record by a factor of 10.
Last week, we enriched isotopes in the world’s most powerful plasma centrifuge, beating the Soviet record that stood for 40 years.
Now, we’re building a fusion reactor in San Francisco to achieve record power density.
New Mexico has issued a public health order that removes federal restrictions to COVID-19 vaccine access so that pharmacies in New Mexico can vaccinate people of all ages and risk profiles. Every state needs to do this!
Sara is one of the most amazing AI leaders I know. She's built one of the most unique labs in the industry, made huge advances in the field, all while being a fierce champion for under-served communities. If you don't know her work, please have a look at her post below!
It has been an incredible honor to spend the past few years leading @Cohere_Labs@cohere .
This has been the adventure of a lifetime. However, after much deliberation, I made a tough decision 2 months ago it is time to say goodbye.
I open sourced Sniffly, a tool that analyzes Claude Code logs to help me understand my usage patterns and errors.
Key learnings.
1. The biggest type of errors Claude Code made is Content Not Found (20 - 30%). It tries to find files or functions that don't exist.
So I restructured my code base for discoverability, and the average number of steps Claude Code needs for each instruction went from 8 to 7 steps.
BREAKING (ACLU): A federal judge reversed NIH's terminations of hundreds of critical research grants that were canceled because of their alleged connection to disfavored topics, including diversity, equity, inclusion, and gender identity.
This is a major win for public health.
As someone who was formally trained in applied statistics, this book legitimately changed my life.
It's old now, but fundamentally it's the intellectual bridge between statistics and machine learning. And I crossed it.
Truly excellent video by @MLStreetTalk about how a handful of providers have systematically overfit to @lmarena_ai.
26 mins of video showcase how easy it has been to distort the rankings.
As scientists, we must do better. As a community, I hope we can demand better.
It is very easy to make mistakes when creating evals for your AI product. @sh_reya and I run through the most common mistakes in this talk (with memes 🌶️!) . Chapter summaries below:
00:51 Foundation model benchmarks are not the same as your application evals
03:00 Generic Evals Are Useless
04:00 Do not outsource labeling & prompting to non domain experts
09:28 You should make your own data annotation app
12:40 Your LLM prompts should be specific and grounded in error analysis
15:25 Use binary labels
18:57 Look at your data
23:41 Be careful of overfitting to test data
25:40 Do online tests
Links more resources in the reply
This is going to be one of the highest signal sessions of the series. Most issues with RAG are poor retrieval.
Search OG @softwaredoug is going to share his bag of tricks from 10 years of optimizing search in industry
https://t.co/un9RS3EnaL
Overview of Large Language Models for Statisticians
They bridge the gap between statistics and AI - identifying key areas where statistical expertise can enhance LLM development.
List emerging statistical challenges in LLM:
uncertainty quantification, decision-making, causal inference, distribution shift, interpretability, fairness, privacy and watermarking.